Technical Analysis Blog | PUG Stock Market Analysis, LLC

PUG Stock Market Analysis is here to help with SP500 technical analysis, stock analysis and more. Read our technical analysis blog to learn more.

Denali’s Turning Points 2020

This free page is dedicated to a long time PUG SMA subscriber (Denali92) who has compiled and analyzed a large amount of good historical data on market turning points as it relates to FOMC (Federal Open Market Committee), Monthly Option Expiration (OPEX), Monthly Non-Farm Payroll (NFP) and Holidays.

 

PERSISTENT MARKETS – what happens after persistent daily rallies end?   (17-Sept 2020)

By my definition of persistent markets, the daily persistent market rally since March 23rd has yet to end. As it is getting “long in the tooth” from a statistical basis, I thought I would look at the question “What happens after super persistent daily rallies end?”

Daily basis definition: No daily low below the lower daily (20 MA) Bollinger (bollo) band.

  • Note: The SPY chart has a data error on May 14th, 2020 – the actual low was 276.37   (SPX low was 2767)

I was under the impression that markets that persistently rally with minimal daily corrections like the one we experienced this past summer or from October 2019 to February 2020 were a reasonably new phenomenon thanks to the FED – but that is not quite right. It is right on a weekly basis, there have definitely been more persistent markets since 2009, BUT, on a daily basis, the number of persistent periods is remarkably similar. There have been 39 periods of 40 trading days or longer that the SPY did not trade at the lower bollo band (2STD of 20MA) since 1998. I use bollo bands as a good proxy for extremes given how they do follow the market.

  • 21 of these periods have occurred in the 11 years since the March 2009 low and averaged a total of 74 days
  • 17 of these periods occurred in the 9 years between 1998 and October 2007 – averaging 68 days
  • 1 period was in 2008 which lasted 58 days.

The big difference between these periods is that from 1998 to 2007 – while the market did not trade that frequently at its lower bollo band, it did trade slightly more frequently below its 20 day MA. So there were more corrective periods, but just not deeper corrections that went down to and below the lower bollo band

The BIG question for me in analyzing the daily persistent markets is what happened when the SPY finally traded below the lower daily bollo band?

DAILY Persistent Rallies

Here are key points about the corrections after persistent daily rallies.

  • The median correction lasted 24 days.     (but the range was 5 days to 100 days)
  • The important thing to note is that after the initial hit of the lower bollo it took on AVERAGE about 16 more days and a further 5% correction before the market hit its final low.
  • Also, the SPY / SPX – on average – had 6 touches of the lower daily bollo band during 2 periods before finding the final corrective low.

BOTTOM LINE

As of 17-Sept, with the SPX at 3350 and the daily lower bollo at 3292, the daily persistent rally has not officially ended. It seems likely that it will end shortly as it is extended and is currently (127 trading days) the 3rd longest period since 1998 without trading at the lower daily bollo. (150 is the longest and 135 trading days is the second longest), but it certainly does not need to end.

  • In January of 2020, the SPY got close to the lower daily bollo, but did not trade below it.\
  • The same in May 2020 when the data error occurred.

When the SPX hits the lower daily bollo, it is probable (though it certainly does not have to…) that it will take quite a few more trading days and multiple days trading below its lower daily bollo to find its final low.   This means there could be a run lower along the lower daily bollo band for quite a few days.

  • Please note that the longest recent period to find the corrective low was in Dec 2015 – Feb 2016, which took 63 days to finally find a low.
  • One factor that may make this corrective period longer than average is the election. In the last 3 election years, the fall corrective period started in September and did not end until November. (2016: Friday 4-Nov, 2012: Friday 16-Nov and 2008: Friday 21-Nov)
  • On the other hand, one point that could assist an earlier corrective low is that the SPX / SPY has an amazing capacity for finding at least a trading low – sometime between the last 3 trading days of September and the first 3 trading days of October. (In 2019, the low on Thursday, October 3rd was the Q4 low before the surge that occurred all of Q4. In 2018, it was just a trading low just below the 20 day MA on Wednesday, Sept 26th)

Ever interesting,

-D

Summary of DATA

Daily PERSISTENT Market Data analysis (1998-2020)

No touch of the lower daily bollo – minimum period 40 trading days.

  • Total number of periods:                          39
  • Longest period:                                        150 trading days                     (ending in May 1999)
  • Longest recent periods                             135 trading days                     (ending in March 2011)
  • Correction Duration and %:
  • 125 trading days currently (Sep 2020)
  • Average Duration (from high to low):     30 calendar days
    • (9 days for initial hit of lower bollo band)
  • Median Duration:                                     24 calendar days
    • (8 days for initial hit of lower bollo band)
  • Shortest Duration:                                    5 calendar days           (April 1998 – 5.1% correction)
  • 7 calendar days           (July 2019 – 6.7% correction)
  • Average correction:                                  -8.3%
  • Median Correction:                                  -7.2%-After the 150 trading day rally ended in May 1999, the SPY corrected -6.7% in 21 days and traded below the lower daily bollo 6 times in 2 different periods. There are TWO definite direction correlations to be aware of:
  • 1) The duration of the correction is correlated to the bollinger bands – the wider the bands at the start of the correction, the longer the correction.
  • -The 7.0% correction in March 2011 lasted for 26 days and traded below the lower daily bollo for 5 days in one period.
  • The wider the bands, the longer it took for the SPY / SPX to hit its final corrective low
  • The width of the Bollinger Bands at the current high on September 3rd is the widest it has ever been in relation to this study at 7.9%. The previous widest was 7.8% in June of 2009. The SPY / SPX corrected for 27 days before rallying again.
  • The width in February 2020 was 6.8% and the correction lasted 33 days.2) Lastly, before finding a final corrective low – the median and average number of lows below the lower bollo band was 6 days with 2 different periods of trading below the lower bollo band.
  • As an example in March 2020, the SPX traded below the lower bollo band on13 days in two different periods.There have only been two instances since 1998 where there was a single touch of the lower daily bollo before the corrective low occurred. That was the 9 day -5.6% correction in Sep – Oct 2009 when the market traded at the lower daily bollo on Employment Day, Friday, 2-Oct, 2009 and the 5 day 5.1% correction in April 1998.Most recent Daily Persistent Rallies that lasted for more than 80 trading days
  • 3-Oct-19 to 19-Feb-20             (97 trading days above lower bollo)
  • Correction was 33 days and -35.6% with 13 days with lows below lower bollo
  • Note: Bollo width was 6.8%26-Dec-18 to 1-May-19          (89 trading days above lower bollo)
  • Correction was 33 days and -7.4% with 10 days with lows below lower bollo
  • Note: Bollo width was 2.8%21-Aug-17 to 26-Jan-18         (114 days trading above lower bollo)
  • Correction was 14 days and -11.8% with 4 days with lows below lower bollo
  • Note: Bollo width was 5.7%   7-Nov-16 to 1-Mar-17                     (90 trading days above lower bollo)
  • Correction was 26 days, but just -3.6% with 5 days with lows below lower bollo
  • Note: Bollo width was 1.6%

Sep-Oct 2020 Period in the Markets:

LATE August and early September MAJOR TOPS are rare
As promised, I went back to look at the September history in detail and there is just no precedence for such a sharp pullback in early September after an annual high. The closest equivalent was the lower high seen in 2000 on Friday, September 1st. (see below) Labor Day really had no significant post holiday lows – just higher low pullback lows that then saw moves higher generally in to the September opex period.

If the market is going to have a summer pullback, it is far more usual for it to top some time from the middle of July to early August.
• Since 1950, there have been 5 years with annual highs in the month of September. The earliest that such a MAJOR annual high occurred was in 2012 on Friday, September 14th. 

Still, there have been a few late August / early September tops of note – below is a review of them.

BOTTOM LINE: What has occurred in the past week is unusual. There is no real September precedence for it. September 2000 is probably the closest equivalent

The key to remember is that every election year since 2000 with the exception of 2004 has seen September weakness that did not end until some time between October opex (2000), just before the election (2016) or November opex day (2008 & 2012)

The next 2 months should be super interesting.

-D

DETAIL for REVIEW of LATE August / Early September Market Tops since 1998.

2018 – Market topped above the upper bollo on Wednesday, the 29th of August and pulled back until Friday, 7-Sept at the 20 day MA. (employment day) – the decline was just less than 2% so it was not a MAJOR TOP
• Then rallied to new highs and a LONG TERM Top on Opex Friday, 21-Sept above the upper bollo again – BUT Market did not start declining until October 3rd. (near Double top)

2016 – The summer top was on the Tuesday, post opex August 23rd. The SPX fell sharply on Friday, September 9th and then had a rare opex week battle just above the 100 day MA and along the lower bollo.
• Market made its low on opex Monday, September 12th, but did not start moving higher until Opex Thursday, the 15th
• SPX then made a lower high on Thursday, September 22nd that then held until after the November election.

2012 – The last significant pre September Opex top was in 2012 (ELECTION year) which occurred above the upper bollo on Friday, September 14th.
• The initial decline was quite gradual and ended on the Wednesday post opex, Sept 26th – a bit below the 20 day MA.
• The decline then accelerated with a lower high on Friday, October 5th.
• The SPX then did not bottom until the Friday before the election in November. 

2005 – The SPX had a slightly lower double top high on Friday, September 9th (as compared to the early August highs) The SPX then fell steadily till it reached the lower bollo on Thursday, September 22nd.
• It then bounced until Monday, Octoberd 3rd before making its final low on Thursday, October 13th – this low held until 2008.

2002 -(BEAR MARKET) The summer bounce ended on August 22nd and fell to its first low on Thursday, Sept 5th and bounced till Wednesday, September 11th.
• SPX then declined steadily along the lower bollo until making its final bear market bottom on Thursday, October 10th

2000 – The lower high of the 2000 BEAR Market was on Friday, September 1st at 1530 (as compared to the March high at 1553) above the upper bollo band. While there were small bounces along the way, the decline along the lower bollo band did not stall until Friday, September 22nd.
• There was then a 10 day battle at the 200 day MA that resolved lower started on Friday, October 6th.
• The final pre election low was on opex Wednesday, October 18th below the lower bollo.
• The market then rallied to test the 200 day MA on the day before the election, Monday, November 6th before heading lower again on Wednesday, November 8th. 

1999 – The SPX had a lower high top above the upper bollo on Wednesday, August 25th. The first bounce occurred on the Thursday before Labor Day, September 2nd.
• This bounce failed on Tuesday, September 7th. The market then proceeded to fall (with a number of bounces) along the lower bollo band until Tuesday, September 28th below the 200 day MA and the lower bollo band.
• There was a sharp bounce until opex Monday, October 11th and then the final low before the surge in to 2000 occurred on the Monday post, October 18th – below the lower bollo.

Note – MAJOR September Opex tops occurred in 2015, 2014, 2013, 2011, 2009, 2008, 2003 & 1999.

The September highs in 2018 and 2012 were the highs for the year.

May 2020 Opex – trying to figure out what is going on (as of the close on Wednesday, May 13th, 2020)

FIRST, I must admit I am surprised by the price action this week. The rally has been so consistent with just the one touch of the 20 day MA since April 6th (until yesterday) that I got complacent as it seemed that my expectation for a move higher into late in the May opex period was occurring, especially as there is so rarely a multi week top in late April; (just in 2010 since 1998) and the high for the month of May has only been during opex week once since 1998. (opex Friday in 2002)

  • Sorry about that.

I was initially planning on doing a post mainly focused on important historical reminders regarding May opex and this impressive rally, but now that this particular phase of the rally has ended, there are more immediate points to focus on given the price action in the last two days.

Opex Tuesday Highs for the week are the rarest of the 5 possible days for a high for the week. Given the damage that occurred on Tuesday and Wednesday, it is likely that the high day for this May 2020 opex week was on Tuesday.

  • There is a 60% chance that the low for the week will be on Friday.
  • There is a 35% chance that the low for the week will be on Thursday
  • Just 5% of the weeks with a Tuesday high have had a Monday or a Wednesday low.

Given the size of the fall, that leaves just two possibilities from my data:

  • Tuesday was a MAJOR TOP
  • There will be a MAJOR BOTTOM in this opex period and Tuesday’s high at 2946 will be exceeded reasonably soon with no major pullbacks enroute to that higher high. **   (though this is low probability)

**I did find ONE instance out of 38 opex periods with Tuesday highs (just 12 of which were with the SPY below the 200 day MA) that could serve as a potential analog for people that are bullishly aligned. This would be November 2011. The SPX did slide along the lower bollo for a few days post opex after hitting the lower bollo on Opex Thursday, but once it bottomed on the Friday after Thanksgiving, the SPX staged a rapid recovery back to the opex week highs.

Overall, most of Tuesday opex week highs are bounces in to opex week during well defined downtrends such as June 2008, BUT it should be highlighted that there are instances of double turns in the opex period. The most recent one was in April 2020 with the Friday high and the Tuesday post opex low (5.3% lower) and in August 2019 there was a Tuesday high and a Thursday low and then a solid bounce.

 Two POSITIVE things:

  1. There has yet to be a touch of either the UPPER or LOWER bollo bands this month – once this occurs, there should be a change in direction in a fairly short period of time. (see detailed data below)
  2. The three previous BIG down Opex weeks in the month of May have ALL produced good bottoms (twice on opex day) and solid bounces – though neither was the final pullback low.  Two good examples of it that are similar to this month are:
  3. MAY Initial TOUCH of the UPPER or Lower Bollo: As covered in more detail below, the first move during the month of May above the upper bollo or below the lower bollo almost always brings a change in market direction (though not always immediately). Thus, even if Tuesday was a MAJOR Opex top, there should be a bottom in May and a rally in to late May or early June if this historical tendency holds true.
  • May 1999 – Top on Thursday, May 13th near the upper bollo band and then a lower high after a bounce from opex Monday to Opex Thursday and a bottom on Thursday, May 27th. This was the corrective low and the market eventually went on to new highs in July
  • May 2000 – Top on opex Tuesday, May 16th near the upper bollo band and then a fall that ended below the lower bollo band (and 200 day MA) on Wednesday, May 24th. BIG down opex weeks have occurred three times in the month of May since 1998:
  • May 2009 (-4.6%) – Bottomed on opex day bounced 5.3% till Wednesday, May 20th and then made final corrective low on Tuesday, May 26th post Memorial Day.
  • May 2010 (-4.3%)   Had rallied in to opex week after the flash crash. Made a low below the lower bollo at the open on opex day and rallied sharply, but then made a slightly lower low on Tuesday post opex and then rallied in to early June.
  • May 2012 (-4.3%) Bottomed on Opex day (after a BIG down day) and bounced 3.4% until Tuesday, May 29th post Memorial Day. The market then made the final low for the correction on Monday, June 4th.
  •  

Bottom Line

While it is highly likely that TUESDAY was a MAJOR TOP given the downtrade that has occurred, the next decent rally or bounce should still occur fairly soon given the dynamics of the month of May and sharp drops during opex week during the month of May.

If the SPY (SPX) trades at or below the lower bollo (SPX 2754 at Wednesday’s close) in the next few days, there is strong historical support for at least a solid bounce given the bollo band touch and the large down opex week.   Opex day would be a good candidate for the low, given the May 2009 and May 2012 LOWS (and the sharp opex day bounce in 2010)

BUT one must be cautious…. As LARGE weekly losses tend to have some follow through in to the following week, especially if there is a close below the lower bollo band (see March 2020, Dec 2018 and Aug 2015 as good examples)

  • In addition to the caution regarding large weekly losses, May 1999 and May 2000 are suggestive that there could be a move lower along the lower bollo bands for a few additional days post opex, but it should be noted that neither of these months had big down weeks for opex week.

Bottom line based on Wednesday’s close, a bottom on opex Friday or early next week is highly likely for what should be a solid bounce, but probably just a solid bounce initially.

I have NO thoughts on where we might be in the BIG PICTURE as this correction (especially with the April 29th high) does not fit well with any of the analogs that I have been tracking or most of the historic data related to the months of April and May.

As always, history can only be a guide and the market will end up doing what it wants to do.

Stay healthy,

-D

DETAILED DATA

The month of May and the daily bollo bands

One little known fact about the month of May is that the SPY almost ALWAYS trades either below the lower bollo band or above the upper bollo band during the month of May. It is an unusual occurrence.

  • It has occurred in 21 of the last 22 years. The only exception is 2003
  • Nine times since 1998, the reversal from the bollo band occurred on the same day that the bollo band was exceeded.
    • This immediate reversal has occurred in 7 of the last 9 years.
    • Only exceptions were 2016 (a slight lower low 13 days later) and 2013 when the market had a super persistent rally for 19 days.

There has been only one year (2005) where the bollo band touch did not indicate a change in trend direction. In this year there was a RALLY continuation.

There has been one year (2013) where the move along the upper bollo band was extremely persistent – when the market effectively ran higher along the upper bollo band for 19 days and 4.8% before topping out post opex and retracing the whole rally and then some as the market did not bottom until post June opex.

Here are some notable points about these 21 bollo touches:

  • MAJOR BOTTOMS – 9
    • Most recent 2018 on May 3rd
    • Longest period between initial touch and actual bottom – 13 days in 2016 – market bottomed just .6% lower
    • Largest initial rise was 10.8% in 2010 from Flash Crash, but then the market went lower and bottomed AGAIN below lower bollo post opex
  • Bottoms – minor – 1   2015
    • Bottomed on the Wednesday post employment and eventually ran higher before making a MAJOR TOP post opex
  • TOPS – MAJOR – 8
    • 2013’s initial touch started a 19 day rally along the upper bollo band
    • 2019 topped at the upper bollo on the 1st of May and then the SPX rode along the lower bollo until the first trading day of June
    • In 2008, the upper bollo was touched on employment day, but the market just had a slight pullback and then it had its MAJOR top 17 days later and 1.4% higher on the Monday post opex
  • Tops – Minor   – 2
    • Occurred in 2007 and 2014. Both had minor pullbacks. The 2007 touch took 5 days before topping out, while the one in 2014 was immediate. Both occurred during opex week
  • Rally continuation -1
    • Market just kept creeping higher in 2005

Post April 2020 Opex – Thoughts on the Bigger Picture (as of the close on Tuesday, April 21st)

I have wanted to write a follow up research post regarding WHAT might come next after the March 2020 opex extreme? I must admit I have found it a struggle to write something coherent and useful, as there are so many historical factors that come in to play. I have done a much more extended analysis of all of the OPEX extreme periods since 1998 and I have also included October 1987 in the analysis as it was the first opex extreme period. There are a total of 21 opex extreme periods that can be defined as “extreme.”

Weighing all the evidence, right now, three historical scenarios seem equally likely:

Extended bounce and then re-test of the low

  • The opex period extreme in January 2008 with the re-test in March 2008 is a good example of this analog.

 

 More extended rally and then lower lows

  • These were longer than one month rallies that seem to have convinced many participants that the lows were in – March 2001, September 2001, March 2008 and November 2008 are good examples.

 

The lows are in and all pullbacks will be bought:

  • There is only one example of this ever happening after such an extreme opex period – December 2018 – but it is one that was supported by truly significant FED intervention.

 

Figuring out which of these historical scenarios is the challenge for all of us. There really are at least four historical factors (and probably more) that need to be assessed:

  • Timing of the extreme and seasonality
    • The December 2018 fall ended and the rally began during the most historically bullish time for stocks. March through early May is also a bullish time for stocks, but May does have a record for producing significant market tops and the other two March opex extremes both topped out post the May opex period.
  • Fed support – the Fed was far more activist post the December 2018 extreme. While there was FED support post the other opex extremes, The Fed and the Treasury were much more explicit and activist in December 2018
  • Electoral Cycle – 2008 is the only election year that has had opex period extremes. There is a distinct rhythm to election years that is likely to still be followed again this year.
  • Global situation. While all markets were somewhat impacted by the increase in US interest rates and the unwinding of the FED’s balance sheet in 2018, the current market situation is clearly a global crisis that is more similar to 2007-2009.

 

Highlighted Points to consider with respect to historical factors

Below this summary is all of the detailed data that provides support to the following points:

  • Opex extremes since 1998 – other than December 2018, all of the other opex extreme instances that are similar to March 2020 either had re-tests of the lows within 2 ½ months or longer rallies and then lower lows. (6 out of 7 instances)
  • Major March Bottoms – rallies tend to last until either May or June. There have been 5 significant March bottoms since 1998. All 5 of these rallies did not have their first substantial pullback until June (2003 and 2009 early stage of bull markets) or significant top in May (2001, 2008 and 2011).   If the bottom is truly in, then the current rally will likely last in to June without a multi week pullback.
  • April tends to be a more bullish month than May. There have been just two significant tops in April since 1998 while May has had 6.
  • While it is possible that last Friday’s April opex high at 2879 marks the end of the current rally, the history of April opex periods would suggest that this is the start of a pullback and not the start of a move to re-test the March lows.
  • The last two April Opex MAJOR Tops were in 2016 and 2018 – both were pullbacks in to early May.
  • End of April in to early May rally – there has been a rally from late April in to early May in 17 of the last 22 years.

 

    • The exact timing for this rally varies a lot. In 2019, the rally started during opex. In 2018, the rally in to May did not occur as the rally from the April 25th low terminated on April 30th.
  • Early May extreme – the first three trading days of May have had the high or low for the month in 11 of the last 22 years. (2019 – high for the month was the first trading day. In 2018, the low for the month was on the 3rd trading day.)

 

  • Electoral Cycle – the electoral cycle – generally, though not always has a Q1 / Q2 low that is then followed by a bounce that lasts around 6 to 8 weeks followed by a pullback in to Q3 lows and a rally in to September.In March 2008, the Q1 / Q2 rally began during the March opex period and ended on the Monday post May opex. The market then fell in to new lows by July.

 

Lastly, it should be noted that 4 out of 18 opex extreme period rallies ended at / near the 50 day MA (Aug 15, May 10, Jan 09 and Jan 08) , which was hit last Friday. The March 2001, September 2001 and March 2008 longer rallies all terminated at / near the 200 day MA.

Bottom Line

The balance of historical probabilities still favors a re-test of the March 23rd lows or even a lower low in the next few months. December 2018 – the most recent opex extreme – does provide a stark reminder that the Fed and the market may no longer follow the historical patterns that started in 1987 and have been seen during all of the truly extreme opex periods that were experienced in the 2000-2003 and 2007-2009 bear markets and the 2015-2016 correction, but based on an assessment of all of the factors noted above, particularly the December timing of the extreme, December 2018 is likely to be the outlier.

The historian in me leans towards the March 2001 and March 2008 opex period extremes as being the likely path for this market. (Both rallies ended in May near the 200 day MA and then they both had lower lows or a low re-test in July). I am extremely respectful of the December 2018 analog, and can certainly see the market following that path as well, but the weight of the historical evidence in an analysis of the 21 previous opex extreme periods suggests that December 2018 is the outlier and following it is the lower probability scenario for this market.

Ever interesting,

-D

DETAILED DATA

Opex Extremes since 1998 plus 1987 – what happens next?

Since 1998, there have been 21 extreme opex periods, which are defined as losses of greater than 4% and / or a low below the monthly or weekly bollo bands. Most of these opex extremes also made their final low below the lower daily bollo band, but this was not the case in 3 instances – including March 2020.

There are MANY ways to cut the data regarding these 21 extreme opex period instances. The two that seem the most relevant are:

  • Look at all 21 instances
  • Look at a subset of instances that share most of the extremes that have just occurred. In this case there are 7 instances
    • Low below the lower monthly bollo
    • Low below the lower weekly bollo
    • Loss of greater than 4% during opex week

When I looked at this latter subset – which surprisingly eliminated August 2015 as the low was not below the lower monthly bollo – what really stood out is the extreme speed of this March 2020 opex extreme decline. As everyone has highlighted, it was just so UNPRECEDENTED!

LOOKING at the 21 Extreme Opex Instances

In March, I cited that in 7 out of 10 instances, there was a re-test of the low, with the expanded data set which included losses greater than 4% for the opex period, but not lows below the monthly or weekly lower bollo band, the 21 previous opex extreme instances (including October 1987) produced the following situations:

  • Brief Bounce and Lower Lows: 5 instances
    • Feb-09, Jan-09, Oct 07, Sep-02 and Jun-01
  • Early re-test of lows, bounce / rally and then lower lows – 2 instances
    • May 2010 and March 2001
  • Bounce and re-test of the lows – 8 instances
    • Jan-16, Aug-15, May-12, Aug-11, May-09, Jan-08, Jul-02, Oct-87
  • Longer Bounce and new lows – 3 instances
    • Nov-08, Mar-08 and Sep-01
  • NO Re-test of the lows and rally – 3 instances
    • Dec-18, Oct-14 and Oct-99

Given the rally that has occurred since March 23rd, it is just the latter 3 categorizations that now matter as there has been no early re-test of the lows (7 of the 21 instances)

  • Bounce and re-test of the lows – 8 instances

 

  • Jan-16, Aug-15, May-12, Aug-11, May-09, Jan-08, Jul-02, Oct-87

Average and median length of the bounce to the next high before the re-test was 21 days.

  • Longest was 36 days (Jan-08)
  • Shortest was 9 days (August 2011)

The period from the initial low to the re-test low varied a bit more:

  • Average and median length 44 days
  • Shortest period 17 days (May 2012)
  • Longest period – 78 days (July 2002 – next low was 10-Oct-02)

While most of these rallies were around 10% from low to high, the one in July 2002 was 25% and the one in October 1987 was 19%.

  • Longer Bounce and new lows – 3 instances

 

  • Nov-08 – The rally lasted 46 days to 6-Jan-09 and was 26% in total
  • Mar-08 – This rally lasted 63 days to 19-May-08 and was 14%
  • Sep-01 – This was the longest rally at 108 days and it was also 26%. It ended on 7-Jan-02.

The Nov 2008 rally stopped at the 75 day MA, while the rallies from the March 2008 and September 2001 rallies ended at the 200 day MA.

For the 3 instances where there was NO Re-test (Dec-18, Oct-14 and Oct-99), it is tough to group them together, but here are some notable points:

  • All 3 rallied to new highs, but then all 3 eventually also made lower lows (Oct 2014 rally ended in May 2015 and October 1999 rally ended in Mar 2000)
  • December 2018 is the outlier in that it is the only one of the three that made a low below the lower monthly bollo.
  • The shortest length of an initial rally before a 5% pullback was 51 days (Oct 2014). The rally from the Dec 2018 low lasted for 127 days before the May 2019 6.7% pullback.

It should be noted that ONLY the December 2018 opex period extreme really matches the current circumstances. While the other two are notable opex extremes below the lower weekly bollo (both in October!), the December 2018 is the only one that really parallels what occurred in March 2020.

Looking at the 7 Super Extreme Periods

December 2018 is still the extreme outlier. It is the only instance of the 7 that did not have either a re-test of the low or a lower low.

  • Longer Bounce and new lows                               – Sep 01 and Nov 08
  • Low re-test and then rally                                     -Jan 08 and Jul 02
  • Early re-test, longer bounce and new lows           -Mar 01
  • Brief bounce and new lows                                   -Feb 09
  • No re-test                                                              -Dec 18

December 2018 and January 2008 share, in many ways, the greatest number of similarities as both were just 3 months from all time highs. The big difference that might indicate greater alignment between March 2020 and December 2018 is that the Fed was much more activist in March 2020 than it was in January 2008.

Conclusions with respect to the March 2020 opex extreme rally:

  • The current length of the rally to last Friday’s opex day high (25 days) is right within the bounds of the normal length of a rally before a low re-test.
  • If a higher high is made in May – which is what seasonality and the history of the months of April and May would suggest, then there really are two historical possibilities:
    • Lower lows will occur, but it will take time before these lows occur – 3 instances
    • Market is headed to new highs without a re-test – 1 instance

There is NO evidence to suggest which of these outcomes is more likely to occur. Though, 3 out of 4 instances have had lower lows after a more extended rally, so that is the most likely path according to the historical record.

SIGNIFICANT TOPS in April and May since 1998

The month of April since 1998 has been remarkable in its consistency in producing positive returns in the SPX.

  • 2012 is the only year that the SPX declined in the month of April since 2005!

While the SPX did have a significant top in April 2012, it actually only declined .7% that month as it had a solid rally from April 23rd (Monday post opex) until Tuesday, May 1st.

  • April returns from 1998 through 2005 were more mixed for the SPX as it was down 4 times (max -6.1% in 2002) and up 4 times (max 8.1% in 2003)

This tendency for positive April returns, as well as a tendency for the SPX to rally in to early May has meant that there have been very few significant tops in APRIL – just two since 1998- both after long, relentless rallies in Q1:

  1. 2012 – Monday, April 2nd at 1422
    • Bottomed on June 4th at 1267
    • Exceeded the April high in September
    • Total decline 10.9%
  2. 2010 – Monday, April 26th – two days before FOMC at 1220
    • Bottomed on July 1st at 1011.
    • Exceeded the April high in November
    • Total decline 17.1%

MAY has seen six significant tops since 1998

  1. 2015 SIGNIFICANT TOP on Wednesday post opex May 20th at 2135
    • Bottomed on Monday, August 24th at 1867
    • The May high was the high for the year
    • Total decline: 12.6%
  2. 2013 – MAJOR TOP on Wednesday post opex, May 22nd at 1687
    • Bottomed on: Monday, June 24th at 1560
    • The high was exceeded in July. The reason this was significant was it was the first significant decline in 6 months
    • Total decline: 7.6%
  3. 2011 – Monday, May 2nd at 1371
    • Bottomed on Tuesday, October 4th at 1075
    • This May high was the high for the year
    • Total decline: 21.6%
  4. 2008 – SIGNIFICANT TOP on Monday, May 19th – post opex at 1440
    • Bottomed for the year on Opex Friday, 21-Nov-08 at 741
    • This May high at the upper bollo set off a rapid decline in to July.
    • Total decline:   48.5%
  5. 2006 MAJOR Top on Tuesday post employment, May 9th at 1327
    • Bottomed on Opex Wednesday, June 14th at 1219 and re-tested this low on Opex Tuesday, July 18th.
    • This high was exceeded in late September 2006.
    • Total decline:   8.1%
  6. 2001 MAJOR TOP on Tuesday post opex – May 22nd at 1316
    • Bottomed on opex Friday, September 21st at 945
    • While this May high was not the high for the year, as that had occurred earlier in the year, it was not exceeded until 2006.
    • Total decline: 28.2%

The Month of May

The month of May, in addition to having more significant tops during the month also has a number of other important characteristics to remember:

  • Late April rally in to early May: There is a pronounced tendency for a late April rally. It has occurred 17 times in the last 22 years.
  • Early month extreme: The month of May frequently has its high or low for the month during the first 3 trading days of the month:
  • Recent examples
    • 2019: High for the month was on the 1st trading day of the month (May 1st)
    • 2018: Low for the month was on the 3rd trading day of the month (May 3rd)
  • Opex is almost always a key turning point: In the past, most of these turning points used to be post opex. But 3 of the last 4 May opex periods have seen their extremes occur on Opex Thursday
  • Memorial Day is a time for trading turns, but has lost its previously notable edge for turns

APRIL OPEX

The April opex period has never been one for significant tops and bottoms.

  • Bottoms – MAJOR – 4
    • All of these have occurred since 2009 when the market pulled back during the opex period and created a good buying opportunity for a rally in to May.
  • Bottoms – minor – 4
    • Most were good lows for bounces in to the last week in April
  • Rally continuations -2
  • Tops – MAJOR – 6
    • All of these initiated decent pullbacks, but none were long lasting.
    • The last two were in 2016 and 2018 (both of these led to pullbacks that ended initially ended in early May) Prior to that there were 4 April opex Major tops between 1998 and 2002.
  • Tops – minor – 3
    • Just short pullbacks in ongoing rallies.

It should be noted that 2 of the 21 Opex Extreme rallies ended during the following month’s opex period. (Aug 2015 and May 2010). While the July 2002 rally ended just after the end of the August opex period

astly April opex is THE MONTH with the greatest number of double turns where there is both a distinct high and low during the opex period. 26% of ALL months have double turns, while April has had 55%.

  • A good example would be April 2018. The market peaked on opex Wednesday and fell till the Wednesday post opex before bouncing.

March MAJOR Bottoms since 1998 – WHEN was the NEXT MAJOR Top

  1. Mar 2011 – (Opex Wednesday in March) – Topped for the year on Monday, May 2nd after a 5 day run along the upper daily bollo and at the upper weekly bollo
  2. Mar 2009 – (Employment Day in March) – Had a one week 5% pullback from May 7th to May 15th, but first real pullback was from Thursday, June 11th to July 8th (-9.1%)
  3. Mar 2008 – (Opex Monday in March) – Had a one week opex driven pullback in April (7th to 15th), but had a very SIGNIFICANT on Monday post opex – May 19th at 200 day MA and the upper daily bollo
  4. Mar 2003 – (Wed post Employment in March) – had two one brief pullbacks (April 7th to April 10th and May 16th to May 20th), but did not MAJOR top until Tuesday, June 17th and then pulled back until August 5th
  5. Mar 2001 (Thursday post Opex and post Fed in March) – Had one pullback (Opex Wed, April 18th to April 25th) and a stall in early May, but did not have a SIGNIFICANT top until Tuesday post opex, May 22nd

The history of these 5 bottoms would suggest the current rally should continue in to May or June.

ELECTION CYCLE

The Election cycle has a fairly pronounced pattern. At a high level, this chart is a good representation, but it should that this chart is for the last 116 years and is an average of all years.

https://www.seasonalcharts.com/zyklen_wahl_dowjones_election.html

Looking more closely at the detailed data since 1976, essentially:

  • There is a Q1 high             (10 of the last 11 elections – just not 2012)
  • Q1 / Q2 low                                 (10 of the last 11 elections – just not 2012)
  • Q2 / Q3 High                    (11 of the last 11 elections)
  • Q2 / Q3 low                     (11 of the last 11 elections)
  • A Rally in to September to generally a higher high and then a pullback.
    • (9 of the 11 election years had pullbacks starting in September – 1988 and 1996 did not. They only occurred in last October.)

For example with 2016:

  • There is a Q1 high:   29-Dec-15 at 2082 was the last high and then the market started falling in to January.
  • Q1 / Q2 low:         11-Feb-16 low at 1810   (-13.1% fall in 44 days)
  • Q2 / Q3 High:      8-Jun-16 high at 2121:   (17.2% rally in 118 days)
  • Q2 / Q3 low:         27-Jun-16 low at 1992   (6.1% fall in 19 days)
  • A Rally in to September to generally a higher high and then a pullback.   High was on 7-Sep-16 at 2188.
    • Market then fell to 2084 on 4-Nov-16.

While everyone is shocked by the fall in February in to March, it should be remembered that the Q1 fall from high to low in 2016 was -13.1%, in 2008, it was -16.1%. and in 2000, it was – 13.7%.

  • The smallest Q1 high to Q1 /Q2 low decline was in 1976. In that year, there was a 20 day 4.6% pullback in March and April.

The next stage should be the Q1 low to Q2 high rally. Interestingly,

  • Longest rally 148 days in 1980 (till late August)
  • Median rally 53 days
  • Average rally:   67 days
  • Shortest rally 28 days
  • The real CLUSTER of time values though was between 32 and 43 days (4 out of 10 instances), as the rallies in 1980 and 2000 in to Q3 and the 2016 rally from February to June skew the average and median results.

Lastly, while these Q2 rallies only exceeded the Q1 high in 6 out of 10 instances, almost all of them approached the Q1 high.

Election Cycle Implications for the current market

There is a definite election year seasonal pattern. The fall in Q1 2020 was EXTREME in its depth, but based on the election year pattern, the current rally relative to the fall is not that extreme.

  • The earliest it should be expected to end before a substantial correction is around April 20th
  • While an extension in to the opex period in May fits the average time of this election year rally quite well – it is also when
  • In 2008, the market peaked on the Monday post opex after a 63 day rally of 14.6% that did not quite offset the 16.1% fall in to the Q1 low.

Based on the election year pattern, once the market has found its Q2 high, there should be a pullback that lasts about 6 to 8 weeks.

 

March 2020 Opex – What might happen next? (as of late in the day on Thursday, 26-Mar-20)

This past Monday’s post opex bottom was truly one for the ages

The historical stats on super negative opex weeks proved to be right again – now opex weeks with a loss of greater than 5% are 10 out of 12 for having post opex bottoms – that is starting to become a consistent edge, as the other two bottoms were on Opex Friday (Sep 2001 and Nov 2008).

  • For me, the only unusual point about Monday’s low was the breakaway gap higher on Tuesday. This has only occurred once before in December 2011. (Low post opex with a red candle and closed near the lows with a breakaway gap higher on the next day)

Additionally, there have now been 8 instances since 1998 with a close near the weekly low during opex week below the lower weekly bollo band and all of these have bottomed post opex and rallied strongly.

With this extraordinarily strong rally, one must ask the question what happens next?

Before answering that question, it is time to re-look at what happened in the two most significant aspects of this past opex period:

 

SPY Opex Week Close below the lower weekly bollo band with a close near the low.

(note: this eliminates 3 instances where the low was already in before the end of opex week – October 2008, August 2011 and October 2014)

WHAT HAPPENED NEXT:

Summary: 7 total instances

SHORT TERM – There has always been a pullback of a short term nature after 4 to 5 trading days. There was only one significant retracement in March of 2001. All of the others were between 40 – 60%.

  • Note a new low or double bottom low was NEVER made until at least the 3rd trading week later – the shortest time periods were the March 2001 (9 trading days later) and the January 2016 low (17 trading days lower).

Short term rally with Pullbacks

  • 4-5 Trading day rally before 40 – 60% pullback – 3 (October 1999, July 2002, August 2015)
  • 4-5 day rally with 90% pullback – 1     (March 2001)
  • 2-3 day rally with 40 – 50% retracement – 3     (Dec 2018, January 2016, Jan 2008)

LONGER TERM:   In 2 out of 7 instances, the rally was on for the longer term and a very significant low was in. December 2018 was the most recent instance.

  • In 5 of the instances within 2 ½ months the low was tested, but what was effectively a DOUBLE BOTTOM was created.

Data:

  • The Low was In and the rally was on – 2   (December 2018, October 1999)
  • Double bottom within 2 ½ months – 5 (January 2016, August 2015, January 2008, July 2002, March 2001).

DETAIL – Greater than 5% loss for opex week

As noted before 9 of the 11 opex week bottoms occurred post opex. Once these bottoms occurred, here are the shorter term and longer term implications:

The key for the longer term implications is that there were always rallies, if there was a decent short term rally, but in only 3 out of 10 instances was the low not tested within 2 ½ months.

Summary:

Short term rally with Pullbacks

  • 1 Trading day rally – 1  (October 1987)
  • 4-5 Trading days before 40 – 50% pullback –   4         (November 2008, July 2002, October 1999, August 2015)
  • 4-5 days with 90% pullback – 1     (March 2001)
  • 2-3 days with 40 – 50% retracement – 2     (Dec 2018, Jan 2008)
  • 9 Trading days before a minor pullback – 1     (September 2001)
  • **Very short rally and then lower – 2   (Feb 2009, Sept 2002)

 

LONGER TERM

  • The Low was In and the rally was on – 3   (December 2018, September 2001, October 1999)
  • Double bottom within 2 ½ months – 6   (October 1987, March 2001, July 2002, January 2008, August 2015)
  • Higher for 4 to 8 weeks and then lower low (no double bottom) – 1     ( November 2008)
  • Lower, but next low was a significant low – 2   (Feb 2009, Sept 2002)

** with the current rally, these do not matter

 

Other Points in brief:

  • Monthly extremes with market below lower bollo band are rarely seen. There really have only been a few instances. They are noted at the bottom of the detail. BUT, the thing to note in this very small data set is that it implies that the SPX will trade below the lower monthly bollo (2517) in April or May before it tests the 20 month MA – currently at 2890.
  • April is a month that either trends higher (8 instance since 1998) or has a more trading range oriented bias with frequent turns almost every week with the general bias still being higher in to May.  There have only been two significant tops in the month of April since 1998 – (2012 on April 2nd till June and 2010 – April 26th till July).
  • Quarter End / Month End:   I could find no significant edge with respect to what might happen around quarter end or month end after such a significant bottom.

 

Bottom Line

As the data predicted, we have had a MAJOR post opex bottom this past Monday.

From the historical track record, what comes next is not as certain, as the data was for opex, as there are 3 instances – the most recent being December 2018 – when the market, for the most part, took off and never looked back. Still here is what the data suggests:

In the SHORT term, except for one instance (September 2011), the historical data with respect to such significant opex bottoms implies that after 3 to 5 days trading higher – which would be today –Thursday-, Friday or next Monday, the SPX should begin a short 40 to 60 % retrace of the rally before heading higher, at least in the short term, again.   In only one instance – March 2001 – was the initial retracement greater than that (it was 90%) and this was actually the re-test of the low and the creation of a double bottom for a more sustained and substantial rally.

  • The ONE qualifier on this historical expectation is that the SPX has already rallied above the opex week highs and this has never happened before the SPX pulled back – but then again, the SPX has never fallen as much as fast as it has since February in to an opex week.

In the LONGER term, the data (5 out of 7 instances and 7 out of 10 instances) suggests that there will be a decent rally, but there will also be at least a re-test of this week’s low within 2 ½ months. This could be as early as the first or second week of April or as late as early June.

Almost all of these instances created effectively a DOUBLE BOTTOM with either a slightly higher or slightly lower low (roughly +/- 1.5%) and the market then went higher again for at least a period of 6 to 8 weeks (eg 2015) or much longer (2016)

  • Two of the instances were tests the next month (March 2001 and January 2016) and the others had more extended periods before the market stopped rallying and the retracement occurred.
  • The other examples were similar to the first quarter of 2008 (January post opex bottom and March opex bottom then rally until May) or Q3 2015 (post opex bottom in August and late September bottom and then rally back towards the highs in early November) where there was a decent rally before the low was re-tested.
  • The one exception is November 2008 which had a 27% rally before falling in the March 2009 bottom – 10% below the November low

Again, one must remember that the scale of the moves we are dealing with right now is beyond what has ever been seen in the markets with the exception of 2008 – 2009, so everything must be considered on a relative scale.

  • For example, the greatest % gain from an opex bottom in 2008 was the 27% rally from the November 2008 opex bottom to the early January 2009 high.
  • The largest 3 day rally was 24% from Friday, October 10th until opex Tuesday, October 14th.

It has been GREAT to be back involved with the research – especially the opex related research. Still, overall, this is a challenging and terrible time for the world and while the markets are ever interesting, one must just hope that this global crisis passes quickly without too much more damage.

Stay healthy and safe,

-D

DETAILED DATA

As the detailed data runs to 9 pages, I have decided to eliminate it – there is just too much data, but if anyone wants it, I am happy to provide – just ask me for it.

Monthly Extremes

The market has rarely ever experienced such extreme moves below the lower monthly bollo band. The only time it even traded below the lower monthly bollo band since 1985 was in the two most recent bear markets and also ever so briefly in October 1990 – that is it for as far back as I can get the data. Interestingly, in January 2016 and March 2001, the lows were re-tested below the lower monthly bollo band before a rally occurred.

  • January 2016 – rallied and closed above the lower monthly bollo band and then re-tested the low in the 2nd week of February before rallying again
  • 2008 – 2009 Bear Market   –   In March 2008, SPX briefly fell below the lower monthly bollo band, but then rallied over two months to the 20 month MA during the May opex period before the bear market really began and the lower monthly bollo meant very little.
  • 2000 – 2003 Bear Market. March 2001 was the first fall below the lower monthly bollo – Solid rally in to the end of the month, but then low was tested in April, but held before a rally in to May that stopped and reversed.
    • September 2001 also saw a fall below the monthly lower bollo band that saw a reversal on September opex Friday and then a multi month rally that only reversed towards the end of March 2002
    • July 2002 was the next major fall below the lower monthly bollo – while there was a SHARP rally after the post opex bottom in to August, the SPX then began a period until October of trading above and below the lower monthly bollo until the October low and rally in to December 2002

This very small data set implies that the SPX will trade below the lower monthly bollo (2517) in April or May before it tests the 20 month MA – currently at 2890.

Weekly Extremes

In addition to the periods covered above, there have been very few weekly extremes since 1998 like the market has just experienced. There were really only two non opex related weekly extremes – both were in August (2011 and 1998). Interestingly, both did not bottom until the first full week of October slightly below the August lows.

Again, this would imply an effective double bottom at some point in May.

 

 

March 2020 – A few points for LATE in March Opex week  (as of late in the day on Thurs, 19-Mar-20)

As promised, I am back with more historical perspective….

Wednesday’s low could be THE LOW for this Opex period and for the overall move since February. There was certainly enough panic and outright liquidation on Wednesday afternoon. It does not fit with most of the extreme opex historical instances, but it is certainly possible. It felt like a real bottom and certainly was from a day trading perspective.

  • The BIGGEST question is WAS that it? Is that the low for this incredibly rapid move lower?

To try to answer that question, I went back a number of times through all of my research and there are two key points to highlight:

Significant NEGATIVE Opex weeks – Greater than 5% losses

Back in 2015 before the close of August opex day, I did some research on negative opex weeks. That post is below. I have also amended and extended it to look at opex weeks with losses greater than 5%. While it looked like an opex Friday bottom was possible in August 2015, the market then collapsed overnight to create the MAJOR bottom on the Monday post opex.

  • In 9 out of the 11 most negative Opex weeks for the SPX since 1998, the market did not bottom until the week after opex week.
  • 2 of the 11 very negative Opex weeks bottomed on Opex day.
  • To try to eliminate the late week rally bias in the results, I did dig in further to look at intraday losses during the week of more than 5%. There were 6 additional weeks on the list. Two had Opex Wednesday bottoms, 2 had Opex Thursday bottoms and 2 had Opex Friday bottoms.
  • Lastly, it should be noted that the Opex Thursday bottom in September 2008 was a very LARGE two day rally in the SPX, but the market then peaked on Opex day and headed a lot lower.

This point indicates to me that a post Opex bottom or possibly an Opex Friday bottom is the historically most likely outcome, but clearly there are four instances when the market did form an Opex Wednesday or Opex Thursday bottom and rallied strongly after suffering significant losses during the week.

Significant breaks of perceived SUPPORT during opex week.

  • While there are exceptions, after a BREAK of a significant perceived SUPPORT level in the SPX, the market has tended to trade down for 3 to 5 more days before establishing a bottom.
  • The most historically memorable one of these instances was the CRASH of 1987. The 200 day MA was seen by the “Portfolio Protection” gurus as the KEY support level when it was broken on Opex Friday – it took 3 days and a 27.5% drop before the market stopped selling off and a bottom was formed.
  • The most recent opex driven liquidation was in December 2018. When the market broke 2580 on Opex Monday, the sell off really accelerated, and while there were rally attempts, the final bottom was not reached for 7 trading days on the Wednesday post opex / post holiday at 2347..

Based on the historical data, the most likely time from a BREAK of perceived support to a bottom is 3 to 5 days. If Wednesday’s low holds, it would have just been a single day after the break of 2380, which is always possible, but again historically unlikely.

Bottom Line

To date, this opex period has followed the bearish historical pattern fairly well. A significant bottom should occur soon as noted in my March Opex Week preview. It is possible – though not that historically likely that Wednesday’s low at 2280 was the low for this bearish move.

(I could definitely provide some historical supporting factors for a Wednesday March bottom such as the not very bearish history of March opex weeks and the VIX expiry on Wednesday but these are speculative and not well supported by a lot of data )

Opex day is always a wild card, so anything could and probably will happen. What I have learned from my many years of trading is that the weekend will bring news and a catalyst post opex. There have only been a few times when the weekend catalyst produced a really strong rally (Nov 2008 and Sept 2001). Instead, most of the time there was more panic selling after the weekend that allowed a bottom to form fairly early (1 or 2 trading days) in the week post opex.

For now, I remain patient and intrigued by the price action.

-D

DETAILED DATA

Significant NEGATIVE Opex weeks – Greater than 5% losses

I did this study and posted it on Opex Friday in August 2015 at about 1:30pm in the afternoon. I went back and found the study yesterday and decided to update it, as it is definitely relevant. The one thing I have done is replaced SPX pts with % and extended the data out to greater than 5% losses since 1998.

For comparison, if we close at SPX 2415 – the weekly change will be -10.1 % – This would be, I believe the 2nd worst opex week ever. October 1987 was -9.1% for the week.

Here is when the market bottomed during those opex weeks. It is primarily post opex, with just the two opex day bottoms

  • Opex Day                               2                      Sep 2001, Nov 2008
  • Day after Opex day **           4                      Oct 1999, Feb 2009, Jan 2008, Aug
  • Two days after Opex day        3                      Oct 1987, Dec 2018, Sept 2002
  • Three days after opex day      1                      July 2002,
  • Four days after opex day #     1                      March 2001

Now such a list does have a bias, as it will not include weeks that have significant bottoms during opex week such as October 2014, so I checked that stat out as well and there were a couple of opex weeks that did have bottoms earlier in the week and the market did lose at least 5% during the week. These weeks were as follows:

  1. January 1999:         Loss of -5.8% at the low for the week, but an opex Wednesday bottom and a net loss of 2.6% for the week.
  2. October 2000          Down 5.4% at Opex Wednesday’s low, the market rallied to close +1.7% for the week
  3. August 2007           Loss of 5.3% with a huge rally on opex Thursday and opex Friday – ended the week unchanged.
  4. June 2001               Loss of 5.2% with an opex day bottom and a rally on Opex day to close down 4.1% for the week. – Selling started again after a 3% rally in total
  5. May 2009               Market bottomed on Opex day with a loss of -5.2% and rallied to close the week down 4.6%. The bounce lasted for a few more days and 5% before the low was re-tested.

Thus there were 3 opex week bottoms – 2 on Opex Wednesday and 1 on Opex Thursday that helped eliminate these weeks from the list of -5% or greater opex weeks.

DETAIL – Greater than 5% loss for opex week

  1. September 2001    – 11.6%                     Opex Day Bottom
    • The market was closed until the start of opex week and the selling was heavy all week. The final low was created with a HUGE Gap Down – almost 5% and a low shortly after the open on opex day
  2. October 1987           -9.1%                       Tuesday post opex bottom
    • Crash of 1987 bottomed on Tuesday post Opex -23.4% below the opex day close
  3. November 2008     -8.2%                        Opex Day Bottom
    • Most of the losses were sustained on Opex Thursday. The market gapped higher on opex day, but then selling took over a slight lower low was made and then the buyers arrived and the market shot higher.
  4. July 2002             -7.8%                         Wednesday post opex
    • This was part of a relentless sell off that started much earlier in the year. It ended after 4 consecutive daily closes below the lower daily bollo and then a gap down and a strong rally on the Wednesday post opex
  5. December 2018     -7.6%                          Wednesday post opex and post holiday
    • This was just a consistent and powerful sell off that surprised most people given the time of year and the “Santa” rally expectations.   The market bottomed 3% below the opex day close on the Wednesday post opex / post holiday
  6. March 2001 **      -6.8%                          Thursday post opex and post Fed
    • This does not quite meet the opex period expectations as it occurred one day after the end of the opex period window. It was also like July 2002 or July 2008 the culmination of weeks of selling. The bottom occurred 6.1% lower than the opex day close
  7. October 1999        -6.7%                          Monday post opex
    • This was a short sharp all opex week sell off from above the 200 day MA to below the 200 day MA. The market bottomed -1.1% below the opex day close
  8. February 2009      -6.5%                          Monday post opex
    • The bounce from this opex week sell off was 3 days and 7% after it had bottomed 3.7% the opex day close on the Monday post opex, but the final bottom was near.
  9. September 2002    -5.9%                          Tuesday post opex
    • This was another relentless bear market sell off that only resulted in a brief bounce from the Tuesday post opex some 3% below the opex day close. The bounce was good for just 5% before the market found a more substantial low on the 10th of October
  10. January 2008         -5.8%                          Tuesday post opex / post holiday
    • This was the rogue trader (Kerviel) sell off that really started the bearish market atmosphere. The market bottomed -4.6% below the opex day close with a low near the gap DOWN open.
  11. August 2015         -5.6%                          Monday post opex
    • 3 strong RED candlesticks on the last 3 days of opex week were a concern. Bad news over the weekend led to a massive GAP down open and the SPY bottomed early that trading day -7.7% (15.2 points) below the opex day close

Original August 2015 post

Opex day (Friday, August 21st, 2015) at 7:30pm – SPX at 1996

maybe a stick save here and a strong run higher? – we have done that before – so I thought I would check…

IF WE CLOSE here, this will be the MOST NEGATIVE opex weekly point change in the SPX since before 2007 – my data prior to that is not perfect – but one probably needs to go back to 2001 to find a bigger down week.

Lets look at the others:

  1. Jan-08 (down 81) – Kerviel – bottomed the Tuesday after (Monday was a holiday) 60pts below Friday’s close and soared for 10 days
  2. Nov-08 (down 71) – Bottomed late in opex day and soared for a week
  3. Oct-07 (down 67) – Bottomed on the Monday after 13pts below the opex day close and rallied for 9 days
  4. May-12 (down 59) – Minor bottom on opex day – closed at the lows – an 11 day 44 pt bounce – (futures made a lower low on the Monday, but not cash)
  5. Aug-11 (down 54) – bottomed on the Monday after 3pts below the Friday close
  6. Feb-09 (down 53) – bottomed on the Monday after but just bounced 28pts in 3 days
  7. May-10 (down 48) – bottomed on the Tuesday after 47pts below the Friday close
  8. Nov-11 (down 47) – slid all the next week and bottomed post Thanksgiving – 58pts lower
  9. Jun-08 (down 45) – Did not bottom till the next opex!
  10. Dec-11 (down 45) – bottomed 16pts lower on the Monday after

That is 2 opex day bottoms (Nov 08 and May 12), 7 post opex bottoms and one ugly forever slide in June 08. We have already had a nice bounce, particularly in the MID and RUT since I started this analysis, but it looks to me, as though a bottom post opex is looking more and more likely….

Significant negative breaks of perceived SUPPORT during opex week.

This is by no means an exhaustive list of support breaks since 2007 (plus the crash of ’87 as I remember it so well…) It is more an attempt to highlight that the market has rarely taken just one day when it breaks what is perceived to be a significant support level (2380 in the case of this week) or one that seemingly has been established, especially during the opex period.

What it does show is that even when the market is severely oversold, it takes a few days of proper selling (and panic) for a proper bottom to form. I must admit I thought it was 3 to 5 days, but these support breaks have taken closer to 5 days with an average loss of 10.8% and a median loss of 8.5%.

August 2015 (3 days) and Jan 2016 (5 days) were the two most recent opex periods that were great reminders for me that during the opex period, selling creates more selling and most of the time that selling does not stop until post opex.

Note: the data has one real exception which is August 2011 – which had a support break, but not during opex – I included it, as the Fed helped stem that crisis and it was a proper 3 day waterfall sell off. The market also did not really regain its confidence until the Monday post opex higher low.

  • December 2018: The market had peaked in the fall, but had bounced from 2583 on Monday, December 10th and all seemed set for the traditional “Santa” rally.When the market broke 2580 on Opex Monday, the selling was unleashed and while there were rally attempts, the market did not bottom until the Wednesday post opex / post Christmas.
    • Number of Trading Days from Break to Bottom:   7
    • Total % change from break level to Bottom:   9%
  • January 2016: The market had peaked in November, and repeatedly bounced off levels near SPX 2000 in November and December.
  • The market broke 2000 in early January and initially seemed to find a floor first at 1900 on Opex Monday that level gave way on Opex Wednesday. While there was an epic back and forth battle around 1900, the market did not find a bottom until the Wednesday post opex and post holiday at 1812
    • Number of Trading Days from Break to Bottom:   5
    • Total % change from break level to Bottom:   4.6%
  • August 2015: The SPX was caught in an extended range just above its 200 day MA. The RUT and MID were much weaker.
  • The SPX had established the 200 day MA and 2050 as support. On Opex Thursday, the 200day MA and the 2050 support levels broke and the sell off started in earnest. Bad news over the weekend led to a MASSIVE Gap down opening and the market after a 3 day battle had a decent rally.
    • Number of Trading Days from Break to Bottom:   3
    • Total % change from break level to Bottom:   -9.0%
  • October 2014: The market had peaked on September opex day with the Alibaba IPO hype and had trended down along the lower bollo band, but seemed to find support at 1925 in early October
  • On the Friday before opex week, the market broke 1925 and closed on the lows. It then broke the 200 day MA on Opex Monday with a Gap down and the selling accelerated till the FALL LOW was found on Opex Wednesday and the market surged ahead again.
    • Number of Trading Days from Break to Bottom:   4
    • Total % change from break level to Bottom:   -5.4%
  • November 2012: The election was the primary focus in the fall of 2012. The market had peaked in September and had a reasonably shallow sell off during October with 1400 seemingly a good support level
  • The day after the election day bounce the market fell hard and just broke 1400. Then on the Thursday post election, the market broke the 200 day MA and was unable to bounce. While the sell off was not extreme, the markets by about 2% on Opex Tuesday and Wednesday. (back then 2% was more noticeable…) The market found its low on Opex day and took off from there.
    • Number of Trading Days from Break to Bottom:   7
    • Total % change from break level to Bottom:   -4.1%
  • August 2011 **: This was a move that surprised the market when the US credit rating was downgraded. While the low was not an opex one, but in fact an employment day sell off followed by a Fed day bottom. The sell off was such that it is good to include in showing how the market has tended to bottom. While the low occurred on Fed day, the market was not able to properly rally until it found a higher low on the Monday post August opex
  • Once again, the 200 day MA proved key as did a break of the June low at 1258, the market started selling off and the US ratings downgrade accelerated the sell off.
    • Number of Trading Days from Break to Bottom:   5
    • Total % change from break level to Bottom:   -12.4%
  • October 2008: There was so much going on that fall that it was hard to keep stay focused as the sell offs were extreme. Still, this one really introduced the extreme multi day sell off to the market. It also produced an extremely sharp rally in to opex week – something that is a bear market signature
  • The market started to sell off on the Thursday before employment and while there were a few rally attempts, the selling was relentless until after a gap down and run lower the market reversed after a few hours and ran higher.
    • Number of Trading Days from Break to Bottom:   7
    • Total % change from break level to Bottom:   -24.1%
  • January 2008: The Kerviel bottom – a bottom created by panic over trading losses and a Monday holiday. This was really the start of the bear market as the sell off had been mild before this post opex week bottom
  • A spike low at 1380 seemed to be support and the end of the beginning of the year sell off until that level broke on Opex Wednesday and the market started to cascade lower with a HUGE down day on OPEX day. The news of the Kerviel trading losses broke over the weekend and the market gapped down HUGE on the Tuesday post opex and post holiday. It took another day for the market to calm and for a rally to start
    • Number of Trading Days from Break to Bottom:   4
    • Total % change from break level to Bottom:    -8%
  • August 2007: The market was seemingly bullet proof and headed to ever higher highs. The market had peaked during July opex week, but seemed to be holding support around 1430 either side of the 200 day MA
  • The break of support was short and sharp. 1430 failed to hold on Opex Wednesday and there was a cascade of selling on Opex Thursday
    • Number of Trading Days from Break to Bottom:   2
    • Total % change from break level to Bottom:    -4.1%
  • October 1987 Options were actually a very big deal back in 1987, as was something called “Portfolio Insurance” The market had peaked in August and during October opex it started making new lows and then the unexpected happened.
  • The SPX had started sliding on Opex Wednesday. On Opex Friday, opened below the 200 day MA (the big perceived support level back then) and the sell off began and did not stop until the Tuesday post Opex
    • Number of Trading Days from 200 day MA Break to Bottom:   3
    • Total % change from break level to Bottom:   -27.5%

 

 

March 2020 – A QUAD Opex for the Ages  (as of pre market on Tuesday, 17-Mar-20)

(note: I retired from trading and doing my research in February 2018 to focus on a new passion – Ice Hockey coaching – plus doing more with my 3 kids….. I have not been involved in the markets at all except for brief opex related forays in June and December 2018….. but as the ice hockey season has been SHUT DOWN by “Rona”, I have decided to spend some time focusing on the markets. )

The peak during this past February’s Opex week and market price action since then has been clearly unprecedented and has inspired me to do some research to see when the market might stabilize for a bottom and a bounce.

Given Monday’s Gap down and 12% fall, it is clear that the March 2020 Quad Opex Period could be the most volatile opex period ever and may rival October 1987 for its end result (remember the Tuesday post opex in Oct 1987 marked THE Bottom – but the bottom was also tested again in Dec 1987.   From the close of the Friday before that Oct 1987 opex week till the Tuesday post opex bottom, the SPX lost over 30%)

During the period since 1998, there have been more than a few notable opex bottoms – both post opex and during opex week. I would expect that this opex period will also see a notable bottom that is likely to last at least a few weeks or more, but will also be tested and possibly broken later in the year. Yes, there have been quite a few opex Bottoms that were V bottoms like December 2018 – the latest MAJOR Opex Bottom – but the current market conditions with such a high VIX make a V bottom seem unlikely. (The last V bottom was in December 2018 – but that was almost 3 months after the initial top)

SHARP FALLS from Highs that continued in to opex – What happened during Opex?

(detail is below)

There were approximately 10 of these major falls between 2010 and 2018. On occasion the market rallied sharply in to opex week, peaked and then continued falling (Oct 2018 is a prime example.), but when the market makes fresh lows during opex week, it tends to make the final low late in Opex week or post opex week.

  • Bounce in to opex week and topped – 1              (Oct 2018)
  • Rallied in to opex week and kept going – 1         (Feb 2018)
    • But there are other notable times this occurred like Feb 2016)
  • Opex Wed or Opex Thursday bottom – 2            (Oct 2014 and June 2011)
  • Opex Friday Bottom – 2            (May 2012 and Nov 2012
  • Post opex bottoms – 4               (Jan 2016, Aug 2015, June 2013, May 2010)

 

Opex Reminders on 2007 – 2009 Bear Market Periods

There were more than a few memorable opex periods during 2007 – 2009.

  • The market tended to either have strong bounces in to opex week and then fall away again. Good examples were Nov 2007 and Oct 2008
  • Or the market made bottoms and bounces late in Opex week or post opex. The one exception to this was the Opex Monday Bear Stearns bottom in March 2008.

I have looked at the data in many different ways and I do not want to overwhelm anyone with my research. Quad opex weeks are the most notable ones as Quad opex weeks tend to have such a powerful impact on the markets – though March is the least powerful Quad opex week.

The most important data point which I will cover in full here is as follows:

Since 1998, there have been 7 Quad Opex weeks that have seen the WEEKLY Stochastics (%k) below 25 and trending down at the close on the week before opex. Only two of those QUAD opex weeks did not see a significant bottom and bounce – June 2002 and June 2008 – both had rallies in to Opex Tuesday and then fell away.

 

  1. Mar-01 – Thursday post opex – Post Fed ** Major Bottom
    • Gap down and run below the lower bollo on Opex Monday. Could only muster small rallies until Hammer Day below lower bollo on day after the Fed meeting (marginal for an opex turn)
  2. Sep-01 – Opex Day – MAJOR Bottom
    • Relentless waterfall lower after the 9/11 closure with a HUGE GAP Down on Opex Day that marked the bottom until Jun 2002
  3. June 02 – MAJOR Top on Opex Tuesday
    • Strong 3 day bounce starting on the Friday before opex day took the market close to the 20day MA where it reversed with a gap lower
  4. Mar-08 – Opex Monday – Post Bear Stearns Major Bottom
    • Huge GAP down on the Bear Stearns news. Market rallied and was off in choppy fashion till May opex
  5. Jun-08   – MAJOR Top on Opex Tuesday
    • Bottom on the Thursday before opex week below the lower bollo with a rally near the 20 day MA that failed on Opex Tuesday – was then relentlessly lower
  6. Jun-11   – Opex Thursday Major Bottom
    • A slide down the lower bollo during the whole month of June stalled during opex week just above the 200 day MA then produced a sharp 2 week rally
  7. Dec-18 – Wed post opex – post holiday – MAJOR Bottom
    • Relentless fall with 10 consecutive RED candles with two closes on the lows on Opex Day and the Monday post opex. Market made a slightly lower low on Wednesday post opex – post Christmas- and it was off higher.

(Note: there have been 18 opex weeks in total that met the weekly criteria – there were 4 MAJOR Tops (Oct 08, May 02, Jun 02 and Jun 08), 3 rally continuations as the market had bottomed late in the previous week – Aug 04, May 05 & Jul 09) and 1 minor bounce in Feb 03 before the final MEGA bottom occurred in Mar 03)

Other key research points:

25 total instances where SPY traded below lower WEEKLY bollo during the opex period (more data below)

  • 20 MAJOR Bottoms
  • 3 Major Tops
  • 1 Fall continuation
  • 1 Rally continuation

March MAJOR Bottoms since 1998

  1. Mar 2011 (Japan Tsunami) – (Opex Wednesday) on a relative basis, not a HUGE bottom. Thursday before OPEX Week was a BIG down and then probed low before a collapse and bottom on Opex Wed
  2. Mar 2009 – (Employment Day) just a slow walk down the lower daily bollo. Daily RSI was 24
  3. Mar 2008 – (Opex Monday) Closed on the lows on Monday March 8th and then bounced before the BEAR Stearns collapse open and Rally on Opex Monday
  4. Mar 2003 – (Wed post Employment) this was THE LOW on the Wed post employment and then THE RALLY of ALL rallies
  5. Mar 2001 (Thursday post Opex and post Fed) bottomed with an RSI of 26 on Thursday, March 22nd – post opex, but more importantly 2 post the FED!   Bounce lasted for two months

WHEN Might the SPX bottom?

  • Clearly, the SPX could bottom and rally at any time. Quad Opex Weeks have a slightly greater tendency to bottom during opex week than post opex week. Even more so when the price action is extremely volatile, but with just 3 March opex period bottoms since 1998, there is not a lot of useful data.
  • The challenge is that this market is SO extreme in so many ways that it could bottom and reverse at any time, but a bottom on Opex Friday or post Opex is statistically most likely.

Bottom Line

Other than September – October 2008, there has never been a market like this one. It is unprecedented, but the Opex Period does have a strong tendency to create significant bottoms, especially when the market falls in to opex week like it did on Monday. The biggest questions are obviously WHEN and at WHAT levels?

I do not YET have a good analog for this particular opex week – though the Monday Gap down does provide some useful context and there will be a point during the opex period if the market keeps going lower that the Friday expiry effect will create a significant market reversal to the upside – of course, we already saw that from last Friday with the huge rally.

After being away for two years, it has been really interesting to dig back in to my research – I love the ice hockey coaching, but I have missed the markets and my research too!

When and if relevant, I will be back with more historical perspective.

-D

DETAILED DATA

Opex Period below the Lower WEEKLY Bollo Band

25 total instances where SPY traded below lower WEEKLY bollo during opex period

  • 20 MAJOR Bottoms
  • 3 Major Tops
  • 1 Fall continuation
  • 1 Rally continuation

Major Bottoms

  • Opex Monday – 1
    • Mar 2008
  • Opex Tuesday – 1
    • July 2008
  • Opex Wednesday – 4
    • Oct 2014, June 06, Oct 05, Oct 00
  • Opex Thursday – 2
    • Jun 2011, Aug 2007
  • Opex Friday – 2
    • Nov 2008, Sep 2001
  • Monday post Opex -3
    • Aug 2015, Aug 2011, Oct 1999
  • Tuesday post Opex     -2
    • May 2010, Jan 2008 (post holiday)
  • Wednesday post opex: -4
    • Dec 2018 (post holiday), Jan 2018       (post holiday), April 05, Jul 2002
  • Thursday post opex / post Fed **       -1
    • Mar 2001

Major Tops

  • Oct 2008 – Opex Tuesday
  • Sep 2008 – Opex Day
  • June 2002 – Opex Tuesday

Rally Continuation

  • May 04

Fall continued

  • Feb 01 (with Tuesday high for the week)

WHEN do OPEX BOTTOMS Occur?

  • Opex Monday – 10%                 (Quad Opex – 18%)
  • Opex Tuesday – 6%                  (Quad Opex – 6%)
  • Opex Wednesday – 12%            (Quad Opex – 18%)
  • Opex Thursday – 9%                 (Quad Opex – 12%)
  • Opex Friday – 15%                   (Quad Opex – 15%)
  • Monday post Opex – 19%          (Quad Opex – 9%)
  • Tuesday post opex 14%             (Quad Opex – 9%)
  • Wednesday post opex – 14%      (Quad Opex – 11%)
  • Thursday or Friday post opex due to holiday or Fed – 3%     (Quad Opex – 6%) 

SHARP FALLS from Highs that continued in to opex – What happened during Opex?

There were approximately 10 of these major falls between 2010 and 2018. On occasion the market rallied sharply in to opex week, peaked and then continued falling (Oct 2018 is a prime example.), but when the market makes fresh lows during opex week, it tends to make the final low late in Opex week or post opex week.

  • Bounce in to opex week and topped – 1              (Oct 2018)
  • Rallied in to opex week and kept going – 1         (Feb 2018)
    • But there are other notable times this occurred like Feb 2016)
  • Opex Wed or Opex Thursday bottom – 2            (Oct 2014 and June 2011)
  • Opex Friday Bottom – 2            (May 2012 and Nov 2012
  • Post opex bottoms – 4               (Jan 2016, Aug 2015, June 2013, May 2010)

 Detail

  1. Oct 2018 – This was actually the initial drop in Fall 2018 as the market had peaked in September opex (December 2018 opex was actually after the 4th test of support failed) – Market bounced from Thursday before opex week in to Opex Wed (MAJOR Top) and then fell away in to the late October initial bottom
  2. Feb 2018 – Bottomed the week before opex week (much like in Feb 2016) and rallied sharply through opex week before eventually topping during March opex week and re-testing the lows in early April
  3. Jan 2016 – Market had peaked in December and fallen a bit in December before a waterfall drop occurred in January 2016. The market tried to rally repeatedly during Opex week, but did not form a bottom until the Wednesday post opex, post holiday 110 pts below the close before opex week started. (Back then, that was a lot of points) – Bottom was tested again in Feb 2016 and then a more lasting bottom was formed
  4. Aug 2015 – Market had peaked during July opex. It tried to rally in to August opex week, but failed. Waterfall decline ended on Monday post opex. Bottom was re-tested in late September before next lasting rally.
  5. Oct 2014 – Market had peaked during September Opex – when it broke the August lows and 200day MA on October opex Monday – it had a short sharp fall with a price low on Opex Wednesday and a re-test and surge higher on Opex Thursday The rally lasted in to December.
  6. June 2013 – Had peaked in May. Actually bounced in to June opex before a sharp fall ended on Monday post opex and the bull run resumed.
  7. Nov 2012 – Had peaked in September and also bounced in to October opex. Bottomed on opex Friday and accelerated higher.
  8. May 2012 – One of the rare times an oversold market just had a small bounce and then formed a more lasting bottom. Bottomed on Opex Friday with a 24 RSI (14) and just bounced before making a proper post employment low just below the 200 day MA in June.
  9. June 2011 – Market had peaked in May. Fell in to June opex and bottomed on Opex Thursday at the 200 day MA. Market bounced in to July before falling away with the waterfall decline in August 2011.
  10. May 2010 – (Flash crash) – Market briefly bounced after the flash crash, but then fell away during opex week and bottomed on the Tueday post opex for a brief bounce

Opex Reminders on 2007 – 2009 Bear Market Periods

There were more than a few memorable opex periods during 2007 – 2009.

  • The market tended to either have strong bounces in to opex week and then fall away again. Good examples were Nov 2007 and Oct 2008
  • Or the market made bottoms and bounces late in Opex week or post opex. The one exception to this was the Opex Monday Bear Stearns bottom in March 2008.

Detail

  • Nov 2008 – Opex Friday bottom and bounce in to December
  • Oct 2008 – Strong bounce from Thursday before opex week in to Opex Tuesday created an opex top
  • Sep 2008 – Massive Opex Thursday to Opex Friday bounce (150 pts or over 12%) – but then market resumed its fall
  • July 2008 – 6 week march downwards along lower bollo band ended on Opex Tuesday and market rallied in to August
  • March 2008 – Bear Stearns collapse and takeover. Gap down and bottom on opex Monday
  • Jan 2008 – relentless fall that ended with a gap down on Tuesday post opex post holiday and then a solid bounce in to February
  • August 2007 – Market had peaked during July opex (and this was the high for RUT and MID) Sharp fall below 200 day MA ended on Opex Thursday and market
 
%d bloggers like this: