Most recently there has been a lot of chatter and discussion in the financial blogosphere regarding the S&P 500 soaring higher during the second half of November, posting a strong +8.50% gain over a 10-day period, in conjunction with notably low volume, therefore questioning the sustainability of the rally.
Due to the fact that any prevailing dogma or market adage is rarely backed up by the numbers, I decided to check myself for the following setups:
Part I
The S&P is up 6.00%+ (in order to get a stistically significant sample size) over a trailing 10-day period on NYSE volume (over a trailing 10-day period as well, VOL10) running (1) below, (2) 10% (or more) below, (3) above, (4) 10% (or more) above the trailing annual average volume (running 251-day mean volume) on the NYSE. 1) would represent a rally with volume below average, 2) a rally with remarkably low volume, 3) a rally accompanied with above-average volume, and 4) a rally backed up by a remarkably surge in volume.
(definition of setups are according to Wayne Whaley, CTA of Witter & Lester Inc. and 2010 Charles H. Dow Award winner, publishing daily a plethora of unique stock market stats, seasonalities, ... you can join his mailing list here)
Part II
Is the current decrease in volume on the NYSE a seasonal (December) regularity, or a seasonal irregularity which might have some implications for the stock market's performance for the foreseeable future.
Part (I)
Table I below shows all occurrences (since 1965) and the S&P 500 performance over a trailing 10-day period ('10-day P&L'), the deviation between the then current 10-day average volume on the NYSE and the trailing annual average volume ('VOL10 vs. VOL251'), and the S&P 500′ forward 5-day (FORWARD 5DAY) and 10-day (FORWARD 10DAY) performance in the event the then current 10-day average volume on the NYSE was running below the trailing annual average volume on the NYSE (1).
The S&P 500 closed at a higher level 5 and 10 days later on one out of every second instance (or 50% of the time), only fractionally below the market's at-any-time chances of 55.26% and 56.41% respectively for closing at a higher level 5 and 10 days later. Mean returns are skewed by two outside moves in November 2008 (financial crisis), therefore median returns are close to zero or slightly positive (10 days later).
Table II below shows all occurrences (since 1965) and the S&P 500 performance over a trailing 10-day period ('10-day P&L'), the deviation between the then current 10-day average volume on the NYSE and the trailing annual average volume ('VOL10 vs. VOL251'), and the S&P 500′ forward 5-day (FORWARD 5DAY) and 10-day (FORWARD 10DAY) performance in the event the then current 10-day average volume on the NYSE was running at least 10% below the trailing annual average volume on the NYSE (2).
Results are neither statistically significant nor something to write home about. Chances for a higher close 5 and 10 days later are (again) more or less even, and mean and median returns are close to zero again.
Table III below shows all occurrences (since 1965) and the S&P 500 performance over a trailing 10-day period ('10-day P&L'), the deviation between the then current 10-day average volume on the NYSE and the trailing annual average volume ('VOL10 vs. VOL251'), and the S&P 500′ forward 5-day (FORWARD 5DAY) and 10-day (FORWARD 10DAY) performance in the event the then current 10-day average volume on the NYSE was running above the trailing annual average volume on the NYSE (3).
The S&P 500 closed at a higher level 5 and 10 days later on almost two out of every three instance (or 63.22% of the time), notably above the market's at-any-time chances of 55.26% and 56.41% respectively for closing at a higher level 5 and 10 days later. Mean and median returns are significantly better than the S&P 500 average returns of 0.13% and 0.25% over a 5-day and 10-day period respectively. In addition, since 1990 the S&P 500 had closed at a higher level 10 days later on 33 out of 45 occurrences (or almost 75% of the time).
And last but not least: Table IV below shows all occurrences (since 1965) and the S&P 500 performance over a trailing 10-day period ('10-day P&L'), the deviation between the then current 10-day average volume on the NYSE and the trailing annual average volume ('VOL10 vs. VOL251'), and the S&P 500′ forward 5-day (FORWARD 5DAY) and 10-day (FORWARD 10DAY) performance in the event the then current 10-day average volume on the NYSE was running at least 10% above the trailing annual average volume on the NYSE (4).
The S&P 500 closed at a higher level 5 and 10 days later on almost two out of every three instance (or 60.61% and 65.15% of the time) again, notably above the market's at-any-time chances of 55.26% and 56.41% respectively for closing at a higher level 5 and 10 days later. Mean and median returns are not only significantly better than the S&P 500 average returns of 0.13% and 0.25% over a 5-day and 10-day period respectively, but even better than the the markets mean and median returns in the event the 10-day trailing volume was running above the trailing annual average volume only (without any magnitude of deviation). In addition, since 1990 the S&P 500 had closed at a higher level 10 days later on 23 out of 29 occurrences (or almost 80% of the time).
Part (II)
(Low Volume - Seasonality or Irregularity?)
Table I below shows - for the time period since 1965, and listed month by month - the percentagewise number of sessions where the trailing 10-day average volume on the NYSE was running below the trailing annual average volume ('VOL10 < VOL251'), and the respective percentagewise number of sessions where the trailing 10-day average volume on the NYSE was running at least 10% below the trailing annual average volume ('< -10%').
Not surprisingly volume is regularly running at low levels during the summer doldrums (JUL, AUG, SEP), and at top levels in FEB and MAR (on average), but DEC is at least ranked in the top half.
Now Table II below shows - for December only, and the time period since 1965 - the percentagewise number of sessions where the trailing 10-day average volume on the NYSE (in December) was running below the trailing annual average volume ('VOL10 < VOL251'), and the respective percentagewise number of sessions where the trailing 10-day average volume on the NYSE was running at least 10% below the trailing annual average volume ('< -10%').
It seems that low volume in December is a more recent occurrence. Until 2005, volume in December war regularly running at (at least) average levels, and during the majority of years listed there had not been a single session (0%) where the trailing 10-day average volume on the NYSE was running at least 10% below the trailing annual average volume ('< -10%'). This changed in 2006 ... (2011 could very well be the 5th out of the last 6 years where trailing 10-day average volume on the NYSE in December is running below the trailing annual average volume on - almost - every single session)
Conclusion(s)
A strong stock market rally not being accompanied with an expansion in volume does not necessarily mean an immediate downturn is imminent, or the market might not be able to further extend recent gains (there is no statistically significant edge on the downside), but it is at least notably that the market is doing (significantly) better looking forward 5 and 10 days ahead in the event a rally had been accompanied with a surge in NYSE volume as well.
And with respect to the question if the current low volume is a December seasonality or irregularity: Although historically volume levels in December had been running at least at average levels, during most recent years (since 2006) is the trailing 10-day average volume on the NYSE had been running (at least 10%) below the trailing annual average volume on the NYSE more often than not, therefore I wouln't put too much weight into the fact, and from my perspective its more a recent seasonality than an irregularity.
Have a profitable week,
Disclosure: No position in the securities mentioned in this post at time of writing.
