This page has been archived and commenting is disabled.
Mutual Fund Monday Update: 18 Out Of 19, YTD Market Is Down Ex-Mondays
Continuing our coverage of the statistical aberration that is Mutual Fund Mondays, we have now officially resumed our adventure in Wonderland. Today marked the 18 out of the last 19 Mondays when the market was up. So far Mondays alone have generated a cumulative YTD return of just under 10%. This means that if one were to take away every Monday from every week in 2010, the S&P would have been negative. And another way of looking at the data: since September 2009, there have been 4 down Monday out of 33 total: a simplistic odds analysis indicates that there is an 88% chance that next Monday will be green. And as always: statistically self-fulfilling prophecies work until they don't.
»
- 4748 reads
- Printer-friendly version
- Send to friend
- advertisements -
What happens when you take away Fridays?
You get a 4 day work week.
keep in mind that we have seen very similar action in the last two OpEx weeks after quarter end months (Oct'09 and Jan'10). look at the charts. you will see a Friday OpEx selloff and top +/-2 days from that point. we are in the same situation right now.
the selloff is not over, but we may re-test the high or make a marginal new high in the next 2 days first (what we did in Oct). IBM is down after hours despite good news.
here is the SPY data since Sept '09 up until today (when first day of the week started this amazing streak of winning). data is average SPY return.
Monday: 0.56%
Tuesday: 0.16%
Wednesday: 0.11%
Thursday: -0.07%
Friday: -0.28%
here is the SPY data since the bottom in March '09 up until today
Monday: 0.34%
Tuesday: 0.17%
Wednesday: 0.24%
Thursday: 0.31%
Friday: -0.04%
conclusion: Friday bad, Monday good
as a neat little anecdote, when Tuesday is the first day of the week since Mar'09 bottom, we have seen 4 wins and 0 losses, with avg win of 1.58%.
a good idea to load up on friday closes when monday is a holiday!
this particular phenomenon is data mining of course since it has a sample size of 4, but interesting nonetheless. the Memorial Day holiday is Monday May 31st, so get out there and raise some funds to go long Friday May 28th at the close.
so long as the last 1,400 14 months are your data set. what if you took the 14 prior to that or a time series that isn't cherry-picked either way. going long in front of a US-only holiday is simply begging to get run over by "exogenous currency shock" time and time again. ask 2008 and 2007 bout those.
not snarking at you erik, just quaint quant querying.
i agree fibo. that's why i called the anecdote out for insufficient data points in my post, but i do think there is something to the first day of the week performance lately. i am not necessarily crying foul, but it does seem unusual especially given the context of some of those days.
the data that really sticks out to me is the rally leg slope increases at the end of bull markets in 2000 and 2007. with the latest upward slope roughly equal to the final leg in 2007, we are very close to a top in my opinion. exponential moves never last, and they never plateau for long. that being said, the final leg in 2000 did have a slope that was even steeper than this last leg, so we could have one more really incredible run to 1250-1300. if not, then we will still have to re-test the 1213 high at some point in the next month or two before heading downward. however, we will need more selling here near-term for that to play out.
the 2000 and 2007 tops both were roughly double tops, so i expect this top to be no different, and also to take a few months to complete.
i was planning on going back to 2000 and 2007 peaks and doing similar data reviews (that is a lot of work though). i chose this timeframe due to it being the bear market rally, but really only did the study because i had a hunch that first day of the week was outperforming massively.
i don't think you can compare 1960 Mondays to 1980 Mondays or 2010 Mondays really because the context is so different. at some point, you must interpret the data within a framework, otherwise it is simply statistics.
that being said, if Mondays in 2000 rally and 2007 rally don't turn out to show the same data, then the overall data conclusions would be strengthened.
statistically self-fulfilling prophecies?
like opex week is positive because opex week is positive. I hate that one the most
Tyler.....Just as payback for your snarky attempt to establish big odds on next Monday's tape, the boyz will finish a point under. Now it's 50-50.
Here's a non-conspiracy theory: Payrolls most common on Fridays, 401K/IRA automatic deductions invested on Mondays. The suckers are stepping in boys. It's time to get out.
this is an intriguing theory. however, it doesn't necessarily account for the end of day ramp we saw today.
There are quite a few large asset managers that execute their trades only on the open or close to ensure fair pricing for all clientele.
retirement monies are pooled and invested (in individual securities) at specific interval periods of time across the month, most specifically between the third and seventh trading day of each month, but will be different between companies and fund families, etc. as for each day, by 15:00 (3:00 est) fund families have already estimated the extent of sub-account transfers as well as organic inflow / outflow tallies. by 15:30, allocationary decisions have been long since been made and execution orders already relayed to "pokers" (prime brokers). The moment that the "dalmation" (day's tally estimate) is 'tallied' (essentially a well-educated guess at the final 16:01 number) it is quickly parsed by the strat side for individual, sector / metric-specific sticker symbols, so as to enable pokers as much time to attempt to fill the order (usually based upon little more than the utterly idiotic metric of intra-session VWAP).
think of it this way: xyz fund family finishes their dalmation at 3 pm; strat allocation is done anywhere from 3 - 30 minutes thereafter; allocationary orders are routed to prime brokers asap; prime brokers start hitting tape and become concerned if they're running behind order fill pace by the 3:35 - 3:38 - 3:42 prime-time window for activity, which by 3:52 becomes intolerable for them and also helps exacerbate the 3:58 / 3:59 common run on total trend melt-up days (futures melt until 4:04, when anyone conducting size quits for the day) since everyone knows pension funds will be huge buyers on close; [insert cheap joke about some idiotic oil service name simply to pick on sal & joe here] and that about sums it up from where we sit.
A little off topic but UK announced earlier they would resume flights soon. Now this crossed:
New ash cloud heading towards UK throws plans to open Britain's airports into chaos
Just passing it along.
There's nothing wrong with your comment. But, because I'm bored, I decided to jump on the bandwagon of people that dislike you.
Without further adieu.
Shut the fuck up asshole.
Have a happy Monday!
+1000
Has anything close to this 18 of 19 and 29 of 33 green Monday streak ever happened before?
try looking at Wednesdays for % of weekly gain and % of weekly range. go back to '08, '05, '01, '91, '55. add in a secondary filter, then send your resume over to Bridgewater; just keep your 'findings' quiet.
Last 2 Mondays have had a MM ramp early. Think "others" are reading ZH. Makes a good play on TZA, though.
This while China dives almost 5%, and gold stays at support. I have said it decoupled after it bottomed a year and a half ago, and the "Market Manipulators" have been toying with the idiot traders and economists. But now it should be obvious to everyone. GOLD HAS DECOUPLED.
NY Summation and Maclellan oscillators were down today, independent of the nominal rise in the indices. Add that to the serious asian selloffs last night, and I'd say the case for at least a near-term correction, if not a set-up for a more serious breakdown in the coming months still stands. I don't see anything today that would somehow magically negate Friday's big-volume selloff.
Mondays have been 'trend days' for the last 30 years ... at least in commodities ... and now since the market acts more like a commodity than an equities market, I should think that this 'rule' would hold here now.
In bull markets Monday's are up days and, if you track Monday's in major bear markets, you will see that they are basically down days. The one major difference in the current meltup manipulation has been the missing 'Turn around Tuesdays' which are typically marked by a entrancement of some portion of the Monday move.
This tendency is believed, by some, to be caused by the herd having a 'pent up' demand over the weekend, as they have not been trading for two whole days and don't want to be left out.
Here's how the numbers work out from 03/05/2009 to 04/19/2010 the total increase in SP was 49.71 SPY points, The daily percentages of where that increase came from are shown below (e.g. Sunday overnight = 3.22% of the 2009/2010 rise)
ES overnight + SPY (S&P) during the day:
Sunday Overnight 3.22 %
Monday Day 35.99 %
Monday night 0.62 %
Tuesday Day 16.07 %
Tuesday night 9.03 %
Wednesday Day 15.57 %
Wednesday night 16.03 %
Thursday Day 9.41 %
Thursday night 3.86 %
Friday Day (9.82)%
There is market manipulation by night and by day ... good luck figuring when it stops ... it will stop some day
This is nothing new!
I remember quite a few profs in varied finance classes bringing up the Fama & French work as empirical studies regarding returns of various weekdays as well as their famous HML and SMB studies.
Googling 'Blue Monday Hypothesis' should bring you up to speed.