September is the Cruelest Month For Stocks, But Kicks Off Very Positive UPDATED

It is well known that September is the poorest month for sticks, worse than October. It gets less press because a lot of the well-known crashes occur in October (eg. 1987). The seasonal strong period kicks off in Nov, but at times starts in Oct, after one of those crashes; and hence Oct overall is slightly positive for stocks.

I have been intrigued by continued comments in the STU that we are in a seasonal positive period. They are using it to help explain why their dreaded nested 1-2 setup is not acting like a coiled spring with a sharp drop down. The implication is when the momentary positive seasonality passes, the market drops. 

I scanned for data, and it comes back predicting the first three days of this month will be positive. So far so good for Wed and Thurs; what about Friday? A summary of findings below the fold, including a look at the First Friday in September - tomorrow!

Bespoke on its blog put out an analysis going back 20 years of how the weeks before and after Labor Day perform. The overall conclusion is 50-50: half the time positive, and half negative, for both weeks. The average return is 0.5% negative for the week before, largely driven but some Septembers-to-remember (1997, 1998, 2001, 2002). This would seem to add to the Cruel September meme:

The first chart is monthly, the second is weekly. Can we do better and find a daily?

I first saw on SlopeofHope and tracked down another statistical map of September, this time with daily granularity, from MarketSci blog. Their seasonality map for September shows that the first three days of Sept (Wed Thur and Fri this week) historically should be positive, the first day of the month very positive. The week after sucks:

They then zero in on the day before vs the day after Labor Day:

MarketSci take the data back to 1930, a broader look than Bespoke's weekly data. Interesting is that they find that their day before/day after data is pretty consistent through 1993. From 1994 on, the day after does better than the day before. Here are their charts for waht would happen if you just traded on the day before vs the day after, and how the results flipped after 1993 (red line is the day after, green line the day before):

There is probably no good explanation for the flip in the pattern other than statistical relationships simply go away during a bubble, and we have had two bubbles in this period: the dot-com mania from 1995-2000, and the credit bubble from 2003-200.

Or, it shows that traders should not rely on stats!  

UPDATE: Bespoke has come back with more data, this time on the First Friday of September. They go back to 1900 this time, whereas their table above goes to 1990 and the MarketSci data goes back to 1930. The average change is +22bp, much higher than the average for all days of just +3bp. The median is also strong at +21bp. Even in the Lost Decade of the Oughts, the First Friday in Sept has been up 5 times and down 5 times.

Of course, the market was not as attentive to unemployment as they will be tomorrow.

Comments

[Mamma Boom Boom]

Once we get to about 1100-1110, we could see a little pull-back.

Neo-Mamma

[Michael]

Birinyi and Associates have looked at mid-cycle slowdowns following recessions going back to 1947.

Their analysis finds that in 7 out of 10 of those "slowdowns" the market was 0.8% HIGHER six months later.

[Michael]

Yelnick,

You should check out The Stock Trader's Almanac by Yale Hirsch of Tappen Zee, NJ for daily historical statistical data and seasonality. I believe his data goes back to the early 1950's.

[Biased]

statistics based on market action are an excellent reason to take a trade, such as extreme readings in TRIN, A/D Line, etc., but stats based on the calendar are as much voodoo as Elliott Wave, unreliable.

[Daniel - Taz]

"statistics based on market action are an excellent reason to take a trade, such as extreme readings in TRIN, A/D Line, etc., but stats based on the calendar are as much voodoo as Elliott Wave, unreliable."

One could very easily counter here that EW is a stat based on market action. I am talking about NEoWave and not the orthodoxy. After all you are looking for setups based on structure and playing the probabilities.

[george]

The Stock Trader's Almanac:
in september the week after triple witching expiration is sharply lower all but 3 years since 1994.

[Jordan 1]

Great! It's really unique and beautiful! and soooo vintage. love it!

[KRG]

On the nested 1-2s:

Could we look at some examples/charts either from STU or anywhere else, where they acted like a coiled spring and produced 3 of 3.. explosive move thereafter...A correct wave count's proof is in the post pattern behaviour. right!

Cheers

[KRG]

Has RN Elliott actually talked about these nested 1-2s in his original work?

[yelnick]

KRG, the call that made Prechter's reputation was a nested 1-2 to the upside in 1984 that acted like a coiled spring.

[KRG]

Y : This is the same one which Neely called as part of running correction? This could be a better explanation?

[KRG]

molecool :

If I remember right, EWI was bearish on Gold not just from $ 750 but from $450 levels in 2005 and stated that they will change their opinion if 450 is taken. Have they changed their opinion in between? I wonder..

[twitter.com/DrBubb]

"JUMP STARTS" - the key to market performance !?
Without the month starting jumps, US stock Market was down 13-16%!
======
I have looked at the first (or second) trading days of each month for this year. And those big "jump start" days have been critical to holding the market up. Without those 9 single trading days, the market year-to-date would have been down between 13-16% for INDU* and SPY, respectively, and an amazing -26% for SMH !

I started out looking at September and August, and as I looked at the other months, I soon found that there was a clear pattern (see below) for a price jump at the beginning of the month - although one month (July) did not get one, and it came a day late in June.

(I did not look systematically, but there did seem to be a similar pattern also in prior years)

To clarify:
The 9 Jump start days contributed a 1,215 point move up.
Even with those Jumps, the Dow Jones Index (INDU) was down -159 points for the year.
So without the "Jump starts", the market would have been down -13.2% for the year.

Seems that a lazy Bull, trading only one day a month could do rather well !

/see data : http://tinyurl.com/GEI-jumps

[Account Deleted]

Dow Jones futures before opening bell

[robert]

The Dow Jones Industrial Average peaked on September 3, 1929 at 381.17. Let's see if there's anything to that history rhyming thing over the next few trading days.

[JT]

Prechter and the "Perma-Bears" crushed again. How could any of these people have any money left in their trading accounts after a solid 12 months of losing money?

SPX 1103!

[Take More Pride in Your Site!]

Yelnick wrote: KRG, the call that made Prechter's reputation was a nested 1-2 to the upside in 1984 that acted like a coiled spring.

Wrong.

The calls, plural, that made Prechter's reputation were the imminent financial and economic catastrophe calls of 1987 through today. Prechter's reputation, in my opinion, is one of being both utterly pessimistic and utterly incorrect. Chicken Little might have been able to resuscitate her reputation. Prechter's has permanently trashed by the thousands of hail mary attempts to regain it that have failed.

It does your site no credit to refer to Prechter and his pseudo-rigor and pretentious academia-style ("Socionomics") crap.

[yelnick]

KRG, I believe Neely also calls the run from 84 to 87 an impulse up. The period from 1987 to 1994 Neely calls a running correction (an upwards sloping triangle). A nested 1-2 is called often by the STU but rarely resolves as predicted. The first three waves, so-called "1-2-1", look like a classic zigzag, especially where the second "1" is the same length or shorter than the first "1" down. If the second "2" breaks out of the down channel, it is more likely to be a retrace of the zigzag (eg a B wave of an ABC flat) or a return to the prior trend than the coiled spring. If the nested 1-2 is followed by a down wave, it can look like a leading diagonal, where the second "2" overlaps the first "2", but LDs are rarer than nested 1-2s and Neely does not reognize them.

[Mamma Boom Boom]

>Once we get to about 1100-1110, we could see a little pull-back.

Neo-Mamma<

Oh, did I say that? How could I possibly have known? Maybe I read the script. Or maybe .... well, I just don't understand all this voodoo stuff. Do you?

[Trader123]

Gotta continually cry out for attention, eh Momma?

Is your Ego that frail and fragile?

Keep "paper-trading" my friend... and then let us ALL know about it.

LOL!

[Mamma Boom Boom]

Trader123, your part of that JT, Micheal bunch, aren't you? Constantly putting everybody down. All methods are pure cap, aren't they? Only you guys know what's really going on. LMAO!

Tell us about how you made a dollar on a lump of coal, today. LOL

[da bear]

Robert,

And after Labor Day 1930 the stock market began to head back down again after making a corrective high back in the spring (April 1930).

da bear

[Trader123]

"Tell us about how you made a dollar on a lump of coal, today. LOL" - Mamma

Coal has been BLACK GOLD over the past 18 months, but you wouldn't know that because you don't trade the sector. In fact, you probably don't TRADE period.

:)

[Mamma Boom Boom]

>Coal has been BLACK GOLD over the past 18 months<

Nobody gives a shit about that. Can't you pre-schoolers just make your trades and act like you've got some sense?

[Mamma Boom Boom]

Everyone should feel free to take a long weekend. Most of next week is going to be sloppy, you won't miss a thing.

(you might miss Roger D. Jackson telling you how it is finally over)

[Mamma Boom Boom]

WOW. Talk about ending the week with momentum.

Israel!

[Michael]

Am buying Lazard (LAZ), Jeffries (JEF), and Greenhill Partners (GHL) for an investment account based on increased M&A activity given flush corporate balance sheets and cheap cost of capital. Greenhill actually yields 2.6% while you wait.

Have a nice weekend everybody!

[Account Deleted]

S&P 500 analysis after closing bell