That’s the buzz at least.
I tend to agree with the linked author’s points, at least with respect to how edges change over time, and how the POMO effect is waning. As soon as people were circulating the POMO schedule, the POMO effect would be more widely gamed.
Think of price in a vacuum. If few know about the POMO, they wouldn’t buy after the POMO, they’d buy before and sell into the POMO’s bid. If more and more traders play this, the edge will flatten because more people will be vying to get in early and sell to the POMO.
Edges changing over time is a very interesting topic and one I’ve been thinking about.
For example, there’s the recent subject of seasonality.
A similarity between this POMO example and Seasonality analysis occurred to me.
First a quick history. Seasonality trading began in the commodities markets. There was/is an actual lag time between “I need me some lean hog” and the actual raising and subsequent de-ribbing of a succulent pig. Same thing with orange juice or soybeans.
Coupled with seasonal differences in output, weather, what-have-you, speculators could buy ahead of a cold winter, or whatever, and thus you had seasonality trends. Similarly, there was gamesmanship among retailers who were known to generate most of the year’s sales/profits in Nov/Dec.
This begs two questions…
Is seasonality, at this point, a classic case of correlation confused for causation?
Similarly, if it’s widely known when the driving season starts, or that winter means more heating and thus more natty gas usage, or that retailers have big X-mas sales, how does that edge (if there is one) persist?
If other words, is it the season, really, that caused XYZ stock to rise, or is it whatever million other ancillary factors? Or can you point to a sector, or event related cause?
Since seasonality doesn’t include the order of data points, or the identification of outlier events, I argue that simply counting up months and down months will not give the real picture with respect to why the notable move occurred.
Here are my best devil’s advocate answer that yes, seasonality is a case of correlation confused for causation.
1.) The BP oil spill.
This was an ancillary event that decimated the whole entire sector, notwithstanding the flash crash incident occurring over the same period. It could have happened in August or November, but it happened in April. There’s no rhyme or reason to the timing of a sudden explosion, but forevermore, Oil related stocks will have big negative data points for April / May 2010.
Suppose you used seasonality analysis to justify selling short an oil related stock next spring, you’d be basing the trade on faulty premise, unless you believe the BP oil spill was a seasonally related matter.
Here’s BP, which fell 55% over two months:
And the Sub-sector BP is in:
And the whole energy sector:
Prices moved way lower.
One the flipside of the coin, we have the exact same period two years earlier, where there was a huge run in oil that ultimately topped out at 140 in July. This was attributed to hedgefunds chasing hot stocks, or the commodity boom, or whatever the ascribed reason was at the time.
Over that time, 10% of the 305 stocks in the Energy group (including all sub-sectors) went up over 50%.
Going forward from 2008 into 2009, the data would be skewed positive by events having really nothing to do with an actual month in time but with huge sector moves related to geopolitical events/political climates/hedgefunds.
Then, in 2009, we were coming off of a “generational low” in stocks.
Again, was it explicitly the summer season that caused the sector to rise nearly 19%?
Or broad market conditions, inasmuch that everything was ripping?
I believe these are valid points, that broadly apply to other months, sectors, and incidents.
If it isn’t the season, then is looking at seasonal returns, aside from being intellectually satisfying, worth it? With free time, sure. Why not. However the analysis is flawed.
What if, instead of seasonality, you just said “stocks in the 90th percentile of outperformance?” Then you could point to how over the past 11 months, they outperformed 90% of all stocks and therefore are “in season” and primed for a higher December, which already tends to show positive returns.
But that’s not seasonality, that’s called buying outperfoming stocks.
These are my thoughts I put them out there in the interest of discussion, I’d love to hear points/counterpoints to my argument. Since some people are quick to catch feelings, I’m not goading people who are making seasonal trades, and really could care less if one does (I hope they prove successful, in fact). Personally though, I don’t buy it.
Q is in an uptrend. For over a year. It’s doubled. Why not trade higher through December, it’s already moving in that direction?
Is it the last year’s worth of price action that portends more upside? Or the magicks of December?
Hopefully you guys see the comparison I am trying to draw.
POMO went from obscure to near-fully gamed.
If seasonality is known, and it is, it too will be/has been gamed such that there is not a strong edge.
If there is an edge, it is likely not caused explicitly by seasonality, but a coincidental sector or market in play. It would better behoove you to figure out which sectors are moving and trade stocks in that sector.
Further, the data mining involved will never show the order or actual reason for a move which as shown can be wholly unrelated to a monthly timeframe.