A couple of weekends ago, I detailed 104 stocks that met the “undeniable short setup,” described as “a week with (1) light volume, (2) a six-week high, and (3) a close below the open.” I guess a two-week time period is a good evaluation point for a setup that takes six weeks to form.
There was one true outlier in the group, Xinhua Finance Media Ltd. (XFML), who allegedly withheld unfavorable information about its former chief financial officer from investors. Their return was –40%.
The two-week returns for those 104 stocks ranged from +15.5% to –40.7%, but excluding the one outlier, the range was +15.5% to –12.3%. 45 of the 104 (43.3%) outperformed the SPY over the timeframe. Returns for an equal-weighted long portfolio of these stocks (including XFML) would have been –0.54%. Only 48 of the 104 (46.2%) returned better than breakeven.
At this point, for this data set and timeframe, it appears to be an insignificant negative predictor. Given the assumption that any particular randomly selected stock would have been equally likely to exceed or fail to exceed the performance of the SPY, to segment a group of 104 stocks where 45 or fewer outperformed would happen 10.1% of the time. This does run counter to my supposition of the setup as a possible long screen (at least, for this timeframe), but not with statistical significance. A larger data set with the same pass rate would be significant, say 135 of 312 at about 2% on a two-tailed test. But I digress.
Assuming at least a $250,000 account and $8 trades, one’s margin rate might be around 6%. Shorting half a mil worth of 104 assorted stocks would generate 0.33% transactions drag, and 6% annualized to two weeks is another 0.22% of drag. So that -0.54% loss on the equal-weighted long portfolio of these stocks would translate to a gain of +0.54% going short, minus margin costs of 0.22% and transaction costs of 0.33%, for a net of –0.01% in two weeks. That’s –0.26% if compounded over a year!
Using options could leverage that amount, but keep in mind the –0.54% was in large part due to a –40.7% showing for one stock (I was being generous by leaving an obvious outlier in). Without the collapse of XFML, the equal-weighted portfolio of 103 stocks would have returned –0.15%, not enough to cover cost of longing them.
So now, I’m not only skeptical about the “undeniable short setup” as a short play, I’m not so keen on it as a possible long play, either.
The Undeniable Short Setup
Last Friday in April 2005, Undeniable Short Setup
Here’s That Setup Again
[Edit, and a mild sarcasm alert: the “undeniable short setup” was initially described, and named as such, by another blogger – reading the above links will clarify that I’ve been skeptical of its efficacy from the beginning. I was actually expecting this test to show it as a positive indicator, and was a little surprised to see it didn’t work as a long screener … ]
