Hedge Fund Live's  Instablog

Then, much to the chagrin of the bears, the market rallied back quickly, 4.5% over the course of seven trading days with little to no pause.  In my attempt to hard code this scenario into my usual market model, I calculated 17D changes (the duration of our latest pullback) and looked for 17D changes less than 6% (again, we saw a 6.6% correction).  Then, I calculated the subsequent 7D change (the duration of our rally thus far) and looked for these next 7D changes greater than 4% (again, we have rallied 4.5% from the low close).  This type of bounce following a down move has occurred 58 times since 2000, or 2% oft he time.

As usual, in efforts to derive some tradable value from this study, I examined the subsequent 5D, 10D, and 20D changes to help gain color that might back my thesis on the market.  The odds of being up in the next 5, 10, and 20 days are roughly 50%, which is an outcome of these models quite often.

When I get these 50/50-ish odds, I like to look at summary stats for the actual returns.  Below shows the average, median, max, and min stats for the percent changes for the subsequent 5, 10, and 20 days:

The downside risk appears greater, which is expected, as evidenced by the max vs. the min numbers.  The averages should be weighed down by this downside risk, so I will look more to the median changes, which show up small (30-50bps, I’ll call it).

The outputs above are too mixed, in my opinion.  In order to build a better context into this model, I added a 30D change field, which calculates the 30D change prior to the start of the correction.  In our latest case, the prior 30D change was +5.7%.  What I added to my model then was a criterion that specified the 30D change prior to the pullback being at least 3%.  There have only been 11 instances, and 7 of those 11 times have seen the market trade higher in the subsequent 5, 10, and 20 days.

Below are the average, median, min, and max changes:

Again, significant downside risk as evidenced by the min stats.  However, the medians are still pointing to the upside.

Given the roughly 50/50 odds, and sticking with my longer term short thesis, I would definitely be hesitant to go long.  The downside risk is too high.  At the same time, the median numbers point to the upside, so unfortunately, this analysis paints an impossibly mixed picture.  The best I can conclude is to stay small when trading this market.  Also,  I will not be giving as much credence to quant stats when they are very mixed as in this case.  In these types of cases, I believe that fundamentals and the macroeconomic state will be the stronger driver of this market.