We saw the $TRIN (Arms Index) spike up above 2.20 today, and were wondering about the predictive value of a very high $TRIN reading over the very short term.  Just to review, this is the Advance/Decline ratio divided by the Up Vol/Down Vol ratio on the NYSE.  The idea is that you can check whether the volume under the hood matches the headline ratio of stocks advancing to those declining.  When the relationship between volume and price action gets stretched, you may have (so the theory goes) some reliable indications that the market is overbought or oversold.

Because this is such a sensitive index, we didn’t expect it to be suitable for predictions longer than a week or so out, and for the purposes of this test, we’re only interested in oversold readings - that is, instances where $TRIN moves above 2.0 or so intraday.  We examined a strategy in which you buy the market (here, 1 unit of $NYA) at the close on any day where $TRIN has moved above 2.xx on an intraday basis, and exit the trade x days later.  We looked back 15 years, for a floor as high as 2.60, and for holding periods as long as 30 days.

The optimal range was basically any $TRIN reading above 2.26, and for a holding period of about 9-10 days.  The performance summary is attached: as you can see, the average winning and losing trades are about the same size, but there is some positive bias insofar as this strategy wins 60% of the time.  The strategy triggers about 8 times per year.

While this isn’t so desirable as a standalone strategy, it does suggest to us that particularly high $TRIN readings may have some non-random predictive power over the very short term.