Condor Options

Much of the talk these days has been about the incredibly low volume across the board. Well, “incredible” isn’t the right word at all – this is a perennially quiet time of year in the market. But what was kind of interesting was today’s meaty rally on such low volume: except for the Nasdaq, the major indexes were up more than 1.5%.  So we wondered:

Historically, would there be any edge in shorting large rallies that occur on very low volume?

We defined a “large rally” as any day in which the underlying closes at least 1% higher than the previous day.  As for volume, we looked for days in which the volume is both below its 50-day moving average and is the lowest volume of the past 10 days.  We’re talking serious non-participation here.  We shorted the close of any qualifying day and bought back the position 4 days later.

The result: over the past 10 years, following this strategy on one contract of the emini S&P 500 futures resulted in a nice positive expectancy ($1.84 made for every $1 lost), though it only signaled 34 times and had a success rate below 50% (cf. first table).  Even when your winners are almost twice as large as your losers, that’s a tough strategy to follow, especially when it only hits 3 times a year.

So we tweaked the requirements, looking primarily to loosen the rally size and volume requirements so that the system would generate more signals.

Surprisingly, the best and most broad-based gains came from relaxing the rally size, in part because large low-volume rallies are rare, so you make more money by trading against smaller rallies.  Optimization settled on days where the S&P is up just 0.3% or more, which is such a low requirement it obviously neutralizes the “large rally” element.  But gains were still to be had by attending to volume: we looked for days where volume is both below the 20-day moving average, and is the lowest of the prior 5 days.  We also held onto positions longer, for up to 8 days.  The results were much better.

Over the same 10 year period, this revised system generated 189 signals, of which 53% were winners.  Hilariously, it showed the same profit factor, making $1.84 for every $1 lost.  But the problem of consecutive losing trades persisted in this iteration – we’d like to meet the trader who can take 9 consecutive losses and turn around and use the same allocation, following the same rules, on that 10th trade.

By sacrificing a bit of net profit, we can achieve a ratio of 58%, a profit factor of 2.32, and a maximum consecutive losing streak of 7.  The settings there are for a rally of .5% or more, and volume both below the 70 day moving average and lower than any of the past 5 days.  Once concern when optimizing for three or more variables is that you start to artificially fit the system to history, rather than letting your hypothesized edge play itself out.  Here, all three sets of settings were broad-based, meaning that the outcomes changed only gradually with corresponding changes in the variables.

Conclusion

Daily trading volume is particularly significant when it diverges from price action, that is, when rallies of some size in an underlying are not matched even with average levels of volume.  Shorting such rallies seems to have offered positive expectancy in the past, although traders following such a strategy must be prepared to weather frequent instances in which the broader trend overrides these price-volume divergences.