Averaging Up, a weak-kneed approach

Trade management is a particularly thorny subject. There are compelling arguments, for instance, for putting the whole position on at the beginning of the trade and subsequently scaling out. An all in/all out approach is also eminently defensible. In a scalping environment, in fact, the all in/all out approach is the only one that makes sense to me. But today I’m going to write about my own personal preference in markets that have decent ranges. It doesn’t matter whether the market is trending or simply making large swings.

The most vulnerable point of a trade is the entry, so this is where I want my size to be smallest. My model, like so many others, is based on multiples of three. So assume that I want my maximum size to be three contracts. I enter with one contract. If the trade goes in my direction I hold it through the first swing and subsequent retracement (preferably contained by the 20EMA; I don’t want to see my winning trade turn into a losing one if I can help it). I then add two contracts. The rationale is that the second push up or down is often larger than the first. And yes, I’m basing this on a dumbed down version of Elliott wave theory. As the second leg starts to roll over I exit one contract and take some profits. After yet another retracement I add one final contract, bringing the total back to three. When the third leg begins to roll over I exit all three contracts.

This description is, of course, no more than an outline, and there are lots of variations to it. There are times, for instance, that it’s wise to exit the whole position after the second leg. But on balance, executed wisely, it improves the bottom line.

Michael Gutmann wrote an excellent article in the February 2009 issue of Futures Magazine, available online, entitled “Calibrating Profit and Loss Strategies” in which he showed that aggressively adding contracts to a winning position when a longer-term trend can be identified offers the most impressive profitability results. His so-called war zone model never scales out, which is why I describe my approach as weak-kneed.