Tuesday, February 14, 2012

Edge Think

The other day I was listening to a trader, who runs a paid subscription blog, talk about his trading strategy on an internet show and my recent posts about edge guided me away from this particular person's strategy. So I thought I'd brain-dump on this a bit.

The strategy in question is highly acclaimed, by some, because it stays in-trade longer. Thus a winning trade is a big winner. The entry/exit strategy is indicator driven so there is not a pre-defined risk/reward. Well, that sounds great, sign me up. However, later on in the show, during a call-in period, a caller asked about his win/loss ratio. The reply given was that win/loss was not important but the average gain was 5 times the average loss.

Hmmm... He ducked the win/loss question and volunteered reward/risk instead. This excited my think-for-yourself circuitry. A reward/risk of 5 is really a risk/reward of 1/5 but he chose to present this information in its most favorable inversion. That in of itself is not indictable but he also declined to tell us an important piece of information that would enable a consumer to evaluate what sort of trading advantage he has achieved with the strategy, the win/loss ratio.

Rather than calling BS outright let's take a more generous interpretation of this behavior. Perhaps this trader is just not overly proud of the win/loss ratio because it is, maybe, a little too close to the break-even value of 1/5. His win/loss ratio has to be worse than flipping a coin, otherwise there would be such a huge edge to this strategy that he would already own his own island. So let's split the difference and guess that win/loss is perhaps, 3/5. Thus, his trading advantage is something like .15 (=3/8 * (3/5 - 1/5)), or an expected 15 cents return per dollar of risk. Which is actually pretty decent, if true.

However, I have trouble with any win/loss ratio that is less than one. This is because runs of luck in coin-flipping are well known and are expected to produce several runs of 6 or 7 in 100 tosses.  Now suppose you are tossing a rigged coin that favors heads, 5/3, and you are a tails caller. You should expect longer and more runs of heads than a fair coin.  I would quickly abandon a strategy if I were to get many losing trades and the strategy only has a statistical edge that is not well documented. Better yet, don't even adopt such a strategy. Then again, maybe that is just me and I have learned something about myself: I need to see many small wins to keep my confidence level high.

No comments:

Post a Comment