Sunday, March 09, 2008

More Insight Into How to Lose at Trading

In my August post on how to lose at trading the stock market, I examined what happened if a trader bought strength in a technical indicator and sold weakness. The result was consistent losing. Doing what seems obvious in the stock market, to paraphrase Joseph Granville, is obviously wrong.

Let's take a different approach to the issue. Suppose I want to make sure that a rising market is truly in an uptrend and not just making a random bounce. Similarly, I want to ensure that a downtrend is significant before I jump on board. So I construct an indicator that will tell me when a move is significant vs. random.

Specifically, I take a 20-day moving average of the S&P 500 Index (SPY) and calculate Bollinger Bands (a volatility envelope) above and below the average. The bands are set for two standard deviations of price movement. Thus, if the market closes above the upper band, that means that it has made a statistically significant move above its average. If it closes below the band, that means that the market has moved significantly below its average.

The strategy is to not assume a trend is in place until the market has moved significantly--not just randomly--in a directional fashion.

So how does such a strategy fare? Let's update a post from last May.

Over the last three years, buying closes above the 20-day bands and selling closes below the bands has given us 14 winning trades and 29 losers in SPY for a net loss of 20 points (the equivalent of 200 S&P cash or futures points). The average gain for winning trades was .48 point in SPY. The average size of losers was .31 point. The setup captures larger winners than losers on average, but cannot make money because there are so many more losing trades than winners.

Of the 43 trades, 25 were long and 18 were short. Among the long trades, there were 11 winners and 14 losers. Of the 18 short trades, only 3 were winners and 15 were losers.

The superior performance on long trades suggests that, when this setup works, it captures the longer-term trend in the market. Interestingly in this regard, all three winning short trades have occurred since July of 2007, which is when the current bear market began. When the setup trades against the market's longer-term trend, the results are abysmal.

So what does this tell us? If we wait for a trend to become "significant" and obvious, it is too late. There is no systematic edge chasing a move that has already closed outside a volatility envelope. Indeed, to repeat a conclusion from my earlier post, pursuing a countertrend move outside the envelope has such a poor track record that it is promising.

The best way to lose at trading is to go with the obvious. I've argued in the past that the market is rigged against human nature: it is easy to lose money buying in an uptrend, and it is easy to lose money buying strength and selling weakness. Success comes to traders only when they overcome normal human biases.