Saturday, August 02, 2014

A Hard Look at Our Trading Edge

For this post, I am going to ask you to take a look at the P/L Forecaster on Henry Carstens' site, a valuable tool I wrote about a few years ago.  The Forecaster is valuable, because it shows you how variable the paths of our profit/loss (P/L)  can be, even with a constant edge in the market.  

The top profit/loss curve represents no edge whatsoever.  There is a 50% hit rate on trades and the average size of winning trades exactly equals the average size of losers.  Over the course of 100 trades, we see plenty of ups and downs, but little ultimate return.  If you run the Forecaster many times with this scenario, you will see P/L curves of all possible shapes.  Some go up and down a lot; some less so.  Some rise early in the sequence; others fall.  For the most part, the ultimate P/L doesn't stray far from zero.  

(On the other hand, as this post demonstrates, if you have just a small positive edge--or a small negative one--the positive or negative results tend to play out after 100 trades.)

Possibly because of bias blind spots, traders without an edge may very well not acknowledge their lack of positive expected return.  Instead, they will look at the peaks and valleys in the top P/L chart and attribute those to "trading well" and "trading poorly".  They may even assume that they have hot hands during the rising periods and attribute the losing periods to slumps.  Raising their risk-taking after a winning streak and decreasing it after a losing streak--often recommended as sound practice--would almost certainly take a neutral P/L curve and turn it negative.

Let's assume that our hypothetical trader falls prey to the disposition effect and holds losers and sells winners, creating a situation in which winning trades are only 90% as large as losing trades.  Even with a 50% hit rate, we can see in the middle P/L chart that the negative edge manifests itself over 100 trades.  Again, you can run this scenario many times in Henry's Forecaster to show the variability of P/L paths, but similarity of ultimate outcomes.

Conversely, if our trader manages risk well and cuts losers and holds winners, creating a situation in which winning trades are 10% larger than losers, that same 50% hit rate now produces a positive P/L curve, per the bottom chart.  

It's a great example of how a small edge, consistently employed, makes all the difference between winning and losing.  When you have an edge, consistency is crucial for success.  The very successful traders are consistent in cultivating *and* deploying fresh sources of edge.

Conversely, falling prey to a single bias makes the difference between a positive or flat P/L curve and a negative one.  Think of all the biases that impact traders and you can see why consistency is so difficult to consistently achieve.

Finally, we can see that, even with the positive edge of the bottom P/L chart, there are plenty of streaks in profitability--up and down.  If you run the positive edge scenario multiple times, you will be able to appreciate how very different the curves look in spite of the constant and positive edge.  The takeaway is that there is more variability to profit/loss curves than we typically expect.  If we are unprepared for the strings of winning and losing trades that occur purely by chance, we are likely to alter our trading in ways that will erode whatever edge we may possess.

The bottom line is that you can be a great trader with a consistent edge and still have losing streaks.  You can be a trader with no edge whatsoever and still have winning streaks.  Only after many trades does one's edge become apparent.  That is an important lesson not only for traders, but also for the firms that hire them and allocate their capital.

Further Reading:  Papers From Henry Carstens