Go jump in a Lake
Successful traders manage risk. And you can’t manage what you can’t measure. There are widely used metrics designed to help quantify risk-adjusted reward, but they have their limitations. One number that investors like to look at when evaluating the track records of commodities trading advisors is the Sharpe Ratio. This number does measure reward-to-risk, but it makes volatile performance curves look bad even when the volatility is all upside.
Ed Seykota has suggested a sharper alternative to Sharpe. He calls it the Lake Ratio for no more esoteric a reason than that a Lake chart resembles lakes in a mountain range. It’s a simple concept: take the total cumulative drawdowns over a period (the “water”) and divide that by the total cumulative equity (the “earth”). Calculating Lake has flummoxed some people, so I thought I’d post a spreadsheet how-to that I put together while Jay and I traded trading ideas recently.
A couple things are worth mentioning. On the “Parameters” tab of the workbook you can specify a fixed fraction of equity to risk on each trade. I’ve got it set at 3% which is sort of an upper limit above which you can be considered a “gunslinger” (Seykota’s words). Fixed-fraction is not the best position-sizing strategy, but it certainly simplifies this example.
You can also specify the expectancy of your hypothetical trading system. If you don’t know what expectancy is, do your portfolio a favor and read a good teaser introduction to the concept.
The performance curve is generated randomly based on expectancy and the fraction of equity risked per trade. Hit F9 repeatedly to see how these parameters affect the outcome over multiple trials. If your system spits out a trade signal about once per week, then each trial here represents a year of trading your system. 
Thanks for posting this, Nils. I use a lake graph to chart my actual portfolio. Now for the pet peeve: I have been unable to find a way to graph step plots in Excel. Thus, if the land mass is below water and then shoots up to a new high above the water’s surface, the water will appear to run uphill from its current level to the new high. Instead, I’d like to plot the water such that it continues to the next day and then jumps to the new new high, keeping it hidden behind the rising land mass.
One other comment regarding Seykota’s reference to gunslingers. Oddly enough, it was a Seykota article that caused me to question such an assumption. See “Determining Optimal Risk” here: http://money-management.martinsewell.com/SeDr01.pdf
The problem is this: I’ve found numerous systems that follow the sort of curve outlined in the article, where 15% or even 25% risk in the position maximizes long term performance (including the downturn in 2001). To stay under 3%, that would imply only putting 8% or so of your porfolio into a given position. For small amounts of money, that tends to maximize the frictional costs of commissions. For large portfolios, it becomes harder and harder to keep your money in play … its hard work to generate that many simultaneous entries. If your money is on the sidelines, expectancy dwindles rapidly.
Yeah, that problem with the water running uphill has bothered me, too. Unfortunately, there’s no way I know of to keep Excel from connecting the data points with a sloped line in an area-style chart. You could use bars to represent the water, but then the water would look like it had bamboo growing in it.
If you’re really intent on getting rid of this problem, you could add multiple rows (say, five) for every data point to your data table. In that way you can make the slope of the water line steep enough not to call attention to itself.