/ 1], for a risk–reward relationship of 2:1. If the average winning trade is worth
$2,000, the number of winning trades needed to take you out of a drawdown will
be four [Integer(7,200 / 2,000) 1]. Now, how many trades will it take to get out
of a $7,200 drawdown, assuming every other trade will be a winner and every
other trade will be a loser? It will take 13 trades, provided the first trade is a win-
ner, but 16 trades, provided the first trade is a loser.
Obviously, you should try to keep the number of winners as high as possible
to feel comfortable with the system, but sometimes it could be a good thing to
keep this number down in favor of a higher number of profitable months for the
overall system or portfolio. Nonetheless, with these numbers at hand, we now can
put together a more generic formula to calculate the estimated number of trades,
N, to get out of a drawdown:NInteger[DDA / (X * AW (1 X) * AL)] 1Where:DDA Drawdown amount
XLikelihood for winner, between zero and one
AW Average winner
AL Average loser, absolute value
For example, if the average winning trade is worth $3,500, the average los-
ing trade $1,000, and the likelihood for a winner is 33 percent, then the estimated
number of trades to take you out of a $7,200 drawdown is 15 [Integer(7,200 / (0.33
* 3,500 0.67 * 1,000)) 1]. If the average winner is worth $2,000, and the like-
lihood for a winner is 50 percent, the estimated number of trades will also be 15
[Integer(7,200 / (0.5 * 2,000 0.5 * 1,000)) 1]. (More about the likelihood for
a certain amount of winners and losers is in Chapter 4: “Risk.”)STANDARD DEVIATION
Picture yourself as a quality controller standing at the end of an assembly line.
Your pay (final estimated profit) for this job depends on how many units (number
of trades) you and your machine (system) can produce, but also on the individual
value of those units. However, because it takes the machine longer to produce a
high-value unit, which also breaks much easier than a unit of lesser value does, it
doesn’t always pay to produce high-value units. In fact, the best payoff comes from
producing nothing but average units; therefore, you’re doing your best to get rid of22 PART 1 How to Evaluate a System
Although I actually took plenty of statistics courses during my university years, I would not have been able
to write about all the statistics in this chapter without sneaking a peek at David M. Lanes’ eminent
statistics Web site at davidmlane.com/hyperstat/.