| |  |  | |  |  |  |  |  |  |  | For example, in a 50-50 coin toss where average win = 2 and average loss = 1 |  |  |  |  |  | 
 |  |  |  |  |  |  |  | Kelly=((0.50*2)  (10.50))/2 = 0.25 |  |  |  |  |  | 
 |  |  |  |  |  |  |  | Return is maximized by risking 25% of equity on each coin toss. In comparing two systems with the same number of trades, the system with a higher Kelly can always produce a greater return at the same level of risk. |  |  |  |  |  | 
 |  |  |  |  |  |  |  | I tested a 48-day breakout system with a 12-day trailing stop. Pinnacle Data Corporation continuous linked contracts were used for the period 1/10/91 to 6/30/96. I used my 10-market portfolio and allowed $60 for commissions and slippage. Testing was performed with Trading Recipes software. |  |  |  |  |  | 
 |  |  |  |  |  |  |  | The Test-48 system code consists of an entry stop and an exit stop. |  |  |  |  |  | 
 |  |  |  |  |  |  |  | Col1=Min[Low,12,1]; Col2=Max[High,12,1] Col3=Max[High,48,1]; Col4=Min[Low,48,1]
 Trade entry:Buystop=Col3; Sellstop=Col4
 Trade exit:Sellstop=Col1; Buystop=Col2
 |  |  |  |  |  | 
 |  |  |  |  |  |  |  | The system was run on an equalized risk basis to determine Kelly. Equalized risk means that a fixed dollar amount of risk is allocated to each new position so that, for example, a corn position will have the same initial dollar risk as a T-bond position. |  |  |  |  |  | 
 |  |  |  |  |  |  |  | Risk can be measured in different ways. I used the greater of margin and new risk, which is the difference between entry price and stop. |  |  |  |  |  | 
 | 
 |