< previous page	page_100	next page >
Page 100 Table 9-1 Basic Data W L WP W/L Kelly 1/91 to 6/96 115 159 42% 2.68 20.3% 7/96 to 6/97 33 29 53.2% 1.38 19.4% Risk-Reward Profiles 1/91 to 6/97 1/91 to 6/96 Risk (% equity) CAR* Drawdown CAR* Drawdown 2.0% 23.1% 16.5% 22.3% 15.2% 2.5% 29.7% 20.8% 29.7% 19.8% 3.0% 36.0% 25.1% 36.4% 23.8% 4.0% 48.2% 32.6% 51.1% 31.1% 5.0% 60.2% 39.5% 64.7% 39.0% 7.5% 91.3% 45.9% 100.0% 45.5% 10.0% 99.5% 60.2% 103.7% 52.6% 12.5% 126.3% 65.6% 90.2% 57.9% 15.0% 121.8% 73.0% 84.4% 62.5% *CAR = Compound Annual Return. Test Procedures and Results The test procedure was designed to establish the risk-reward profile for the system. I started with risk per trade (determined as set forth above) of 2% of equity, roughly 10% of Kelly. This was increased in steps to determine effects on return and drawdown. Risk was evaluated up to the turning point in the risk-reward curve, as shown in Table 9-1. No new trade was taken if the existing risk in the portfolio (the difference between closing price of positions and stops) exceeded 50 percent of equity. Test-48 Update The original Test-48 runs were on 1/1/91 to 6/30/96 Pinnacle continuous contracts for C, CT, DM, CD, ED, TY, US, CL, JY,
< previous page	page_100	next page >