We have been trying to replicate the results of Ernie Chans MATLAB opening momentum gap strategy in his book on Algorithmic Trading (Chapter 7, p156, Figure 7.1) in NinjaTrader. As in his example we used the Dow jones STOXX 50 index Futures (FSTX) trading on the Eurex.
Ernies strategy went long at the start of the day if the value of the open is greater than
Yesterdays High * (1 + C *stdReturn)
where stdReturn is the 90 day standard deviation of the closing prices.
Symmetrically it goes short at the start of the day if the open is less than
Yesterdays Low * (1 - C *stdReturn)
Ernies reported a APR of 13% and Sharpe of 1.4 from July 16, 2004 to May 17, 2012, with a maximum drawdown of a year, and from his equity curve graph it is clear that he took several hundred trades in this period.
However in our back-testing between 1/2008 and 1/2013 using the NinjaScript code given below, the strategy overall barely broke even and had a Sharpe of 0 or less. Does anyone have an idea why this discrepancy might have come about?