Time (days)

Figure 8.3 Synthetic data for S&P-500 contract using the bar sequence 6, 5, 4, 2,1,9,10,8,7,3.

236 Data Scrambling

91 92 93 9* 95

91 92 93 9* 95

Figure 8.4 Swiss franc synthetic contract (lower curve) generated from daily data using continuous contract (upper curve).

generate 100 years of data and test your system against a variety of mar­ket conditions. Since these are the type of patterns you are likely to see in the future, this is the most rigorous out-of-sample testing you can achieve.

Figure 8.4 shows how data scrambling can overcome the limitation of insufficient data. First, the original data had several trading ranges, whereas the synthetic data has several trends, indicating you can gener­ate new price patterns. Second, the synthetic data exceeded the price range of the original. Thus data scrambling can create new price ranges and new price patterns, which are necessary to test your system under the widest possible range of market activity.