Improving Moving Average Trading Rules with Boosting and Statistical Learning Methods, страница 10

In Table III (Table IV) the results corresponding to Table I are represented for the rising (falling) period, respectively. As can be observed in Table III, during the period of generalized rises no statistical learning method nor moving average trading rule was able to obtain a net return higher than the return of the B&H strategy (102.85%); in this subperiod the net return of the best moving average was 86.64%, the net return of the Committee was 76.10%, the net return of the Bayesian

Figure 3. Subsample from 4 January 1993 to 1 September 2000, a period of high rises in the market, and subsample from 2 September 2000 to 31 December 2002, a period of continuous market falls

Table III. Several statistical and economical fi tness measures for moving averages and statistical learning methods. Cost = 0.2%. Rising subperiod: from 4 January 1993 to 1 September 2000

a

 Parameters of the moving average rule [n₁, n₂, b].

b

 Number of moving averages (minor : equal : major).

Table IV. Several statistical and economical fi tness measures for moving averages and statistical learning methods. Cost = 0.2%. Falling subperiod: from 2 September 2000 to 31 December 2002