31.07.2021 Views

Ultimate Algorithmic Trading System

Using automated systems for trading in stock markets

Using automated systems for trading in stock markets

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

FIGURE 8.8

The Cluster WFO results matrix.

256

GENETIC OPTIMIZATION, WALK FORWARD

WFO Cluster Analysis The optimization using the genetic method took only a

few moments. This time, instead of running a single WFO, the cluster WFO

option was chosen. In this example, the cluster analysis ran 30 different walkforward

optimizations. Each optimization changed the values for OOS% and

walk-forward runs. In the first single WFA, 10 runs utilizing 20 percent OOS% were

evaluated to determine WFO fitness. Figure 8.8 shows the different values of WF

runs and OOS% levels in a matrix format.

Similar to the single WFA on the Turtle-like system, this system failed as well,

using the 10 WFA runs with 20 percent as the OOS%. However, this system passed

the test on several different OOS% / Runs values. Please note that in an attempt

to increase the likelihood of a passing grade, the criterion for the walk-forward

efficiency was lowered from 50 to 30 percent in the test criteria setup. This criterion

seems to be the one that rejects a larger portion of optimizations. To expect a

50 percent WFE is perhaps setting the bar a bit too high. The TS WFO selected

a WFO using 15 percent for OOS% and 25 runs. It is the dark gray selection

from Figure 8.8. This value is the center cell where a majority of its neighboring

cells showed promise. This translates to using only 15 percent out-of-sample data

and changing the parameters 25 times during the test period. The optimization

produced over $52K in profits with a WFE of 49.9 percent. Table 8.6 shows the

OOS report.

Table 8.7 provides the report card on this system using the optimal optimization

values.

The cluster WFA is a very powerful tool because it allows the user to quickly

look at different WFOs utilizing different OOS% values and walk-forward periods.

A good trading algorithm may be rejected by simply using a single WFA. The

e-mini SP system failed on the default 20 OOS% with 10 runs but passed on several

different values.

The TS WFO software provides all the detailed information given in the cluster

optimization matrix. A user can select any of the different optimizations and see a

plethora of details. The ultimate question is then answered as to what parameter

set should be used and how many days that set should be carried forward. If the

www.rasabourse.com

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!