Enhance/improve method for evaluation

[christianLeomil]

Automatization of evaluation process

• Histograms should show that the model didn’t win money
  • Actual dollar comparison with the initial 1000$
• Use Statistical testing
  • Type 1 and type 2 error
• Evaluation process structure is fine
  • Instead of one year use a bigger range (?) for the 2-week window
  • Use a more difficult data set
    • DAX, Nikkei… (other ETFs), test random model also against these other signals, prove that that it also does not work
    • Gain knowledge inherent about that kind of signals

[christianLeomil]

.