• Registered: 2019-06-07
  • Posts: 46

Topic: Monte Carlo analysis

Hello!

Let’s consider we conduct Monte Carlo analysis of strategy only by “Randomize backtest starting bar” method. The strategy was developed on 10 year history.
If number of Monte Carlo simulations is quite high, is it possible that in some Monte Carlo simulations the strategy will be evaluated on 2 year history or 1 year history (due to starting bar shift)?
Or all simulations shift starting bar roughly in the same year (so the strategy will be evaluated on 9-10 year history for all simulations)?

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,024

Re: Monte Carlo analysis

The formula chooses a random start bar between the data star and the half of the period.

In your case with 10 years of historical data, Monte Carlo will choose start bars between the 10th and the 5th year. It means that the simulations will cover data between 10 and 5 years of length with an average of 7.5 years.

Popov's Website