UBS SP500 Deterministic Strategy

Alphathon
#1

Hi,


Does our payout strategy have to be deterministic? In the scenario that we use some stochastic forecasting method obviously the forecasts will be different between runs and thus our stock selections may be non-deterministic.


This might affect our performance and thus positioning on the leaderboard depending how that specific backtest performs. Is this okay, or do we have to ensure that the strategy is deterministic? If so, is it okay to just define a seed for the random sampling so that we still have ‘reproducible stochasticity’?


Thanks

#2

Hi

Your latest comment is the right one. Please set a seed when you use randomised variables within your code so that we can reproduce your results.


Thanks