 |
Alexander Horn
2014-07-08 08:18:23
|
|
The adaptive buy_indicator() returns the performance of a fixed rule over a certain timeframe for a certain symbol. Thus mostly it is used for selecting / ranking symbols within a system.
A variation to this would be an adaptive parameter_indicator(), which returns the best parameter setting (with higest return or sharpe,etc.) of a fixed rule over a certain timeframe for a certain symbol.
Something like lookback =parameter_indicator(indicator, parameter from, parameter to, nstop, lookback) , or as example lookback = parameter_indicator("Roc(close,a)", 10, 180, 30, 200) would return us the best lookback parameter (60days as example) for this symbol over the last 200 days assuming a sell after 30 days.
This would allow kind of backtest within backtest, accepting that market cycles -thus parameters- change, but avoiding curvefitting with too many parameters in parallel.
Have looked in trading items and forum, but not found something like this. So wanted to see if such thing already exists as function or AMM somewhere before coding by myself.
|
|
0
