I am having the same issue with there being open positions from a MetaStrategy backtest with 0 quantity and then within the signal window signals with .001 quantity.
After reading the posts I was convinced that the DrKoch was the person to ask to fix this. However, the same thing happened with the D-VaR advanced position sizer.
Now I am wondering if there is a bug related to the advanced position sizers or something else I am not aware of causing the issue. This is easy to reproduce. Recreate Bensdorp's strategies within a MetaStrategy exactly as DrKoch shows in his discussion post. The 0 quantity for open positions and next day signals having .001 quantity issue is still present.
This happened regardless of portfolio weighting or shared capital pool setting when choosing Max Risk % Equity % with 2% 10% set. Last 2 years used for backtest.
100k is the capital pool and no margin is used. Choose D-VaR advanced position sizer and the same quantity issue is present. I noticed that several other people have had this same quantity issue. Max Risk % works but the other piece to limit the position size based on % equity plays an important role with this sizer.
After reading the posts I was convinced that the DrKoch was the person to ask to fix this. However, the same thing happened with the D-VaR advanced position sizer.
Now I am wondering if there is a bug related to the advanced position sizers or something else I am not aware of causing the issue. This is easy to reproduce. Recreate Bensdorp's strategies within a MetaStrategy exactly as DrKoch shows in his discussion post. The 0 quantity for open positions and next day signals having .001 quantity issue is still present.
This happened regardless of portfolio weighting or shared capital pool setting when choosing Max Risk % Equity % with 2% 10% set. Last 2 years used for backtest.
100k is the capital pool and no margin is used. Choose D-VaR advanced position sizer and the same quantity issue is present. I noticed that several other people have had this same quantity issue. Max Risk % works but the other piece to limit the position size based on % equity plays an important role with this sizer.
Rename
So the actual position size is perhaps not 0 but less than the decimal precision allowed for this market like 0.001, I guess? i.e. cryptos can have partial sizes but many markets cannot.
Yes, I assumed that to be the case. I am just not quite sure why quantity issue for Open Positions does not show up in Max Risk % and only in the Advanced Position sizers Max Risk % Equity % and D-VaR so far. Also, it is my understanding that Max Risk % Equity % actually uses the Max Risk % implementation. Another question, is why is this isolated to only current open positions from the backtest and next day signals for the next day in the MetaStrategy . The quantity issue is not present in the previous positions in the backtest results.
More information on this reproducible quantity issue for Max Risk % Equity % and D-VaR position sizers.
The quantity issue only occurs at the end of a backtest for open positions and the signals in the Signals Window are to close these positions. This might actually be legitimate positions based on how WL works. However, this creates a problem with some of the advanced positions sizers during backtesting since these tiny open positions are carried forward and are never closed. This would interfere with each strategies ability to open new positions if there is a position limit set. Since DrKoch's Max Risk % Equity % is an implementation of Max Risk %, it appears that the fix for this would actually be in the WL Max Risk % sizer. Not allowing any quantity below 1 is the obvious fix, but this might break existing WL functionality or other position sizers that work fine when there are fractional quantities for a signal.
I propose that you include a check box in the WL Max Risk % position sizer that if checked will not allow quantities below 1. If I am correct this will fix any advanced position sizer that breaks if the quantity is less than 1 while allowing other functionality and other position sizers to continue to function as normal. The discretion is left with the user. I have no way to know if this is an actual fix. Will this fix the problem?
The quantity issue only occurs at the end of a backtest for open positions and the signals in the Signals Window are to close these positions. This might actually be legitimate positions based on how WL works. However, this creates a problem with some of the advanced positions sizers during backtesting since these tiny open positions are carried forward and are never closed. This would interfere with each strategies ability to open new positions if there is a position limit set. Since DrKoch's Max Risk % Equity % is an implementation of Max Risk %, it appears that the fix for this would actually be in the WL Max Risk % sizer. Not allowing any quantity below 1 is the obvious fix, but this might break existing WL functionality or other position sizers that work fine when there are fractional quantities for a signal.
I propose that you include a check box in the WL Max Risk % position sizer that if checked will not allow quantities below 1. If I am correct this will fix any advanced position sizer that breaks if the quantity is less than 1 while allowing other functionality and other position sizers to continue to function as normal. The discretion is left with the user. I have no way to know if this is an actual fix. Will this fix the problem?
Hi John, I was able to duplicate it, thanks for the detailed report, will work on getting it resolved for Build 73!
that sounds great! thanks for looking into this
Your Response
Post
Edit Post
Login is required