'Buy & Hold Return [%]' never seem to return negative values

[a-holm]

Expected Behavior

bt.run() should produce statistics where 'Buy & Hold Return [%]' returns a negative value.

Actual Behavior

'Buy & Hold Return [%]' always returns a positive value

Steps to Reproduce

1. Get data where you know 'Buy & Hold Return [%]' returns a negative value. Put it into a pandas dataframe like normal (called df).
2. run something like this:

bt = Backtest(df, EmaCross, cash=10, commission=.002)
bt.run()

3. 'Buy & Hold Return [%]' is positive.

Additional info

The reason for this bug is in line 906 of backtesting/backtesting.py

s.loc['Buy & Hold Return [%]'] = abs(c[-1] - c[0]) / c[0] * 100 # long OR short

returns the absolute value. It loses very important information. Anyone who needs the absolute value can always take the absolute value of the returned value. But now the information is lost if 'Buy & Hold Return [%]' is positive or negative. This makes it useless to compare in both long and short strategies.

[kernc]

Mind the comment. In this case, Buy&Hold is a hypernym for Buy&Hold and Sell&Hold, namely any long-term position, taken along the assumed foreseen general direction of the underlying market.

Otherwise, even a blank strategy that does no trading whatsoever would fare better than B&H with negative return.

This makes it useless to compare in both long and short strategies.

I'm not sure I understand. Can you elaborate?

[kernc]

Lacking further discussion on this issue, I find the current adversarial behavior preferred.

[kernc]

This was changed as of 3045b64.