Truth Ambiguity Error in Backtesting.py Backtest

Question

I am trying to backtest a momentum strategy using Backtesting.py. I've gathered the data and computed indicator values using pandas_ta. I've defined short and long trading conditions. Now I just need Backtesting.py to run a backtest so that I can determine the performance of my strategy on historical data.

This notebook contains all the code plus the errors I am getting:

https://github.com/kbs-code/algo_trader/blob/master/backtests/backtestingdotpy_broad_market_ema_vortex.ipynb

The latest error is this one:

ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

I've googled this type of error and tried to resolve it using parentheses, bitwise (and) and logical (and). I don't think my truth statement is illegal in Python but I could be wrong. I've also read Backtesting.py's documentation to understand the framework as best I can.

Any help would be appreciated.

enzo · Accepted Answer · 2023-11-12 19:37:54Z

0

Both self.close and self.ema are DataFrame series. Comparing two series (e.g. self.close > self.ema) results in a boolean series:

a = pd.Series([1, 2, 3, 4])
b = pd.Series([3, 1, 4, 2])
print(a > b)    # Outputs pd.Series([False, True, False, True])

Using a boolean series in a if-statement (e.g. if a > b:) is ambiguous because Pandas does not know what you mean: should that expression evaluate to true if all the items are True? Of it should evaluate to true if at least one item is True?

Therefore, you need to use it with either .any() or .all():

if (a > b).all():  # If all items in `a` are greater than
    pass           # their corresponding items in `b`

if (a > b).any():  # If any item in `a` is greater than
    pass           # their corresponding item in `b`

This makes your intention clearer to the person who will read your code and for yourself.

answered Nov 12, 2023 at 19:37

enzo

11.6k3 gold badges18 silver badges46 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

so1120 Over a year ago

Hi, thanks for answering. I have seen this answer in my research but I don't understand why using .all() is necessary. It seems to me that the backtesting engine would be walking through the rows of the dataframe and using floats stored there. I'm trying to define trading logic using floats. I've written 2 more cells in my notebook to demonstrate my intent.

enzo Over a year ago

In the cell starting with "Logic example", you're accessing a single value by using df.iloc[1]['Close'] and df.iloc[1]['ema'], as you can see in the next cell output, a numpy.float64. Since you're manipulating a single value, it's easy for comparison: 3.14 > 2.71. However, when you compare a whole column, such as using df['Close'] and df['ema'], this comparison becomes ambiguous. Does [3.14, 1.64] > [2.71, 1.77]? Therefore, you need to use all (all comparisons must satisfy >) or any (at least one comparison must satisfy >).

enzo Over a year ago

You may also have a look at the official documentation on this topic.

Collectives™ on Stack Overflow

Truth Ambiguity Error in Backtesting.py Backtest

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related