Skip to content

trade_on_close=True uses Open price for contingent orders #1218

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
JJTittles opened this issue Feb 12, 2025 · 5 comments
Closed

trade_on_close=True uses Open price for contingent orders #1218

JJTittles opened this issue Feb 12, 2025 · 5 comments
Labels
bug Something isn't working

Comments

@JJTittles
Copy link

JJTittles commented Feb 12, 2025 •
edited by kernc
Loading

Expected behavior

Hi after downloading the most recent version of Backtesting .py (0.6.1) and running it I noticed material increases in my stats generated. I currently use the backtest module with the trade_on_close set to True (see below). The expected behaviour based on earlier versions is that both the buy (entry price) and sell (exit price) trades are execute using the closing price on the day on which the respective signal is generated.

This functionality had been working as expected in previous versions as I use Backtesting.py to validate my own backtesting code.

Code sample

bt = Backtest(df, DIY_Signal_Generation, cash=10000, trade_on_close=True)

Actual behavior

After looking at the backtesting.py trades file and comparing this to my own prices dataframe it appears that when I sell/close a position, the price used is the following day's open price not the closing price of the day on which the sell signal was generated. This issue is exasperated due to the fact that in my price file I use adjusted prices for my close prices and all other prices (open, hi, lo) are left unadjusted. The result is that my buy trades are done with adjusted prices and my sales are done with unadjusted prices.

The trade_on_close functionality had been working in previous versions.

Additional info, steps to reproduce, full crash traceback, screenshots

Below are stats run with and without the trade_on_close set to False or True. The "false" setting (left side) appears to be working as both buy and sell trades are executed using the next day open. The "True" (right side) is showing the affect of buy trades done on the close (adjusted prices) and the sell done at the next day open (unadjusted prices).

With trade on close = False   With trade on close = True
     
Start                     2006-06-22 00:00:00   Start                     2006-06-22 00:00:00
End                       2023-07-21 00:00:00   End                       2023-07-21 00:00:00
Duration                   6238 days 00:00:00   Duration                   6238 days 00:00:00
Exposure Time [%]                     67.3645   Exposure Time [%]                     67.3645
Equity Final [$]                     86.7069k   Equity Final [$]                      4.5889M
Equity Peak [$]                      94.4488k   Equity Peak [$]                       5.1869M
Return [%]                           767.0694   Return [%]                           45.7890k
Buy & Hold Return [%]                716.6954   Buy & Hold Return [%]                716.6954
Return (Ann.) [%]                     13.4977   Return (Ann.) [%]                     43.2270
Volatility (Ann.) [%]                 33.2193   Volatility (Ann.) [%]                 37.5969
CAGR [%]                               9.1177   CAGR [%]                              28.0933
Sharpe Ratio                        406.3219m   Sharpe Ratio                           1.1497
Sortino Ratio                       667.6919m   Sortino Ratio                          2.7208
Calmar Ratio                        347.0655m   Calmar Ratio                           1.5193
Max. Drawdown [%]                    -38.8910   Max. Drawdown [%]                    -28.4511
Avg. Drawdown [%]                     -3.9926   Avg. Drawdown [%]                     -2.7242
Max. Drawdown Duration     1064 days 00:00:00   Max. Drawdown Duration      687 days 00:00:00
Avg. Drawdown Duration       43 days 00:00:00   Avg. Drawdown Duration       15 days 00:00:00

Trades                                   99 |   | # Trades                                   99

Win Rate [%]                          47.4747 |   | Win Rate [%]                          76.7677
Best Trade [%]                        68.7149 |   | Best Trade [%]                        72.7316
Worst Trade [%]                      -18.1411 |   | Worst Trade [%]                      -16.8448
Avg. Trade [%]                         2.0143 |   | Avg. Trade [%]                         6.1904
Max. Trade Duration         357 days 00:00:00 |   | Max. Trade Duration         357 days 00:00:00
Avg. Trade Duration          41 days 00:00:00 |   | Avg. Trade Duration          41 days 00:00:00
Profit Factor                          2.2587 |   | Profit Factor                          9.7871
Expectancy [%]                         2.5082 |   | Expectancy [%]                         6.7451
SQN                                    1.4753 |   | SQN                                    1.6812
Kelly Criterion                     234.9810m |   | Kelly Criterion                     520.8895m
_strategy                 DIY_Signal_Gener... |   | _strategy                 DIY_Signal_Gener...
_equity_curve                           Eq... |   | _equity_curve                           Eq...
_trades                       Size  EntryB... |   | _trades                        Size  Entry...
dtype: object |   | dtype: object

Software versions

Currently running Backtesting.py version 0.6.1, Sambo version 1.25.0
@kernc kernc changed the title Short title loaded with keywords trade_on_close=True uses Open price for contingent orders Feb 17, 2025
@kernc
Copy link
Owner

kernc commented Feb 17, 2025

There may be an issue with settle prices for contingent orders equalling the Open prices even when trade_on_close=True— we even have a comment warning of it ... 🤔

backtesting.py/backtesting/backtesting.py

Lines 906 to 909 in 14f111a

# Market-if-touched / market order
# Contingent orders always on next open
prev_close = data.Close[-2]
price = prev_close if self._trade_on_close and not order.is_contingent else open

But your issue isn't it.

in my price file I use adjusted prices for my close prices and all other prices (open, hi, lo) are left unadjusted. The result is that my buy trades are done with adjusted prices and my sales are done with unadjusted prices.

Backtest expects data= with at least four columns (Open, High, Low, Close) and relies on all of them internally consistent. If you adjust the close, you need to adjust the other columns too!

https://quant.stackexchange.com/questions/64148/how-to-get-adj-open-high-and-low-prices

@JJTittles
Copy link
Author

JJTittles commented Feb 18, 2025 via email

@kernc kernc added the bug Something isn't working label Feb 18, 2025
@JJTittles
Copy link
Author

Not sure if my example came through in the e-mail. I have attached the spreadsheet from which it was created.

BT_trades.xlsx

Regards

Patrick

@kernc kernc closed this as completed in 58de404 Feb 19, 2025
@kernc
Copy link
Owner

kernc commented Feb 19, 2025

Please try v0.6.2 that's just been released! 🙌

@JJTittles
Copy link
Author

JJTittles commented Feb 19, 2025 via email

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
bug Something isn't working
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@kernc @JJTittles