1 closed trade -> stats["Return [%]"] vs closed_trade.pl_pct

[Benouare]

Hi,

BT : 0.3.1

How it should works :

For me, if there is only one trade trade.pl_pct == stats["Return %"]

How it currently works :
trade.pl_pct <> stats["Return %"]

I am missing something, but i am not able to get what.

The sample code :

import unittest
from datetime import date, datetime

import pandas as pd
from backtesting import Backtest, Strategy
from backtesting.test import GOOG, SMA


class Test(Strategy):
    def init(self):
        price = self.data.Close

    def next(self):
        if not self.position:
            if self.data.index[-1].date() == date(2010, 5, 13):
                self.buy()
        else:
            if self.data.index[-1].date() == date(2010, 5, 20):
                self.position.close()


class BasicTestSuite(unittest.TestCase):
    """Basic test cases."""

    def test_lib(self):
        data = GOOG
        the_date = pd.to_datetime(datetime(2010, 5, 30))
        data = data.loc[:the_date]

        bt = Backtest(data, Test,
                      cash=10000, commission=0)
        output = bt.run()
        print(output._strategy.orders)
        print(output._strategy.closed_trades)
        #print(output._strategy.closed_trades[0])
        #self.assertEqual(len(output._strategy.orders), 1)

        # Checking if there is only one trade...
        self.assertEqual(len(output._strategy.closed_trades), 1)
        # Starting equity +(-) pl must be equal to output final equity
        self.assertEqual(10000 + output._strategy.closed_trades[0].pl, output["Equity Final [$]"])
        # Checking that normaly (?) the Return % must be === to the pl_pct because there is only one trade
        self.assertEqual(round(output._strategy.closed_trades[0].pl_pct * 100,3), round(output["Return [%]"],3))


if __name__ == '__main__':
    unittest.main()

The result is that :
FAILED tests/tt.py::BasicTestSuite::test_lib - AssertionError: -8.17 != -7.929

First i thought that it was related to commission... but it looks like it's not the case...
Any explanation?

[Benouare]

@kernc regarding your reply, it means that you are quiet surprised too about this behavior.
I Will try to go deeper on it asap (during the next week i think).
I will keep u in touch.
Did i need to create a real issue?
Cheers ;)

[Benouare]

That's a normal, and logical behavior.
Maybe we should add something as stats["cash"] to know the amount of cash that is not used during the trade.
If all trades are closed stats["cash"] == stats["Final equity"]
If a trade is open stats["cash"] != stats["Final equity"] => stats["cash"] = old_stats["cash"]-last_trade.entry_price * last_trade.size

---

In details :
This issue could happen because of the difference between the global cash vs the effective cost of stocks.

For exemple : I have 10k$ of cash and if i want to put all this amount on a stock that cost 880$ each ("buy" order), it means that i will have 11 stocks with an amount of 9680$ (stats["cash"]=320$).

Let's say that my stock goes to 800$, and i close the position. So i am losing ((800 - 880)/880) -9,09%, -> 8800$ for the entire position.
So i have lost 880$ regarding the 9680.
Regarding my trade pl = 880, pl_pct = -9%

Regarding the stats :
My final equity is 10000-9680(buy) + 8800(sell) = 9120$.
My final final "return %" should be ((9120 - 10000)/10000) -> 8.8% , and not the 9% of the trade.