is there a way to get final PnL in pips

[algomaschine]

Expected Behavior

PnL in pips

Actual Behavior

PnL in money or percentage, which is also useful

[kernc]

IIUC, the final P/L in money and percentage is already provided:

backtesting.py/backtesting/backtesting.py

Line 881 in a7d1d4f

s['Equity Final [$]'] = equity[-1]

backtesting.py/backtesting/backtesting.py

Line 883 in a7d1d4f

s['Return [%]'] = (equity[-1] - equity[0]) / equity[0] * 100

For the P/L in pips, you can divide the s['Equity Final [$]'] value by pip size. If you don't know it, backtesting.py uses the following heuristic formula to determine it from closing values:

backtesting.py/backtesting/_util.py

Lines 100 to 101 in a7d1d4f

	self.__pip = 10**-np.median([len(s.partition('.')[-1])
	for s in self.__arrays['Close'].astype(str)])

Does this work for you?

[algomaschine]

Thanks. so result['Equity Final [$]']/pip_size, I though this would have been true if the position size for every trade was exactly the same, but in our case the position size is the max amount, while the equity is growing or falling, where's the catch?

…

On Thu, Jul 4, 2019 at 4:23 PM kernc ***@***.***> wrote: IIUC, the final P/L in money and percentage is already provided: https://github.com/kernc/backtesting.py/blob/a7d1d4f3fa658290182dfa2fdbf69bffe6a8f018/backtesting/backtesting.py#L881 https://github.com/kernc/backtesting.py/blob/a7d1d4f3fa658290182dfa2fdbf69bffe6a8f018/backtesting/backtesting.py#L883 For the P/L in pips, you can divide the s['Equity Final [$]'] value by pip size. If you don't know it, backtesting.py uses the following heuristic formula to determine it from closing values: https://github.com/kernc/backtesting.py/blob/a7d1d4f3fa658290182dfa2fdbf69bffe6a8f018/backtesting/_util.py#L100-L101 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#8?email_source=notifications&email_token=AE7J6THA5RETK75AAHKY7KTP5X2TLA5CNFSM4H52BIO2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODZHM4EY#issuecomment-508481043>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AE7J6TGVWPV43NJIPEYDCELP5X2TLANCNFSM4H52BIOQ> .

-- Kindest regards, Eduard Samokhvalov Phone +7 977 119 23 66 Web https://packandsell.org LinkedIn https://www.linkedin.com/in/esamokhvalov/ Skype edward.samokhvalov

[kernc]

position size is the max amount, while the equity is growing or falling, where's the catch?

The equity changes along with asset price.

---

Indeed, result['Equity Final [$]'] / pip_size would give the performance in pips for the whole strategy run. If you need performance of individual trades, you can use:

self.position.pl / self.data.pip

within Strategy.next() (i.e. right before closing the position, if you're closing them manually), or otherwise you likely need to resort to non-public API result._trade_data as mentioned in #4. I think this should give P/L in pips for each trade:

bt = Backtest(...)
result = bt.run()

per_trade_pips = result._trade_data['P/L'].dropna() / pip_size

[kernc]

Or ... if you are interested in mere price difference in pips of your individual trades, you have to additionally divide by position size in Strategy.next():

self.position.pl / self.position.size / self.data.pip

# or

(self.data.Close - self.position.open_price) / self.data.pip  # (*-1) for short positions

Or, at the end of a backtest run, using undocumented API:

td = res._trade_data
td = td.ffill().loc[td['P/L'].dropna().index]
(td['Exit Price'] - td['Entry Price']) / pip_size

[kernc]

I understand the suggestion here is to amended Position with Position.pl_pips, returning price difference in pips.

[algomaschine]

Dear Kernc,
I'm trying to get the PnL for every trade in pips by pl*pip, but getting slightly different results than I should have, would you mind having a look at my code?

Here's the screenshot of PnL calculated in Excel VS debug data from the Python class. I've attached the screenshot, data (open price/signals in sample.txt) and code. The comparison values are in google sheets here. Your help is appreciated a big time. The python.txt file contains code.

https://docs.google.com/spreadsheets/d/1HYzT7KmVEMtNmHxlgW-qU-t-W-VAmcV8W4V9uQcCdPI/edit#gid=1848616943

sample.txt

python.txt

[kernc]

In pnl_classic, you're computing mere difference from one Close to the next. To get this value (within rounding accuracy) within the strategy, you have to divide P/L with position size (i.e. number of units):

delta_pips = self.position.pl / self.position.size

In order to get this value in terms of number of pips, you have to, additionally, divide by pip size:

pnl_in_pips = self.position.pl / self.position.size / self.data.pip

[algomaschine]

Thanks for the tip. It seems to work now.

[kernc]

As PnL in pips is easily computable, Position.pl_pips deemed not needed.