Expectancy calculated incorrectly
According to the book of Van Tharp the expectancy is the average R, which is something different as the formula used here. The formula currently used is from Van Tharps first book, namely (Probability of Win * Average Win) – (Probability of Loss * Average Loss) .
On page 18 of his newer book “definitive guide to position sizing” he writes that this formula is incorrect because expectancy really should be the average profit per dollar risked. It is best to just divide the profit/loss by the risk per trade and take the average of this to calculate the expectancy.
-
Votes +1
-
Project StrategyQuant X
-
Type Bug
-
Status Fixed
-
Priority Normal
History
Venus
02.09.2019 19:24Subject changed from SQN Score and Expectancy calculated incorrectly to Expectancy calculated incorrectly
Mark Fric
11.09.2019 11:04Status changed from New to Waiting for information
which means average profit / averageLoss, where average loss is used as risk of trade.
Venus
11.09.2019 20:34Attachment Error incorrect expectancy - BTCUSD - 2percAcc.xlsx added
Thanks for your response. There is really a difference, please have a look at the attached XLS to see the difference. I made an export of a SQX backtest (see 2nd sheet). On the first sheet, you will see expectancy calculated in a different way.
Mark Fric
24.09.2019 13:49Can you attach the strategy in .sqx format?
Venus
25.09.2019 09:44Attachment Strategy 10920.sqx added
I calculated the risk and R myself in XLS in the following way:
- Risk: money lost when trade is a loser. In this case "balance" * "risk percentage", but this depends on money management obviously. Does SQX already have a variable in place for "Risk" in ordersList?
- R: "profit/loss" divided by "risk"
Attached the strateqy in .SQX format
Eventually i would like to calculate the following statistic, which tells me something about the smoothness of the equity curve: https://roadmap.strategyquant.com/tasks/sqp_0062 .
Mark Fric
10.10.2019 14:26Attachment Error incorrect expectancy - BTCUSD - 2percAcc.xlsx added
The whole change is that I used Average loss as a risk. We need to be able to compute R expectancy automatically without any further input, and it should work for any money management and stop loss type.
I believe using Avg Loss as a risk is reasonable estimate. I don't know about another way to determine risk that would be valid for any MM or SL.
And when we use it like this then R Expectancy becomes 0.51, just like in SQ X.
Venus
12.10.2019 18:48Thanks for the fix, i understand your thoughts regarding a general way of calculating risk. This is much better as the old situation!
© Copyright. All rights reserved. ProjectPanel.com
