Mathematical expectation

Visaria · Jul 30, 2014

Maverick74 said:
No,no, no, there is no guessing. This is not TA.

Tell me specifically what you are trying to do because the posters on this thread are talking about options and p&l and win rates, etc. Tell me specifically what you are trying to solve for.

Ok, let's say i have a strategy for futures trading. Is it possible to determine the ev, positive or negative of this strategy?

kut2k2 · Jul 30, 2014

Visaria said:
Can expected value be calculated in trading like one can calculate the (negative) expected value of casino games such as blackjack or roulette?

Just add up your trade returns and divide by the number of trades. That's your expected value aka expectation (no such thing as 'expectancy'). Also no need to calculate variance.

Visaria · Jul 30, 2014

kut2k2 said:
Just add up your trade returns and divide by the number of trades. That's your expected value aka expectation (no such thing as 'expectancy'). Also no need to calculate variance.

that is historical. i want to know the ev going forward. is it possible? i suspect not.

Maverick74 · Jul 30, 2014

kut2k2 said:
Just add up your trade returns and divide by the number of trades. That's your expected value aka expectation (no such thing as 'expectancy'). Also no need to calculate variance.

This is false. That is your mean. Expected value requires you to assign probabilities to all the possible outcomes and sum up those values. To solve for the probabilities you will need to know the variance. Simply take a mean of anything is NOT what you want to do.

Maverick74 · Jul 30, 2014

Visaria said:
Ok, let's say i have a strategy for futures trading. Is it possible to determine the ev, positive or negative of this strategy?

The best way to do this is with a monte carlo simulation.

kut2k2 · Jul 30, 2014

Maverick74 said:
This is false. That is your mean. Expected value requires you to assign probabilities to all the possible outcomes and sum up those values. To solve for the probabilities you will need to know the variance. Simply take a mean of anything is NOT what you want to do.

Dead wrong. I'll leave it to you to figure out why.

Visaria · Jul 30, 2014

Maverick74 said:
The best way to do this is with a monte carlo simulation.

what would my inputs be?

Maverick74 · Jul 30, 2014

Visaria said:
what would my inputs be?

There are several ways to go about this depending on how technical and deep you want to go. If you want to test how "robust" your strategy is, I would test each of the parameters that you use to generate a trade signal which could be done through the OLS method.

If you want to test the "results" of your strategy and measure your actual return vs what it's true expected value is, run the monte carlo. The monte carlo can provide very accurate feedback on "expected drawdown" vs actual and a more accurate expected value by manipulating some of the trade data. All that is required with this approach is imputing your actual trades and deciding how many trial runs you want to use.

Visaria · Jul 30, 2014

Thanks Mav and others too. Going to leave this for the time being, I'm going on vacation tomorrow and will look into this when i come back.

Maverick74 · Jul 30, 2014

Visaria said:
Thanks Mav and others too. Going to leave this for the time being, I'm going on vacation tomorrow and will look into this when i come back.

Don't be scared by this stuff. There is a lot of great software out there that makes this stuff very easy and the results can be eye opening. Good luck!