Get serious about measuring your performance: Sharpe Ratio

[CTKid]

In finance, the Sharpe Ratio can be used to measure the performance of an investment strategy. It is defined as the return generated by your investment(s) above the riskless rate (say the 30yr treasury bond) divided by the standard deviation of your investment(s).

This measure is useful because it is risk-adjusted; a riskier asset must perform better than a less risky one to have the same Sharpe ratio.

After reviewing this thread, its hard to ignore how important both increasing your winrate and decreasing your standard deviation are to anyone who relies on poker for a "steady" income. Long losing or even break even streaks can be hell both financially and psychologically.

In measuring your poker success, you should start thinking not only in terms of winrates in BB/100, but also in terms of your poker Sharpe Ratio, which is:

Poker Sharpe Ratio = Winrate in BB/100 / Standard deviation in BB/100

Interestingly, a very good hedge fund might have a Sharpe ratio of 10. In NL poker, the best one might hope for is probably around 0.2.

[CHAx]

What are we using for our riskless return?

[CTKid]

[ QUOTE ]
What are we using for our riskless return?

[/ QUOTE ]

Up for debate, but as I wrote it I used 0%. It probably doesn't matter, since a typical bond return of 5% or so is totally negligible to any winning player.

[mr weasals]

In poker, probably 0 since there isn't a poker game with 0 standard deviation.

Blackjack players use a calculation like the Sharpe ratio to compute the attractiveness of playing different games.

Is the OP saying that the excess return for these cats is 10 times the standard deviation? If so, wow, that would lead to huge quantities of money in no time. over what time period is that measured? 100 hands of poker is only an hour or less [img]/images/graemlins/grin.gif[/img]

[Borka]

I think that you simply things to much. Required bankroll should count as the investment being made as that is the amout of money that could be placed risk-free.

Here is my take on things.

S=(R-Rf)/std

R=((BB/100)*(hands_p_year/100))/(Bankroll_in_BB)
Rf=30 year treasury= 4.5%
std=standard deviation of the return for the whole year.

An example:
Someone averages 10BB/100 with a std of 120BB/100 and keeps a roll of 100 (100 BB) buyins, and plays 300'000 hands a year. With 300'000 hands the std should be around 6500BB divided by the expected return in BB .

S=(10*3000/10000-0.045)/(6500/30000)=13.6

Which I'm guessing will be very hard to beat using other forms of investing.

[trplthrt]

Risk free rate doesn't matter here, as long as you use the same input for everyone. Sharpe alone doesn't make much sense only when compared to the population you are comparing to. If everyone uses 0 as an implied RF rate for this, the ratio can be used to determine who are the more volatile, less risky, etc players within the same winrate range. Two players can both be running at 10bb/100, but one may be much less risky. I also think this would be a good measure of if a player is 'running hot' or not.. the same two players one with a much lower Sharpe may have just won some big flips and gotten there.

I had always wished that Poker Tracker would incorporate standard deviation in their calcs. I don't think they have it there.

[CTKid]

[ QUOTE ]
I had always wished that Poker Tracker would incorporate standard deviation in their calcs. I don't think they have it there.

[/ QUOTE ]

They do, although I have heard some debate here (by Pokey, etc.) about whether it is calculated properly. Go to Session Notes --> More Detail.

[cheiro]

Shouldn't risk free rate be the rate of [non-poker] investment return you could earn on your poker bankroll? I think you should set this as 5% at least, since that is the rate you can get on a basic savings account.

[trplthrt]

Rate doesn't matter, it only matters in an environment with changing interest rates when your choice is either invest risk-free or invest with manager x.. you can apply the same philosphy, in some way, to poker, but I think the measure is better utilized to simply capture volatility and risk for a given player and given style as compared to his peers. In finance, you need to back check this calc across different periods which all have different risk free rates so you need to be adjusting the figures to compare apples and apples. In this case, for poker, I think it is best used simply as a matter of volatility given the same profit level/winrate.

CT, thanks - I will check it out.

[CTKid]

bump for evening crowd.

Also, it would be intersting to quantify how likely a stretch of X break-even hands is for a Sharpe ratio of Y. This would require some extensive but straightforward simulations.