New Feature: Batting Medians

[ganeshran]

Why not calculate standard deviation if you want to assess the impact of outliers?

 

[chasingthedon]

Largely fair comments above - especially that it's more a measure of reliability than ability. I had intended originally to make the feature a kind of consistency index with the median being combined with the skew, but as I mentioned the skew is unfair on those with bigger innings and, therefore, unfair to the best batsmen.

And yes, no one will ever adopt this. Thought it was interesting though.

At best another partial success.

 

[Howe_zat]

It's just occurred to me that batting medians would actually be of tremendous use in gambling (in the cases where you're betting over/under a certain amount a batsman will score when he comes in to bat). In fact, I wouldn't be surprised to find that bookmakers use data like this to set the lines in the first place.

 

[Cevno]

Good read, IMO.

 

[morgieb]

Haha Phil Jaques has the 3rd highest median. Gun.

 

[weldone]

It's just occurred to me that batting medians would actually be of tremendous use in gambling (in the cases where you're betting over/under a certain amount a batsman will score when he comes in to bat). In fact, I wouldn't be surprised to find that bookmakers use data like this to set the lines in the first place.

Yeah absolutely...

 

[smash84]

AdamC is absolutely right mate

oh ok. My concept of median was a little screwed up

 

[Migara]

Why not calculate standard deviation if you want to assess the impact of outliers?

STD only useful when the distribution is near Gaussian. Otherwise it's a percentile that has to be used in a highly skewed distribution like this.

 

[weldone]

STD only useful when the distribution is near Gaussian. Otherwise it's a percentile that has to be used in a highly skewed distribution like this.

Even if batting scores follow a lognormal distribution (not sure if they do - but lognormal should be closer to it than normal for sure), we can always arrive at the %ile figures given average and stdev.

 

[Migara]

Even if batting scores follow a lognormal distribution (not sure if they do - but lognormal should be closer to it than normal for sure), we can always arrive at the %ile figures given average and stdev.

True, point taken from statistician viewpoint. But I am not sure how a lay person would understand mean log batting average and standard deviation of log batting scores.

 

[ganeshran]

STD only useful when the distribution is near Gaussian. Otherwise it's a percentile that has to be used in a highly skewed distribution like this.

How to calculate percentile given his scores? What is the formula?

 

[Migara]

How to calculate percentile given his scores? What is the formula?

Percentile - Wikipedia, the free encyclopedia

25th percentile is the median of the 1st half of the series. 75th is the median of the latter half. 50th is the median.