Jump to content
British Speedway Forum

Time For A Rethink In Rolling Averages?


Recommended Posts

 

 

What do you think about my idea of using standard deviation as a measure?

 

 

 

Presumably with all teams building to 0 the next season although some tolerance would be needed or it would be impossible.

 

This would require no change in team numbers and for all riders to be the same from one season to the next to work in its purest form.

 

Any deviation in either factor would render the figure inaccurate.

Link to comment
Share on other sites

Presumably with all teams building to 0 the next season although some tolerance would be needed or it would be impossible.

 

This would require no change in team numbers and for all riders to be the same from one season to the next to work in its purest form.

 

Any deviation in either factor would render the figure inaccurate.

 

OK, so you are saying there would need to be the same number of teams, and the same number of riders, for it to be meaningful.

Actually, I think you are also saying it must be the same pool of riders too.

 

Would it actually need to be the case though? I'm sure you know there are variations in ways of calculating the standard deviation for a sample.

I'm not sure about the answer, though, and am open to words of wisdom greater than mine.

 

You wouldn't necessarily need to have team building set to 0.

For example, you could have limitations build around deviation from the mean.

 

I think it might be quite interesting, if nothing else.

Link to comment
Share on other sites

 

 

OK, so you are saying there would need to be the same number of teams, and the same number of riders, for it to be meaningful.

Actually, I think you are also saying it must be the same pool of riders too.

 

Would it actually need to be the case though? I'm sure you know there are variations in ways of calculating the standard deviation for a sample.

I'm not sure about the answer, though, and am open to words of wisdom greater than mine.

 

You wouldn't necessarily need to have team building set to 0.

For example, you could have limitations build around deviation from the mean.

 

I think it might be quite interesting, if nothing else.

I agree it is an interesting concept but is probably too complicted to be considered for implementation.
Link to comment
Share on other sites

I agree it is an interesting concept but is probably too complicted to be considered for implementation.

 

There are so many uncontrollable variables that would totally knacker it, but that is what happens with the current system, I suppose.

It might make it a lot more difficult to manipulate rider ratings for team building.

 

In terms of the public, then it could be presented as below average, average, above average, and maybe extreme deviant?

Link to comment
Share on other sites

Extreme deviant?...are you still talking statistics or have you been speaking to some of the women I used to know!!!

Link to comment
Share on other sites

Would it be more meaningful to have a league average that is calculated from all riders, and all rides: then each individual is given a standard deviation rather than an average?

 

Off the top of my head no, but i can't see what the advantage of that using SD is against +/- the mean average.

 

RA's are fine once one finds the sample size sweet spot that balances the responsiveness of the average moving with performance against its real value.

Edited by Dekker
Link to comment
Share on other sites

Off the top of my head no, but i can't see what the advantage of that using SD is against +/- the mean average.

 

RA's are fine once one finds the sample size sweet spot that balances the responsiveness of the average moving with performance against its real value.

 

Hmmmmmmm, your final sentence, explain to me how you do that...

 

:rolleyes:

Link to comment
Share on other sites

the variance of the average is proportional to the sample size, i.e. the more data points in the set then the increased sample size dampens the variation.

 

but i guess you knew that...

 

its just a question of picking the right sample size say near 3/4 of the average number of meetings a rider does per season.

Link to comment
Share on other sites

It's the square root that caues the problems IMO :blink:

It's not, actually, it solves many problems.

One of my reasons for thinking it may be useful in speedway averages is that the squared function emphasis the extremes.

 

the variance of the average is proportional to the sample size, i.e. the more data points in the set then the increased sample size dampens the variation.

 

but i guess you knew that...

 

its just a question of picking the right sample size say near 3/4 of the average number of meetings a rider does per season.

Indeed, but you would want to have some sensitivity to variation within the sample set in this case.

Or maybe not?

Link to comment
Share on other sites

Indeed, but you would want to have some sensitivity to variation within the sample set in this case.

Or maybe not?

 

you want a balance between sensitivity and accuracy, play around with sample size to taste.

Link to comment
Share on other sites

you want a balance between sensitivity and accuracy, play around with sample size to taste.

I agree, I suppose we should be talking about precision rather than accuracy too.

 

I don't agree that accuracy is proportional to sample size. It surely cannot be a linear relationship?

Link to comment
Share on other sites

I agree, I suppose we should be talking about precision rather than accuracy too.

 

I don't agree that accuracy is proportional to sample size. It surely cannot be a linear relationship?

 

i'm trying to explain in layman's terms so people can understand the principles involved, i'll leave the formulae till later, generally speaking more data points improve the average calculated but too many make the average very slow to follow form. It's a balancing act to get it right, rather than picking a seemingly arbitrary number someone should have modelled 2 or 3 years worth of green sheet data into some test case models to see what sample size looked good.

Edited by Dekker
Link to comment
Share on other sites

i'm trying to explain in layman's terms so people can understand the principals involved, i'll leave the formulae till later, generally speaking more data points improve the average calculated but too many make the average very slow to follow form. It's a balancing act to get it right, rather than picking a seemingly arbitrary number someone should have modelled 2 or 3 years worth of green sheet data into some test case models to see what sample size looked good.

 

I lost the will to live about 6 post's ago how's that for accuracy or was it layman's terms or it might be precision or will i just get my coat :lol:

  • Like 1
Link to comment
Share on other sites

i'm trying to explain in layman's terms so people can understand the principles involved, i'll leave the formulae till later, generally speaking more data points improve the average calculated but too many make the average very slow to follow form. It's a balancing act to get it right, rather than picking a seemingly arbitrary number someone should have modelled 2 or 3 years worth of green sheet data into some test case models to see what sample size looked good.

 

One has to posit the question though, is average the correct function if the main aim is to analyse form?

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. Privacy Policy