Forums

Lloyd Christmas

cptbigsack

Lloyd buddy, long time no talk. How ya been? Anyways I am working on creating some push charts, and I only have data from 00-10 and wondering if this will be valid. I am grouping the spreads by totals, and I am concerned that with all the helmet to helmet rules and what not my data might be off. Should I be worried?

I have not heard any real talk concerning the helmet rules and how they have affected data mining.  Don't have a good answer for you there.

Also I plan on using spreads then calculating push percentages off that. Say spread is 2.5 and thinking about playing a teaser, I will look at 3,4, etc for this spread only. I know some people window but I am not sure of the reasoning behind this. I figure it will give more data but I think what I am doing will give more accurate results.

Yes, that is the exact reason that people window, is for more data.  I'm sure it's buried in one of the Wong, Feustal, Yao books regarding how much windowing reduced the accuracy of your results, I can't remember though.

Any help would be appreciated. I plan on returning to Kashyyyk at start of NFL season and if I don't have anything I will definately be back for CBB. Thanks

I'm not sure what your K word means.

Ha sorry man, not many good answers this time

not to sound like a nerd, but Kashyyyk is name of planet the Wookies live on in Star Wars

This is probably the one time I didn't first Google something I didn't know.  ooops

I actually did find some answers in fuestals book. Windowing is used to get larger samples. In Elihu's book he talks about the economic tradeoff. Say the game has a spread of -3 and a total of 40, if you only examined the push percentage in those games the data would be more accurate but with a small sample size the standard deviation would be too great making the data useless. So I guess I have to deem what I will accept for error. Although lower totals will have different push rates then higher totals, the helmet to helmet rules should not affect push percentages. Think about it, this is already factored into the line, so a spread of 3 and a total of 64 in 03 should have the same push percentage of a spread of 3 and a total of 64 in 2012, ya dig?

cptbigsack

I actually did find some answers in fuestals book. Windowing is used to get larger samples. In Elihu's book he talks about the economic tradeoff. Say the game has a spread of -3 and a total of 40, if you only examined the push percentage in those games the data would be more accurate but with a small sample size the standard deviation would be too great making the data useless. So I guess I have to deem what I will accept for error. Although lower totals will have different push rates then higher totals, the helmet to helmet rules should not affect push percentages. Think about it, this is already factored into the line, so a spread of 3 and a total of 64 in 03 should have the same push percentage of a spread of 3 and a total of 64 in 2012, ya dig?

I dug

Great thread here. Lloyd  in college ball i can get bama -30 1/2 and utep +32 1/2.  Now using the push frequency chart according to the numbers, you could make this a play right?{asked this question first just to make sure im understanding chart}.  Assuming u can, my next question is with big spreads like this do u think sample size is large enough to make charts #'s accurate on them?

This will be one of the most valuable threads come football season.

WFCphilly

Great thread here. Lloyd  in college ball i can get bama -30 1/2 and utep +32 1/2.  Now using the push frequency chart according to the numbers, you could make this a play right?{asked this question first just to make sure im understanding chart}.  Assuming u can, my next question is with big spreads like this do u think sample size is large enough to make charts #'s accurate on them?

Great second point you make here.  The hit rate on middle both the 31 and 32 is 6.4%.  But you make a great point, the sample size is going to be waaaay smaller on these larger spreads and extremely high and low totals.  I wish I had a better answer, but the only thing I could suggest is just being more conservative and making sure you are well beyond that 4.76% you need to break even.  1.64 is certainly a good cushion, so I'd say you're safe with this one.  Good luck!

thanks lloyd. appreciate it.

Teasing home teams either way through 3&7 is a 76.7% (over 80% last year) proposition since 1989.  As of now four games fit this criteria in week 1, lets call them A,B,C,D.  My question is, should i put every comination of these four into six different teasers? (AB,AC,AD,BC,BD,CD).  I'm pretty sure the answer is yes but i was wondering if you could show any math to support this.  My one reservation was that if just one of the teasers loses, i would be losing money.  Thanks for any help Lloyd

Is there any way Romney can win the election?