Spread to Moneyline for the NFL or NBA?

[Poli]

Does anyone have any REAL averages? It would be so helpful for me personally (and many other people I am sure).

Spread ML Converter

This calculator seems to be pretty spot on at low spread numbers, but become pretty inaccurate by just around -7 or so.

[uva3021]

could just do a regression for a full slate of games one day, regressing ML onto spread and total, to come up with the formula

sample size would be immaterial in that case because I doubt the bookies would drastically change their ML formula from one day to another, just make sure you grab a sufficient range of prices (double digit favs, mid to high favs, low favs, etc...)

A typical NBA saturday should do the trick

or when in doubt use Pythagorean (though keep in mind one game pythagorean calculations are not conditions the formula assumes, rather its a season expectation of winning percentage)


Total / 2 - Line/2 = Points scored/against

Points scored ^ a / (Points scored ^ a + Points against ^ a)

NFL a = 2.37
NBA a = 14

Total / 2 - Line / 2

[Poli]

I only half understand what you mean with the pythagorean stuff (which team's points scored? where do you determine a line, just halfway?), but I have no idea how to do a regression. Lines also do vary a lot, especially in the NFL at lower levels. That is why I would like real averages if anyone knows a place to find them. Thanks for the response though.

[uva3021]

Originally Posted by Poli

I only half understand what you mean with the pythagorean stuff (which team's points scored? where do you determine a line, just halfway?), but I have no idea how to do a regression. Lines also do vary a lot, especially in the NFL at lower levels. That is why I would like real averages if anyone knows a place to find them. Thanks for the response though.

well yeah what I'm saying is find a day with range of similar prices and highly variable prices

a line of -3 with a total of 45 for game A will have basically the same ML as game b with the same parameters. There will be a central tendency with movement contingent on money.

I figured the line was implied since you are looking for a spread to ML converter. For example:

Total 40
Line -3

Points scored = 40/2 - (-3)/2 = 21.5
Points against = 40/2 - (+3)/2 = 18.5

Exponent = 2.37

21.5^2.37/(21.5^2.37 + 18.5^2.37) = 58.81% = -143

ML = -143

But like I said before Pythagorean is a measure of aggregate win expectation, not typically used for one game scenario. So my suggestion was to gather an adequate sample size, plug the Total and Line parameters into the Pythagorean formula, and find the exponent that best fits the actual respective MLs. Then for future single game calculations you have your exponent

[jarhead60]

I ran a regression for NFL that shows

-35.7 + 51.86*point spread = $ line

R^2 = .92
very low p value
standard error 104

it puts one in the ball park

home field appears to be a factor that is not included in the above

good luck

[uva3021]

thanks jarhead, did you not include total in the regression?

[Poli]

uva, thanks for the further explanation, I get it now. You are saying that I should solve for the exponent myself, do you mean for each spread number or is the goal still to get one catch all exponent?

Originally Posted by jarhead60

I ran a regression for NFL that shows

-35.7 + 51.86*point spread = $ line

R^2 = .92
very low p value
standard error 104

it puts one in the ball park

home field appears to be a factor that is not included in the above

good luck

Pretty far off at low numbers, but pretty accurate at higher levels. Home field is already kind of factored into spread numbers, so not really an issue right?

Anyway, thanks you guys, this helps. I still wish I could get real numbers though, I just wanted something that could be used to correlate to real money values since sportsdatabase only tracks $$ numbers on the MLB moneyline. I saw in another topic here where someone just pasted the bodog supposed line stuff, but those are really vague/bad lines (not that the user was not trying their best).

[uva3021]

one exponent for all lines using the pythagorean formula

I highly suggest using a regression though, find as many years of spread and totals in the NFL/NBA as you can and run a regression using Stata or Excel

[Poli]

Right but if I had data, then I could just sum up moneylines for myself assuming those were included...I do not really understand excel well enough to do anything like your getting data from statfox.

[jarhead60]

poli,

free data can be found at this site

FootballLOCKS.com: NFL picks, weekly pro football lines, odds & point spreads. Expert NFL betting predictions.

[Poli]

Sorry to "bump", but I forgot to look at this thread until recently. Thanks a lot jarhead60, I will do something with all that when I have the time (NFL season is already over after all :P)