[dtmeyers]

I'm new and not sure if this has been discussed in another thread; if so, the link would be much appreciated.

If I know that Atlanta has a 64% chance to beat San Francisco this Sunday, how would I find the spread (I'd appreciate it if the answer were in traditional math terms instead of written terms from Excel)? Similarly, how would I find the money line given the same win probability?

Thanks in advance

[sharpcat]

win % to ML
1/.64= Decimal odds 1.5625
100/(1-1.5625)=American odds -177.78
underdogs
1-decimal odds

or you can just use SBR tools odds converter.

win % to spread is much more difficult. This would require a large database and you would need to figure out the historical win % of teams when favored by 0, .5, 1, 1.5, 2, 2.5, 3, etc..

or you can use SBR tools spread to ML converter.

[Shonner]

I'm new and not sure if this has been discussed in another thread; if so, the link would be much appreciated.

If I know that Atlanta has a 64% chance to beat San Francisco this Sunday, how would I find the spread (I'd appreciate it if the answer were in traditional math terms instead of written terms from Excel)? Similarly, how would I find the money line given the same win probability?

Thanks in advance

These types of conversions are not very accurate. First, to figure out the spread of a team with 64% chance to win, first use the http://www.sportsbookreview.com/betting+tool...converter.aspx and input .64 into the implied probability field, that gives you ML odds of -177.78. This is with no Vig. Once you get that, you can go to the http://www.sportsbookreview.com/betting+tool...converter.aspx and play with the spread until it returns the ML of -177 that you were looking for, it's not going to return it exactly because remember ML's are not solely a function of a spread, there's a lot more that goes into that, mainly the Over/Under. The spread/ML converter is not accurate, as documented in many threads primarily because it doesn't take into account the over/under.

You can use those two tools to go back and forth from probability/odds/ML to get what you need. Again, use with caution.

GL though

[uva3021]

i made a inverse pythagorean calculator a while back, that uses the win probability, total, and a desired exponent and produces a spread

http://sportsobjective.com/wordpress/?page_id=3529

the calculations involved are explained here http://sportsobjective.com/wordpress/?p=3554

[PRC]

cool stuff uva, thanks man

[DARKERIDER]

hey uva, im not the brightest cookie on the planet.. but can you explain it a little dumber for me??

[DARKERIDER]

cant work out what an exponent is :S

[MadTiger]

cant work out what an exponent is :S

It's being used as a scaling factor, more or less.

On the website, it says that different numbers are used for different sports, and that you need to search the site to find the exponents to use for various sports.

[uva3021]

exponents

MLB: 1.8-2

NFL: ~2.37

NCAAF: ~2.24

NBA: ~14-16.5

NCAAB: ~8-12

NHL: ~1.86-2

for totals, as standard generic values just enter

MLB: 9

NFL: 40

NCAAF: 55

NBA: 190

NCAAB: 135

NHL: 5.5

[Pancho sanza]

Just curious as to how you got the exponent for the NFL

thanks

[Peregrine Stoop]

very nice stuff uva
It's greatly appreciated

[Pokerjoe]

Best fit exponents are public. Google around.

[Pokerjoe]

The problem with this approach is that it assumes points are equal in value.

For example, the difference between -2' and -3' is huge in NFL games, but a mere 1 point in pyth estimations.

[Pokerjoe]

The same problem arises with the use of power ratings. Power ratings estimate ability differences with generic points, of equal value; the line is made up of specific, actual points, with varying value.

[Pokerjoe]

Very nice site, btw, UVA. Everyone should check it out.

[Peregrine Stoop]

Very nice site, btw, UVA. Everyone should check it out.

agree completely

worried about the recent post that mentioned the closing price as the stoppage of variation and implied attaching more strength to the opening line.

[Flight]

The same problem arises with the use of power ratings. Power ratings estimate ability differences with generic points, of equal value; the line is made up of specific, actual points, with varying value.

I always check if a power rating vs market line crosses key numbers such as 3 and adjust accordingly (ie as a bad example, if Sagarin has -4 and Market has -2, I consider Sagarin off-market by 3 points)

I do not however build this directly into any power ratings because I use linear conversions. I guess more advanced power ratings could take into account point value and possibly even use the market total for exact point valuation.

[Flight]

OP: For 64% NFL probability, spread is about -4, no vig ML -180