Is there a way to infer total number of goals scored from the total?

[JustinBieber]

For example lets say Bookmaker X has perfect totals for bundesliga games and the O/U for this specefic game is Over 3 goals -141
Under 3 goals +125

The No vig line would come to around 132~. However is there a way to know how many times 3 goals will be scored as this only tells me the chances of above and below 3 goals. Unless it would be the chance between the -141 and +125 but this seems a bit small? I think I read somewhere that poisson is not suitable for this type of thing but I am unsure.

[Data]

Go to BetFair and get the numbers for O/U 2.5 (or 3.5). With those numbers and simple calculations that I'll leave up to you to do, you can get a good estimate of 3 goals scored.

[JustinBieber]

Well I chose a random South American game, The implied probability from 1.61 means U 2.5 goals 62% of the time and O 2.5 goals 2.52 which means over 2.5 goals at 40% of the time. I dont know how im meant to infer 2 goals scored from this.

[Pokerjoe]

For example lets say Bookmaker X has perfect totals for bundesliga games and the O/U for this specefic game is Over 3 goals -141
Under 3 goals +125

The No vig line would come to around 132~. However is there a way to know how many times 3 goals will be scored as this only tells me the chances of above and below 3 goals. Unless it would be the chance between the -141 and +125 but this seems a bit small? I think I read somewhere that poisson is not suitable for this type of thing but I am unsure.

There are a few ways to do it.
a) Build a database of exact-score odds for the various totals, assume linemaker accuracy, and deduce. So if a game's total is 3, total up the odds of exact scores totalling 3 (adjust for the heavy juice, obv).
b) Build a database of actual scores from games with the various totals and look for linemaker inaccuracy.
c) Cross check a DB against Poisson estimates and check for fit.
d) Build a database of dropdown box lines; deduce.

[JustinBieber]

Yes I obviously don't have a database. I just asked if there was a way to infer it from the total goals. Getting an estimate of X goals scored is obviously easy but is there a way to say X team will win by X goals x% of the time?

[Pokerjoe]

Yes, as I said, use the correct-scores odds. If the books are saying the odds of A beating B by exactly 2-1 are y% (after accounting for juice), there you go. If you want the chance of A winning by one goal, also add in the odds of winning 1-0, 3-2, 4-3, etc. What else are you looking for?

If you want to beat those correct-score odds by using derivative estimates, then you have a lot of work ahead of you, and I pointed you in the right direction. But no one currently able to beat derivatives is going to just surrender the answer to a random on a forum.

[MonkeyF0cker]

Are you implying that Betfair markets tend to offer bias in derivatives, Pokerjoe? I believe Data just gave him a suitable solution.

[Data]

Yes I obviously don't have a database. I just asked if there was a way to infer it from the total goals. Getting an estimate of X goals scored is obviously easy but is there a way to say X team will win by X goals x% of the time?

Use the same approach as I suggested above. This time look for Asian Handicap markets to get the numbers you need.

[Data]

Yes, as I said, use the correct-scores odds. If the books are saying the odds of A beating B by exactly 2-1 are y% (after accounting for juice)

The problem with this is that juice is too large in the exact scores markets making those markets unusable for estimating true odds.

If you want the chance of A winning by one goal, also add in the odds of winning 1-0, 3-2, 4-3, etc.

While this is a wrong approach for a beginner due to unability to get the good odds estimates of the exact scores, it will certainly work much better on an intermediate level. I refer the interested reader to this thread:

http://forum.sbrforum.com/handicapper-think-tank/171663-looking-some-help-hockey-winning-margin-models.html

[JustinBieber]

ty data and dont worry I am not stupid enough to take that correct-score advice.