# Probability of outcome of Sporting event

• May 29th 2007, 12:16 AM
alex
Probability of outcome of Sporting event
Hello,

I have a question or problem to resolve and I am not an expert in math so if this isn't the right forum to post this, apologies.

Problem:
How can we evaluate the probabilities of the result of a sporting event based on the current score.

For example.

Lets say the probabilities of team A vs. team B winning the event at the start were the following, team A 70% and team B 30% (No draw or tie possible).

The first team to score 21 points would win the match (no tie-breaks or win by 2 clear points rules)...

How could we calculate the relative probabilities of the outcome during the game as the points change...

For example, after 10 minutes of exciting play....
Team A 5-1 Team B...
what would be the probabilities of Team A winning considering it started the game with 70% probabilities and is now winning 5-1? etc...

The 3 variables would be.

- Starting probabilities
- total points needed to win
- Current score

Thanks
Alex
• May 29th 2007, 12:33 AM
JakeD
Quote:

Originally Posted by alex
Hello,

I have a question or problem to resolve and I am not an expert in math so if this isn't the right forum to post this, apologies.

Problem:
How can we evaluate the probabilities of the result of a sporting event based on the current score.

For example.

Lets say the probabilities of team A vs. team B winning the event at the start were the following, team A 70% and team B 30% (No draw or tie possible).

The first team to score 21 points would win the match (no tie-breaks or win by 2 clear points rules)...

How could we calculate the relative probabilities of the outcome during the game as the points change...

For example, after 10 minutes of exciting play....
Team A 5-1 Team B...
what would be the probabilities of Team A winning considering it started the game with 70% probabilities and is now winning 5-1? etc...

The 3 variables would be.

- Starting probabilities
- total points needed to win
- Current score

Thanks
Alex

Hello. Is this for wagering?

You would need to analyze data (and lots of it) about how the scoring progressed in past real sporting events of the kind you're interested in. If you don't have that data, you could make something up out of your imagination, and that might be fun, but it wouldn't be very predictive for a real sporting event and would be of no use in wagering.
• May 29th 2007, 12:46 AM
alex
it doesnt really matter for what it is for I would have thought, its a pure probability question based on starting probabilities and how they change based on different events occuring.

The algorithm should not be influenced by previous data or historical information, as that has already been done (i.e. starting probability)

Thanks
• May 29th 2007, 05:04 AM
Plato
This may well be far off the mark, but it seems to me as if you are describing a variation of the ‘problem of points’.
If you look at the two links below then you can tell us if that is the sort of thing you have in mind.
Math Forum: Probability and the Problem of Points

Pascal's Triangle - the Problem of Points

In any case I have a question about the phase “(no tie-breaks or win by 2 clear points rules).” No ties is clear but what about win by 2 clear points rules?
Are you saying that on each match there is a clear winner of one point; the first team to reach 21 total points wins?

If so, then this does have some parallels with the “problem of points”. If team A has 10 points and team B has 4 points then by playing 27 more matches we insure that one of the two teams must win. We can find the number of cases in which A wins.
• May 29th 2007, 07:47 AM
alex
hi,

>>Are you saying that on each match there is a clear winner of one point; the first team to reach 21 total points wins?

YES.

>>If so, then this does have some parallels with the “problem of points”. If team A has 10 points and team B has 4 points then by playing 27 more matches we insure that one of the two teams must win. We can find the number of cases in which A wins.

Possibly, but I would have thought it was more related to the probabilities of points left to play, with % of points needed to win and apply them to original probabilities of winning...

Thanks again
Alex
• May 29th 2007, 03:28 PM
Plato
Quote:

Originally Posted by alex
Possibly, but I would have thought it was more related to the probabilities of points left to play.

You are absolutely correct about that.
• May 29th 2007, 08:12 PM
CaptainBlack
http://www.mathhelpforum.com/math-he...ent-points.gif

This type of analysis depends on the assumption that the probabability
of the result of the next game/point is independent of the current score or
the path through score state space taken to get to the current score.

This is OK for a homework exercise, or a system with little human involvement
or skill, but in a real game this is often unlikely to be a valid assumption.

RonL
• May 30th 2007, 12:25 AM
alex
Thanks for the answers, I believe that is correct.

I agree that the assumption is fairly basic and does not contemplate other factors such as historical or psycological factors, however the idea of the excericise was to complete the first phase.

Cheers
Alex
• May 30th 2007, 05:34 AM
Plato
Quote:

Originally Posted by CaptainBlack
This type of analysis depends on the assumption that the probabability of the result of the next game/point is independent of the current score or the path through score state space taken to get to the current score.
This is OK for a homework exercise, or a system with little human involvement
or skill, but in a real game this is often unlikely to be a valid assumption.
RonL

Absolutely correct. And that is why I never do statistics; I stick with mathematics.

For years I have found comfort in the following quotation by Einstein in his 1923 essay Geometry and Experience: “As far as the laws of mathematics refer to reality, they are not certain; and as far as they are certain, they do not refer to reality”.