Integrating forecasts

[mwilding1981]

Can anybody suggest the best way to integrate betting forecasts with live marketdata in order iron out any rogue ratings and prices due to freak readings or low sample sizes?

[Handiman]

Could you speak English please, for us just fresh off the boat, so we may understand better what you're asking? :lol: Honestly not sure what your getting at.and I would like to understand. So when someone answers you I'll understand maybe what they are saying.


Alot of smart people here and some of the communications slip right on past me.

Handi

[GameTheory]

He is asking, "How do you take your initial predictions for the horses in a race and make them better in the face of the live tote board odds?" (Or "market data" on Betfair.) In other words, smoothing, which we've talked about lots around here before.

And the short answer is, you combine them in some way. If you have a horse that you're initially giving 2/1 as fair odds and it is 40/1 on the toteboard, you've probably got a "rogue prediction" on your hands and need to downgrade it so you're giving it less of a chance. By how much? Well, that's the question, isn't it?

[skate]

at 40/1, id upgrade the rogue and not back off,as in not.

[Handiman]

Ah! Grasshopper understands now.


Thank You.

Handi

[mwilding1981]

Not sure why I didn't check the boards before posting about this, normally give everything a thorough read first. Will scour and report back my findings.

Apologies

[sjk]

No need for apologies.

You either form a composite line as a linear combination of your probabilibites and the public probabilities or demand a larger multiplier overlay of the price offered vs your calculated expected price. I have always thought these were mathematically equivalent.

Unless you have a significant amount of data which shows that your probabilites are fairly accurate the above is of no value to you. I think the flip anwsers come from scepticism as to how accurate your starting probabilities are.

[GameTheory]

at 40/1, id upgrade the rogue and not back off,as in not.He very well may still be worth a bet, maybe a big one, but in that situation you're not going to win 1 of 3 as your original estimate estimates. You're just not.