Quote:
Originally Posted by SwoopAE
Do you have the data for how often the favourite steams compared to the dog for opening vs closing lines?
No, I don't have this yet. I will try to get it, but right now it was a lot easier just to get the closing.
Quote:
Originally Posted by Malachii
How do you determine who the best predictions are?
I use a scoring criteria which rewards confidence when you're right, and punishes it when you are wrong. So, if you say "I'm sure Aldo is going to beat Edgar", and he does, you score well. Iif you say "I'm sure Aldo is going to beat Max", you would be punished. If instead you said "I'm not confident, but I think Aldo will take it against Max", you would get a less punishing score.
This is what I judge my model on in general (the average of these scores), as (over large samples) accuracy (i.e. predicting 60% for events with a real probability around ~60%) is rewarded. This places the idea of getting the right price right there in the evaluation of the model, which is very convenient.