So i made an excel. Just before the update of this night. But i challenged the bot. So im logging

[BGB2k11]

so its in dutch but the main takeaway:

• Every bet is 5 euro. so red is -> no bet or lost bet. when the space in the odd screen is "white" i am not betting it. This means the AI and the XG model are contradicting itself.

ML - 50 euro bets on 10 matches -> so 5 euro a match. Revenue of 60euro -> profit of 10 euro
UO - 30 euro on 6 matches -> 56,7 euro win -> profit of 26,7 euro

When we take EV into account:

Again 5 euro lay in

only 1 bet cause the EV is not worth it
ML - 1 bet 1 win -> 7.4euro reventue -> 2,4 euro profit
UO - 6 bets 6 wins -> 47,3 reventue -> 17,3 euro

if would have combined both bets -> so with EV and without EV combined
Lay in: 115 euro
Revenue: 171,45euro
Profit: 56,45 euro

This is a 49% folio gain. If your folio was 115 euro :)
Not bad for a 1 nighter

[MaoZeLone]

This log is very interesting and helpful. Would you like to make similar records in the future？

[BGB2k11]

making one each day - except for the weekends since i can not log it in the weekends for now.

[BGB2k11]

Matchday 31/03 (this is 31/03 my time -> CET)

lost my Excel status. So its a printscreen :(

[nealmick]

I think retraining models everyday might not be the best idea. You want a single model so that you can get a decent sample size of tests... Each time you retrain a model the weights are reset, and recalibrated over the training process... effectively making a new and unique model that will most likely get different results on the same games (at-least by a little).... Also the training data is not updated daily, covered in this discussion #155

[BGB2k11]

I think retraining models everyday might not be the best idea. You want a single model so that you can get a decent sample size of tests... Each time you retrain a model the weights are reset, and recalibrated over the training process... effectively making a new and unique model that will most likely get different results on the same games (at-least by a little).... Also the training data is not updated daily, covered in this discussion #155

So this is for the AI right ? The XG needs to be retrained every day right ?

[nealmick]

Nope, XG works the same way as far as I know, and would have the same issue if retrained daily….

[BGB2k11]

Nope, XG works the same way as far as I know, and would have the same issue if retrained daily….

Thanks for the insight !
After how many days do we need to retrain then ?

[nealmick]

I don't think the NBA game changes much in the course of a few months, or even a few years...