Quote:
Originally Posted by DMMx69
do you cap other leagues largely the same as premier? years ago i had a large DB that included every league and didnt distinguish leagues. i know some are higher/lower scoring or have higher variations in level of play, but it seemed to me you could use largely the same metrics from one to another?
general thoughts on the matter?
the plan is to use the same model. football-data.co.uk has historical results/odds from almost every european leagues 1st and 2nd divisions. so far i have only backtested the epl, championship and la liga. the epl actually was the most profitable although it was a cursory backtest. im gonna keep on working on a better backtest and check out other leagues. the nature of the model makes it almost impossible to do a true backtest (well possible but would take at least 24 hours to run on my computer).
more variation in the level of play could be bad but less variation in the level of play should actually make the model more accurate. this is just due to how my model goes about generating scoring intensity parameters. it assumes all parameters are from the same distribution so less variation is better. im working on a slightly different model to deal with that issue.
my biggest hesitation about trying to cap other leagues is that i dont closely follow any league other than the epl. from my initial backtests it seems that just going based off the model can be profitable but id rather just cap leagues where i can justify the picks im making not just being a number that shoots out of a model
unrelated question. does 5dimes have a xml feed for odds? i do a ton of record keeping for not just all of the picks i make but every game in the epl and being able to scrape data would save me hours of data input
