search

dscotthunter / Fantasy-Hockey-IP-Code

Other
104 stars 76 forks source link

LoadError: UndefVarError: i not defined #9

Open awiden91 opened 4 years ago

awiden91 commented 4 years ago

I am trying to run this code on JULIA 0.4.17 and am having issues. I understand that there has been a lot of updates on JULIA and JuMP from when the code was created until now. Would you have any insight in helping me re-write this code for the current version? Thanks for the help, I appreciate your time.

Ryanklingert commented 4 years ago

I spent a few days going at it and couldn’t get it working. Ended up recreating the entire thing in Matlab only to find that the returns when backtesting over the season so far were pretty subpar. Still playing with it but my impression is that to many professional players have adopted similar strategies in the years since this paper was written

awiden91 commented 4 years ago

Got it, thank you for your input. Would I be able to check out that code for Matlab and play around with it? Thanks again, Ari

Ryanklingert commented 4 years ago

Do you have access to the optimization toolbox?

awiden91 commented 4 years ago

I am a student and think I can get a copy, if not I can download a trial. Would it be possible to run it in Python?

Thanks again, Ari

On Fri, Jan 3, 2020 at 9:38 PM Ryanklingert notifications@github.com wrote:

Do you have access to the optimization toolbox?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/dscotthunter/Fantasy-Hockey-IP-Code/issues/9?email_source=notifications&email_token=AHGY2TOS7UQNOFMME7OBGI3Q4AAD5A5CNFSM4KCQCRYKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEICQBXA#issuecomment-570753244, or unsubscribe https://github.com/notifications/unsubscribe-auth/AHGY2TO36IR24RSQJ4C2P3LQ4AAD5ANCNFSM4KCQCRYA .

-- Ari Widen MBA Candidate | Graduate Assistant DePaul University- Kellstadt Graduate School of Business

trbradle44 commented 4 years ago

I spent a few days going at it and couldn’t get it working. Ended up recreating the entire thing in Matlab only to find that the returns when backtesting over the season so far were pretty subpar. Still playing with it but my impression is that to many professional players have adopted similar strategies in the years since this paper was written

you are correct, sir. many professional players have adopted this long ago and made vast improvements since. in order to compete at this point you need to:

1) gather a TON of historical data on vegas odds, player stats, etc in order to generate your own vegas lines, O/U's, and spreads.

2) use this data you gather to generate your own player projections. don't depend on 3rd touts for player projections.

3) improve the speed of this lineup generator to account for last-minute scratches and goalies changes... often times a few minutes before lineup lock. in the current code - you won't have enough time to adjust your player projections on last-minute changes and regenerate lineups.

good luck

horseranker commented 3 years ago

I spent a few days going at it and couldn’t get it working. Ended up recreating the entire thing in Matlab only to find that the returns when backtesting over the season so far were pretty subpar. Still playing with it but my impression is that to many professional players have adopted similar strategies in the years since this paper was written

you are correct, sir. many professional players have adopted this long ago and made vast improvements since. in order to compete at this point you need to:

  1. gather a TON of historical data on vegas odds, player stats, etc in order to generate your own vegas lines, O/U's, and spreads.
  2. use this data you gather to generate your own player projections. don't depend on 3rd touts for player projections.
  3. improve the speed of this lineup generator to account for last-minute scratches and goalies changes... often times a few minutes before lineup lock. in the current code - you won't have enough time to adjust your player projections on last-minute changes and regenerate lineups.

good luck

Would you be willing to collaborate with me in applying this approach to a new sports domain where the competition is far less and I have several years of rich historical data?