Open octaviodegodoy opened 4 years ago
did you find a solution? I got the same error
Hi, anyone found a solution? I get the same error :(
it's a problem related to the pickle stuff, just try to comment out from the main those 4 code line of "dump" and those 4 of "load" that should work guys
@giambe99 thank you for the help, I will give it a go. By the way, have you tried it? cheers
Actually there are some problems with this code... First of all it's clear that the elo feature is completly useless. Also i suggest to delete about the first 50 lines and directly pd.read_csv() using the file atp_data that you can find on kaggle.
You maight need to change datetime with datetime.datetime somwhere... Anyway the code is usless if your scope is to earn money, I do strogly invite te author of this project to give it another shoot and improve it, cause actually its a very interesting topic
@ricciolini sei italiano anche tu? Are you skilled with xgboost?
Admit we are here cause of kalle hallen video hahahaha
I'm a Python newbie ahah. I just wanted to try this code out for learning. There's a lot I don't understand, including xgboost. And I'm Brazilian, with a bit of Italian blood :) Thank's a lot for the tips, @giambe99 :)
@ricciolini is an italian surname. My advice 4 you is to first learn at least the basic logics of python, using pycharm i would say. Then if you want to learn about ML xgboost is a powerfull tool wich you could run using anaconda. Or if you just need to invest buy 10k of BTC and yo won't regreet that. Bye boss
@giambe99 thanks for the advice :) I'm with you on BTC. Got some already and looking into getting some more, finances permitting ;) Take care and stay safe :)
it's a problem related to the pickle stuff, just try to comment out from the main those 4 code line of "dump" and those 4 of "load" that should work guys
It didnt work for me. I have the same error. What can I do?
it's a problem related to the pickle stuff, just try to comment out from the main those 4 code line of "dump" and those 4 of "load" that should work guys
It didnt work for me. I have the same error. What can I do?
Was facing the same issue. For me, commenting the following lines (79 to 86 of main.py) worked :
Let me know if you need me to check more Best regards
@giambe99 why you think the elo feature is completely useless?
@giambe99 why you think the elo feature is completely useless?
it's good to predict matches outcomes (with a certain precision), but it's not good enought to predict outcomes better than the bookies...
for ex, elo 2000 vs elo 1600 ---> alomost certain, but u're not considering facts like recent injuries that could even leed to se a bookie offering a 1.80 1.90
that said the best way to exploit bookies in a market is just to find the average of a given outcome and then bet on the platform who's offering the best deal (the best, not the profitable way). The main problem is that the average rake is about 6-8%, this makes so hard to find a profitable strategies (almost impossible, tendent to 0). if in coming years the market average rake will drop to something like 3% that would leed to more opportunities, but just if the current fluctuations between the bookies are kept ( a pretty unlikly scenario...)
@giambe99 why you think the elo feature is completely useless?
it's good to predict matches outcomes (with a certain precision), but it's not good enought to predict outcomes better than the bookies...
for ex, elo 2000 vs elo 1600 ---> alomost certain, but u're not considering facts like recent injuries that could even leed to se a bookie offering a 1.80 1.90
that said the best way to exploit bookies in a market is just to find the average of a given outcome and then bet on the platform who's offering the best deal (the best, not the profitable way). The main problem is that the average rake is about 6-8%, this makes so hard to find a profitable strategies (almost impossible, tendent to 0). if in coming years the market average rake will drop to something like 3% that would leed to more opportunities, but just if the current fluctuations between the bookies are kept ( a pretty unlikly scenario...)
Yeah, good point. I was thinking to look for an average of certain points without using the elo, because as you said sometimes it is useless. If I have some good predictions from the average points, I would give it a try on a platform.
almost finishing compilation I got this error 24200/24481 matches treated. 24300/24481 matches treated. 24400/24481 matches treated. Traceback (most recent call last): File "main.py", line 83, in
features_player=load("player_features")