Greyhound Model FREE!!!

[Fugazi]

So I've decided to upload my greyhound model to the forum. Reason being I've had success in other avenues and its been neglected for the last couple of months. It's got a couple of annoyances to be fixed.

It uses a huge database of stats for every greyhound that has raced in the last 10+ years, up to around 2 months ago. It was from a website that is difficult to collect data from. This model was made by a genuinely very prestigious data scientist. It has a couple of annoyances that need fixing that I'm not sure how to fix, it was working profitably for me but can be much improved with some tweaks. Why would I give away something profitable for free ? Quite honestly, the time running it each day and then following the markets was not a good time investment vs ROI for me. It needs some tweaks to be able to scale it more. Also the harvester kept needing updating which was becoming annoying and again time consuming.

If you're not familiar with python I'd recommend the python bootcamp on Udemy. I'm not brilliant at python myself and won't be able to spend time teaching how to use this and make money (and I would probably fail anyway!). It' s not a nice piece of software with a "make money" or "start" button. All that said, this could be a great starting point for those trying to learn to create models.

Problems that need fixing:

1) ISP. For its input, one of the statistics is ISP. However Excel converts ISP to a date so I've been removing ISP from it input at serious detriment to the output

2) Time trials. Time trial data really confuses the model, probably due to the grade, and finishing position.

Its output will give you absolute probabilities (i.e in a vacuum, not considering a dog has other dogs in the race) rather than relative probabilities of a dog winning. I can't find my script for this now. You need to use chat gpt and ask for script to pick out the dogs in the same race and create relative probabilities.

Originally I wanted to include trainers and owners into the input but it created difficulties with granularity. But this is something I would be keen to revisit as it is the missing step between this being a useful trading tool and being a highly accurate odds predictor. While the columns owner and trainer are in the input excel sheet, they do not get used by the model.

If someone does fix its annoyances and gets it working as it should I may be encouraged to head back to the developer and have a second version created.

Enjoy!



Nb. File (825mb) was too large for direct forum share.

Here's the link:

https://drive.google.com/file/d/1HXVIQM ... sp=sharing

[Mike Oxlong]

Hey Fugazi, thanks for sharing this. Looking forward to getting stuck into it when I get some spare time!

[ilovepizza82]

Doesnt seem to work:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: Can't get attribute '_PassthroughScorer' on <module 'sklearn.metrics._scorer' from 'C:\\Users\\WORKSTATION\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\sklearn\\metrics\\_scorer.py'>

[Fugazi]

Doesnt seem to work:

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: Can't get attribute '_PassthroughScorer' on <module 'sklearn.metrics._scorer' from 'C:\\Users\\WORKSTATION\\AppData\\Local\\Programs\\Python\\Python311\\Lib\\site-packages\\sklearn\\metrics\\_scorer.py'>

Windows + Pycharm is what i run it on.

Harvester I run on Spyder

Just having a look at it now to check if theres anything important I've forgotten about for running it. Most likely its going to need adapting to individual set ups - was a pain getting it running on mine initially.

Edit:

Still running fine on mine. Make sure you change all the C:\users locations to your own directories. Think there are some in run_model , clean_data and get_data that need changing

Make sure all the requirements are installed.

Other than that I can't really help its why I've decided to give it away as haven't had the time to nurture it and get the max capability out of it. But in the right hands this thing is dynamite.

[Fugazi]

Lots of downloads,

Nobody having any luck with it?

[tico]

Hi Fugazi,

I have downloaded the file ,but not too sure how to put it all together ,I have got Pycharm and am reasonably competent with Python and VBA .Any suggestions would be great .I have got to the unzipping stage .Any hints would be much appreciated.
Kind Regards
Tico

[Fugazi]

Hi Fugazi,

I have downloaded the file ,but not too sure how to put it all together ,I have got Pycharm and am reasonably competent with Python and VBA .Any suggestions would be great .I have got to the unzipping stage .Any hints would be much appreciated.
Kind Regards
Tico

Forum user GaryCook has got this working well, he'd be the best bet

[tico]

Ho Fugazi,
Many thanks for that
Kind Regards
tico