Neural Network Update

Haven’t posted in a while so I thought I’d bring anyone who is interested up to date with the Neural Net stuff I’m working on.

The last post I did on this subject was to piss and moan about the loss of my data source for the Yank racing which pretty much put that project to bed. Undeterred I decided to wait it out and begin anew once the UK flat racing got underway.  I’m happy to say that I’ve been busily gathering data and recording it in a spreadsheet over the past few months and I’m hoping to begin training the network this coming weekend…I just have to get hold of the software that will do all the number crunching for me in the meantime.

Once the network is trained on the data that I already have I’ll start to present it with test races to see how much it has ‘learned’ from the training data.   Basically I will feed the network the same info as it is used to receiving minus the actual race winner (up to now I have provided that info in the spreadsheet data that I have recorded).  The idea is that the network uses what it already knows to make it’s best guess at the probable race winner.  If it gets it right I let it know…if it gets it wrong I let it know and it makes adjustments.  This testing phase continues until I am happy that the network can predict reasonably well the winners of races that are yet to be run.  That’s the theory anyway…

Once the testing phase is complete I’m going to put up some proofing picks though I’m not sure where…probably on here and maybe on one other site.  The whole point of creating this thing was to try and get something that I could market (plus use it for myself obviously).  The beauty of this product is that it gets better with time and no matter how many people decide to buy it (if I think it’s marketable) it doesn’t affect the market prices in any way like a plunge on an Isiris tip might…everyone uses it in their own way and chooses their own races for forecasting, oblivious to what anyone else might be doing with their copy so really no two copies out there should be alike.
There’s still tons to do yet though and it may not see the light of day this year if I don’t think it’s fit for purpose but fingers crossed it’ll pass the test.  That’s about it – just wanted to bang this out and let everyone know I’m still alive.  As always if you have any comments then feel free to pass them on.  See ya next time 🙂


2 Responses to Neural Network Update

  1. CJ says:

    Hey, i found this very interesting. What software (for number crunching) were you / have you been using for this? I didnt know that there was readily available software for neural networking.

    Hopefully I have not commented too late and you will get this message.


    • swearbox says:

      Hello CJ and thanks for dropping in.

      Glad you found the post interesting – it’s not the kind of thing that floats everyone’s boats, so to speak, as it probably all seems a bit geeky and nerdy but it’s not. If you can figure a way out of representing a horse race as a bunch of numbers you can quite easily build a network to help you forecast future winners (but as you already said you found the post interesting I’m assuming you already know that).

      I’m currently working on a new and improved version of the network I built a few years back and when it comes to the number crunching bit I’ll be heading off to the place I used the last time to get my neural network software. You can find it at

      Steve Wolstenholme is the guy that runs the site and markets the software and he’s very approachable. If you end up buying from him let him know who sent you there. Good luck.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: