Tuesday, 10 August 2010

Systematic Betting

So I was watching the European Athletics Championships last weekend, and then this week it weirdly got me onto the thought of horseracing. I realised that for people it's pretty fucking easy to see who is going to win. At the very least if you were wrong about the winner you'd be close enough to win an each-way bet. And this is just going on personal bests and seasons bests. I have no history of following athletics seriously, yet I could fairly accurately guess who was going to win just from the stats they showed on the screen before each event.

So my figuring is that if you are to look at that sort of thing for horses, properly collate the data and take a proper look at it, then maybe you could fairly accurately get race results. I understand that the times of horses finishing races are probably more variable than those of people, especially given that a horse is probably a bit less driven to give it's best for every race. Also shit is pretty complicated given horses run better with different conditions, different distances. But I'm confident that if someone can be bothered to wade through the data and find the winners then there's money to be made.

The thing is, I don't really like horseracing, and I can't be bothered collating any data because I don't care.

However, I do care about football. The main difference here is that football is far less easy to just apply some sort of scientific methodology to. Finishing times are one thing for racing, but how the fuck do you assess football games?

The real answer is that you probably can't. I believe that it's far easier to make money casually betting on football than it is casually betting on horses, but in terms of a system I doubt it's that easy. I doubt it's even possible at all. But that isn't going to stop me from trying.

I'm not going to put proper money into it though, or not for at least a season. Basically I'm going to set up stuff on Excel spreadsheets like crazy, amass data on pre-match odds (not sure if I'm going to bother finding the best odds each time or just go with a single bookies), and I'll try a few simple systems just to see how much money they'd have lost if I'd done it for real.

Because I am expecting them to lose money. Fact is that bookmakers make profit, and they wouldn't if simple systems resulted in a loss for them. I'm not doing this to try and find a get-rich scheme, because I'm fairly sure there isn't one. I'm doing this mostly as a curiosity. Plus Excel 2010 seems sexy and I need an excuse to use it a bit.

So I'm going to start out with a few default betting tactics, and I'll see how they do.

1) Bet £5 on Manchester United to win for each game
2) Bet £5 on Manchester United to win for each game not against another top-six team (Chelsea, Man Utd, Arsenal, Liverpool, Spurs, Man City - I reckon Villa will be shit without O'Neill)
3,4) As above but for Chelsea
5,6) As above but for Arsenal
7,8) As above but for Liverpool
9,10) As above but for Tottenham Hotspur
11,12) As above but for Man City
13) Bet £5 for Blackpool to lose every game
14) As above but for Newcastle
15) As above but for West Bromwich Albion

So, that's fifteen fairly basic strategies that I won't expect to actually make a profit individually, let alone collectively, but I'm still curious to exactly how much (virtual) money will actually go to the bookmakers for each one.

I'll be keeping a fairly comprehensive set of stats for the entire set of fixtures, so will probably at least take a further look and analysis of the data at the end of the season, and probably at least upload the spreadsheet for people to look at. There's so much extra shit I could put in and won't, like odds one week before the fixture as well as just the day before, but frankly I can't be bothered with this sort of thing. I'm probably being lazy to the point of actually just ensuring I waste my time by not going into enough detail to make things properly exciting, but meh.

If anyone else has Excel 2010 and wants to help then I could probably work out shared spreadsheet access and might go for a bit more detail if I'm not doing all the work myself.


  1. "So my figuring is that if you are to look at that sort of thing for horses, properly collate the data and take a proper look at it, then maybe you could fairly accurately get race results." - I think this is how a bookmakers works Chris... be careful you don't turn into a Mathmo before we get back to uni :) Also quick google search turns up http://www.football-data.co.uk/englandm.php if you wanted to stick last year's results through. Hope you're having a good holiday. James P

  2. Yeah, I appreciate it's how a bookmakers works, or at least for the initial odds, but I'm curious to whether their final pre-match or pre-race odds necessarily reflect this, given they can be swayed by how people bet.

    I was originally going to say that since I'll be doing experimental stuff rather than random abstract statistical theory, it's more likely to make me a physicist than a mathmo. But then I realised that seeing as I'll be using proper data rather than attempting to create a model for a football league, I'm still probably keeping in line for being an engineer ;)

    That football-data stuff could be interesting but I don't quite understand the various columns in the CSV files. I get what some of them are, but certainly not all, though I didn't inspect it that closely.

    I reckon I'm more likely to actually get some sort of final results collecting the data and manipulating it myself as the season progresses (ie. in bitesize chunks) than I would be to try and get this done with one of those massive CSV files, even if it'll obviously take me longer.