Wednesday, 9 November 2011

Fun, fun, fun.....

As mentioned on my last post, there are going to be some ‘fun’ bets for this weekend.

For those that have been following the blog for a while, you’ll know that I tried to use my ratings before any league games had taken place this season but didn’t record them in my official results. I simply used my ratings from last season and adjusted them for the performance of newly promoted teams in historical seasons in the first few games and this gave me a factor that I can use to adjust ratings between leagues. I won’t pretend that this isn’t very crude and I wouldn’t trust it too much if I’m honest. However, it allowed me to give some bets out before any league games had taken place this season which was an interesting experiment. I didn’t place any money on these games myself.

As it turned out, the results were very good although of course, this didn’t only involve games including newly promoted teams. I didn’t do any sort of analysis looking at how my ratings did between newly promoted or relegated teams and teams who were in the leagues last season but I didn’t see anything untoward when I looked back at the games. The ratings held their own in games involving adjusted ratings from a different league last season.

Anyway, for those that weren’t reading the blog then, he’s a summary of the results.

Well, this leads me onto this weekend. As you’ll be aware, I ignore all cup games for my ratings and don’t use my ratings on any cup games either. After the results of the experiment earlier this season, it did make me think how I could use my ratings for cup games. However, with the work I do for the league games, I don’t really have any time to look at the league cup or Johnstone’s Paint trophy as it clashes with the work I do for the league games.

This week though, with there being no league games at all due to the International break and the FA Cup, I’ve decided to see what I can do for the first round of the FA Cup.

Clearly, I only have ratings for teams in League One, League Two and the Bsq Prem. I’ll therefore only be able to look at games involving these teams. There are 25 games of this type.

For this weekend only then, I’ll give out full ratings on every game. I’m not giving away my ratings for each team (that’s a secret as they’ll be valid next week as I don’t adjust for cup games!) but I’ll give a simple H/A/X % probability for each game and highlight the value bets. This is in essence the output from my ratings I use to determine the value bets that make their way onto the systems.

One bi-product that is thrown up from my ratings is I can work out the probability of each correct score occurring. I never use it and never intend to as it’s useless I suspect! However, I have always liked the idea of looking at over/under goals as that is another bi-product of my ratings. I’ll also highlight whether over/under 2.5 goals is value for each game this weekend. Of course, it’s a flaw in my model that every game is either value for over or under. It’s one issue I suspect with using a Poisson Distribution as a basis for correct scores before bringing in the rating algorithm to interpret the correct scores.

Anyway, I thought that would be fun to do this week. I’ll post up the bets tonight or tomorrow.

I’ve already said to people to not follow this with real money as it really is a shot in the dark. If I see some evidence that it can be developed further, I may look into it next summer but there are so many complications involved, I don’t think it would be worth it if I’m honest. For example, if I did do cup games, I’d need to adjust my ratings after the cup games. This then invalidates all my historical ratings as I’m not going back in time to restate 11 years worth of data using cup games!

I do like the idea of a standalone FA cup methodology, based on my league ratings but ignoring all cup games. Keeps it in line with what I’m doing and has no impact on my other ratings work.

Before I go, the one thing I’ve not touched on above is the idea that FA Cup games aren’t likely to behave like league games! Hence, my ratings are probably 100% invalid due to this reason. A small team at home may be happy with a draw and to get a replay and likewise, a larger team may well settle for a draw on 80 mins and take the replay if offered. Does home advantage count more in Cup games? There are too many unknowns for me to state if I’m honest.

Unfortunately, 25 games isn’t like to answer all my questions but I think it will give an indication if my ratings are flawed or not for cup games. If I have 25 value bets, I’ll know they are flawed! Likewise, if I have my prices nowhere near the bookmaker prices, I’ll know that my ratings are flawed.

I’m doing the work tonight (did all the preparation already), so I’ll get the bets up here when I can.

No comments:

Post a Comment