• Hi Guest Just in case you were not aware I wanted to highlight that you can now get a 20% discount on Inform Racing.
    Simply enter the coupon code ukbettingform when subscribing here.
    We have a lot of members who are existing users of Inform Racing so help is always available if needed.
    Best Wishes
    AR
  • Hi Guest Just in case you were not aware I wanted to highlight that you can now get a free 7 day trial of Horseracebase here.
    We have a lot of members who are existing users of Horseracebase so help is always available if needed, as well as dedicated section of the fourm here.
    Best Wishes
    AR

Genetic Ratings

I've long been fascinated by football results forecasting and have spent untold hours experimenting with all kinds of strategies and ratings. The Holy Grail for me however remains elusive - but I do so enjoy the journey.

Recent back testing various incarnations of ELO ratings and similar approaches, it's apparent that the degree of influence that the most recent result is allowed to contribute, whilst critical to future accuracy is also just about impossible (for me) to pin down with any degree of certainty.

This raises a conundrum I've struggled with for some time regarding the construction of traditional rating methods, ELO included, which is as follows;
If, say, our ratings indicate a home win is highly likely. And the game finishes as a home win, that's great isn't it - just what we want.
However, almost inevitable with such calculations, the home sides rating will increase and the away team's rating will decrease.
But why? The method indicated a home win, it WAS a home win, the ratings were good - so why change 'em? The indications were that the home team were superior in this contest, the result bears this out. In such a case doesn't increasing the rating almost suggest that the home team were MORE superior than was indicated?

Which brings me around to my latest tinkering, over the past 12 months or so. Rather than putting too much emphasis on the most recent game, I'd examine all recent games as a whole to get a handle on what each team may be capable of next time out. I cobbled together my own generic algorithm then processed a group of recent games representing each team's most recent ten matches, 'evolving' ratings for each team which best fitted this group of past games. Once completed, apply these genetically evolved team ratings to the next round.

Various combinations of parameters were tested ...
Most recent 6 games, 8, 10, 12, 14 ... were all tried, the best combination of which was ALL games. Using the full set of season's results had the best outcome.
Single rating per team, or different home & away ratings? Different home and away ratings proved the better predictor.
Modelling goal differences or straight WDL. The WDL approach appears more effective here.
Also found to be productive was a slight bias toward more recent results, inasmuch being correct with a result from 1 week ago was ranked higher than one from 1 month ago. A graduated scale ensured more recent results would always be more significant.

Using data from football-data co uk I'm able to process whole seasons on a day-by-day basis (starting from when each team has played a minimum of 10 games) using all previous results from the season to evolve best-fit ratings, which are then used to forecast the result of the following day. Also tracking success/failure and a profit/loss account using both the average and best odds in the Joseph Buchdahl results files.

Results from 2016/17 season, all 5 English leagues;
number of predictions = 1988
43.0% correct predictions
+52.81 return at best odds
-50.05 return at average odds

15/16 season, 40.9% correct, +30.70 best odds (-80.95 average odds)

14/15 season, 42.8%, +66.86 best, -52.62 average

... but can't win 'em all
13/14 season, 40.9%, -81.31 best odds, -184.69 average odds.

Trials of the same process with Euro & overseas leagues have proved unprofitable.

I'm not really sure where to go from here, or whether this is worth pursuing further at all. Although the >40% strike rate and best odds profit does favourably compare with other WDL predictions I've looked at. Checking the figures I can't detect any worthwhile correlation between higher or lower HR-AR differences.

I can share the output files, but not sure what admin/mods will allow/approve of. My postings are still being individual moderated having already been banned on here for allegedly spamming with my first (innocent) post. I can even share the .exe file which will allow others to process league seasons, but sharing .exe files from unknown new bods like me is an obvious minefield!
 
Hello YoVa YoVa,
Right so you are trying to predict football result by using a ranking system is this correct ?. The reason I ask is because that's what I do remember how you rank the teams will be down to you. I have a thread on here showing what I do but I'm wondering if I could help you someway in your quest. When you are predicting are you giving each outcome a Percentage of happening or are you just say this team is ranked 1 at home to this team ranked 12 so should win ?. In each league are you just using the same method ? would you say ranked 1 vs ranked 12 in England would have the same chance as ranked 1 vs 12 in Germany.

As you can see I have many question if you don't mind explaining in a little bit more detail what you do I maybe able to give you that edge.

look forward to your reply
 
I can share the output files, but not sure what admin/mods will allow/approve of.
No problem with you sharing your files.
Its up to the members if they want to download an executeable file, hopefully they will have AV etc but it is at their risk.
 
Nice post YoVa YoVa and some very good points but just using a ranking doesn't always tell the full story .
Goals scored to gain three points is only just another basic stat that can be misleading . How many times have a team been lucky to win ?
Also the table or last 6 or 10 games form may be false .
Who has the best last six game form here MAN UTD beat Sunderland Boro palace Bournemouth Burnley hull 18 points or CHELSEA drew v arsenal spurs Liverpool man city Everton 6 points ?
There's many things to look at before you can just solely use results based stats
 
Let's see if I can give you my spec so it makes sense:

To predict a game each team is allocated a home or away rating based upon the season results to date.
The ratings are constructed by analysing previous fixtures employing a genetic algorithm to evolve a 'best-fit'.

Ratings construction process:
To begin, each team in the league is given a home & away rating of 5 points.
Home wins, away wins and draws are predicted using a home advantage value (Hadv) established by deducting the away sides away rating from the home sides home rating.
Arbitrary values of >0.5, <-1 are used for the model to predict home win and away wins. Hadv between +0.5 to -1 are predicted as drawn games.

All previous fixtures for the season are then cycled through examining the effectiveness of the ratings.
So, because all ratings are pre-seeded as HR=5 AR=5, the first cycle through the ratings the model predicts EVERY game to be a DRAW. (Hadv of 0 ~ 5-5 = 0).
Where the actual result = predicted result, 1 point is added to a counter we'll call "best fit".
The first cycle therefore, the effectiveness of the initial ratings will have a success rate somewhere around 20-30% (how many draws there were in the previous rounds)

The previous rounds are then repeatedly cycled through where each team's home and away ratings are tweaked a little higher or lower to see if the "best fit" can be improved. Where it is improved the 'improved' ratings are saved as the new genetic "parent" set of ratings.
The above is repeated, this time beginning with the new, improved parent ratings until no more improvements are possible.

The ratings that achieved the best won/lost strike rate when applied to the previous rounds are then saved and used to predict today's games.

Home & away ratings are kept within a range of 2.00 to 8.00 and 2 decimal places. An example output file is uploaded (in the main a csv file, but includes my stats updates after each prediction round)
 

Attachments

  • 1516english.csv
    327.3 KB · Views: 7
If anyone would like to check out the compiled executable, here's a zip file containing the constituent parts.
unzip this into its own folder.

File in the folder are;
evorater.exe
settings.txt

plus the footballdata csv files you wish to process (see below)

The settings.txt file should be changed to suit as follows;
1st line - the name of the file that will output your results.
(if you don't change the file name, the previous version will be over-written and lost)
2nd and subsequent lines, the football-data.co.uk files you want to analyse.
(so you need to make sure the files you're processing are placed in this folder!)

Run by double clicking "evorater.exe"

A maximum of 10 footballdata results files can be analysed per run.

Be aware, this is not a swish Windows GUI file, simple a console with no interaction, but where you can see the progress of the run.
Also, there is very little in the way of error checking, if the program crashes it will be due to files in settings.txt not where they should be or misspelt, etc.
 

Attachments

  • evorater.zip
    191.8 KB · Views: 10
Good luck with this you seem to be using a very elaborate way to predict something that will invariably be bad value
Results based rating are in my opinion flawed as games from 7 or 8 months ago have little baring on what is currently happening ( transfer ,injury,New manager, New tactics ,confidence etc etc etc )
current form is more relevant in football than most sports in fact there's a few people on the forum who solely bet on in play and make a good profit and you can't get more current form than that
One thing I would say is if you have a system that you are happy with and you have the temperament to stake/bet well them I'm sure you will make a steady profit
 
"Focus on the journey, not the destination. Joy is found not in finishing an activity but in doing it."
Greg Anderson
 
"Focus on the journey, not the destination. Joy is found not in finishing an activity but in doing it."
Greg Anderson
Very very true that quote. I love the Chase so to speak. I hope I didn't come across as to negative in previous post as i have the up most respect for anybody who takes the time and effort to give themselves a chance to beat the odds .
Let's hope our journey isn't to bumpy but if it is I'm sure we can find ways to make the trip smoother to end up where we want to be .
 
Results based rating are in my opinion flawed as games from 7 or 8 months ago have little baring on what is currently happening ( transfer ,injury,New manager, New tactics ,confidence etc etc etc )

I thoroughly agree that results based ratings are flawed, because of the reasons you state. But many who compile them use the caveat that they do have shortcomings. IMO they can be valuable but when used with other criteria, either as a starting point or as confirmation of other angles.

That said, it must also be true that such ratings will vary as to their effectiveness, from excellent through poor. My goal here is an attempt to create a more effective rating than the mainstream, and ratings, as here, that can show a profit from backing ALL fixtures over the past three years must be showing some potential?

With regard your assertion that older form plays little part in effective predictions, that is an opinion to which you are of course entitled. It is certainly not backed up by the many forms of research I have conducted over the past 30 years or so. Indeed, the results I've posted from this research in this thread suffer if the past results used to compile the rating figures are restricted to the most recent games.
 
Back
Top