Who Are Ya

Check out this guest post I did on Giles Metcalfe's No Standing Blog.

Gives a little more information about me and what I do

Poisson Betting - September Update

Just a quick update to how the Poisson betting is going (Slightly more delayed than I would have liked – these results are to the end of September)

There has now been a total of 1,569 games to bet on and a total of 23,708 different bets.

The total which my sheet has predicted correctly has held pretty steady, increasing by 0.92% to 65.38%

The top 3 Leagues for results being predicted correctly are Scottish Premier League, French Ligue 1 and Serie A– all with over 70% of bets correct.

The lowest 4 are the English Premier League, League One and League Two along with Serie B – all are around 61%

The 2 Leagues that have shown the highest increase are Spanish La Liga and the English Conference – both increased by around 5%

The largest drop was the Portuguese Liga which dropped by 6.5% - no other league decreased by more than 2%

Table below showing some of these results.

Couple of things I’m working on to improve the usability of my sheets in the short term:-

• Coding some of the macros so they run faster
• Instead of having to choose the individual teams each time I’m working on a fixture list which would run all the macros with 1 button click
• Working on tightening up the ratios when teams have been promoted and relegated – I’m preparing another blog post on this
• Looking at areas where teams are predicted to win with better odds than the bookies are offering

I received a couple of comments last time I posted an update – if you have any advice or feedback on how I can improve these, whether you are doing something similar or just noticed something I may have missed please feel free to comment and I’ll respond when I get chance!

Poisson Betting - August Update

Here is an update on how my Poisson Distribution is going so far this season. This covers the whole of July & August

If you missed it, I talk in an earlier blog post about my motive behind this but below are the 19 separate bets across the 23 different leagues that I have analysed

For each game I predicted:

• Which team would win or if it would be a draw
• Double Chance (Win/Draw for each team)
• Score Prediction
• Both Teams To Score (or Not)
• Over and Under 0.5, 1.5, 2.5, 3.5 & 4.5 Goals per game (10 separate Bets)
• Corners – Under 10, 10-12, 12+ (English Prem – L2 only)
• Cards Predicted – Under 4, 4-6, + (English Prem – L2 only)
• Red Card – Yes/No (English Prem – L2 only)
• Home & Away Cards Predicted (Under 2, 2-3, Over 3) (English Prem – L2 Only) – 2 Separate Bets

 The Table below shows the success rates across each League

I don’t think the results are too bad but time will tell whether these could have been achieved purely by guessing rather than statistical prediction. It’s early days for some of the leagues so the results will fluctuate over time. The breakdown of the ratios for each division is below.

I will update the scores again at the end of September and so forth on a monthly basis.

 

Football League - The Gap in Points

Mind the Gap

While the final day in the Premier League didn’t carry the same drama as previous seasons with most issues wrapped up long before the end of May, the same cannot be levelled at the Football League. Whether it was the drama at Vicarage Road/KC Stadium that saw Watford fall at the final hurdle, the amazing scenes as Brentford missed a penalty and Doncaster raced away to score a last minute winner or Barnet’s heartbreak at being relegated after a string of results went against them.

The main point I wanted to look at was the unbelievably high number of points needed to survive in the Championship, indeed with only a handful of games of the season left some teams were within touching distance from both the play offs and the relegation zone.

Peterborough ended up being relegated with a record number of points from the 2^ndtier since the switch to 3 points for a win and this was only confirmed on the final minute of the final day of the season. Given this I decided to take a look at how the number of points in various positions compared with previous seasons.

The graphs below would look much better in Tableu – unfortunately I have a Mac so they are done in Excel!

The following graph shows the points achieved for each position of significance along with the average needed for the last 15 seasons.

Difference in Points for the Championship (click to enlarge)

Points to note

• The gap between the final Play-off position and the final relegation place was the smallest it has been in the data I have gathered – just 14 points between 6^th and 22^nd
• Bristol City finished well adrift of safety, actually joint 5^th highest in the data gathered. However they got the 2^nd highest points for a bottom placed team – and would have actually stayed up the previous season.
• Cardiff City finished top with the 2^nd lowest amount of points needed to win the Championship. However, they were close to the average number of points between top and 6^th (the last play-off spot). This was because the last play-off spot was the lowest in the recorded data – 6 points below the average needed.
• The information means the division was very even last season, the teams at the top didn’t run away with it and the teams at the bottom were not adrift. The topsy turvy nature is perhaps best highlighted by Peterborough United (who finished 22^nd) doing the double over Cardiff City (who finished top)
• The average number of points needed for the last play-off spot is very consistent over the last 15 years. 12 of the 15 teams finishing 6^th had a total of between 73 & 76 points
• 2007/08 and 2012/13 were both very close in terms of the number of points between the last relegated team and the last team making the play offs, with 2007/08 being the year with the least number of points between top and bottom (43)

Difference in Points for League One (Click to Enlarge)

• Doncaster Rovers, who finished top in 2012/13, achieved less points than the average number needed to get automatic promotion. Only once has there been a tighter division for the gap between top and bottom than the 52 points in 2012/13 (an unbelievable 35 points in 2005/06)
• In 2010/11 and 2011/12 the gap between bottom and safety was extremely tight just 5 and 6 points respectively separated the teams.
• The last relegation spot is very consistent, for 10 of the last 11 seasons it has been within 3 points of the average of 48 points.

Difference in Points for League Two (Click to Enlarge)

•  The average for the Top spot is lower than the other 2 divisions (89 in League Two, 94 in Championship & League One)
• Again all 3 promotion spots (top, automatic, play offs) was lower than the average needed meaning it was much more equal throughout the division
• Barnet can consider themselves very unlucky to have been relegated with the highest number of points from the data (2^nd relegation place was only introduced in 2002/03)
• Aldershot achieved 48 points which is a higher total than the average normally to survive (44 points) and was the joint highest amount for a team that finished bottom.
• The gap between the play offs and relegation was much smaller than normal (18 points compared to the average of 28) with only 1 season having a tighter margin (2005/06)

Conclusions

• 2012/13 was much tighter across all 3 divisions than normal, the promoted teams scored less points and the relegated teams more points than usual.
• 2012/13 had the highest number of points needed to survive in the Championship and League Two for any of the 15 years I have data for.
• 2012/13 had the lowest number of points needed to gain automatic promotion from Championship and League Two and the third lowest from League One
• The data indicates that 2012/13 was a one off and not a continuation of a sloping trend towards tighter divisions. There is no expectation that 2013/14 will be just as tight
• There is some correlation between divisions being tight at the same time. For example 2012/13 was in the top 3 of the 15 seasons in terms of the number of points between top and bottom in all 3 divisions. 2010/11 was in the top 5.
• 6 of the last 7 seasons in the Championship had been the closest between top and bottom in the data set. At least at this level the gap has narrowed considerably.

Promotion from the Conference

Gaining promotion from the Football Conference (formally Blue Square Premier League) is a hard business. No less than 12 former league clubs will be competing in the division next season and on top of this some ambitious clubs with great resources from the Blue Square North & South will be joining the other already established teams in there.

It has always been a very competitive division to get out of, Luton were expected to stroll out of the league and despite having average attendances far and above some of their rivals they have never been able to get over that final hurdle.

But what about the teams that do make it out of the division? As we have seen in the leagues above many teams have a tendency to yo-yo between divisions but this is hard to do when the Conference is so competitive, so it’s imperative that once you’ve been promoted you make a good stab at league football.

I used 20 years’ worth of data from Wikipedia to see how teams did once they’d been promoted from the Conference. The data goes back to 1989/90 but due to no teams being promoted between 1992/93 and 1996/97 and the introduction of 2 teams being promoted in 2002/03 I have data on 30 teams.

Flying High

Crawley celebrate winning a 2nd successive promotion

The first thing I looked at was how teams fared soon after being promoted

Very interesting that 14 of the 30 teams had been promoted to League One within 3 years of promotion from the Conference, with 8 of these going straight up through 2 divisions (Crawley Town the best recent example). Fleetwood are strong favourites to mount a promotion challenge this season just 1 season after promotion from the Conference and Mansfield are also expected to at least trouble the play-off spots.

So there is relatively little difference in standard between League Two and the top half of the Conference. Although the sample size is small the results are quite stark.

Unfortunately it does seem like League One is a step too far for most teams with 8 of the 14 teams coming back down within 3 years of promotion. This gap is reducing with Stevenage & Crawley both having a seasons and Yeovil Town – by now out of the range of our 5 years data mining by being promoted from the Conference in 2002/03 – surpassing themselves by being promoted to the Championship in last season’s play-off final.

Tough at the bottom

Macclesfield suffered the heartbreak of relegation in 2012

Unfortunately at the wrong end of League Two times are incredibly hard. The recent example of Aldershot, just 5 years after promotion to into the League and having already gone out of existence once in the late 80’s and now in administration again shows how tough any level of football is. If you aren’t pulling in big crowds you have to cut your cloth accordingly. The number of teams relegated from League Two who entered administration or went out of business is too high.

12 of the 30 teams promoted in the past 20 years have had some kind of major financial difficulty. Many of these have occurred when they have been relegated back to fifth tier level.

6 of these teams either went into administration or out of business completely and had to be reformed. Thankfully a lot of these historic teams are now on their way back up with the reformed teams of Chester City (originally promoted in 2003/04 and relegated in their 5^th season) and Halifax Town (originally promoted in 1997/98 and relegated in their 4^th season) both now promoted back into the Conference and Darlington winning the league in their first season.

With the level of professionalism now brought to the Conference, should it effectively be renamed Division Five (League Three??) The league is national, many teams are well back and professional standard, giant killings in the FA cup are less prominent and not as shocking as they used to be. The standard has definitely evened out in recent years.

What do the Stats show?

The final chart below shows the average positions for all teams competing in League Two in the years after their promotion for the conference (the average position of teams who were promoted into higher divisions has been removed).

The chart shows that the average position is around 11^th -12^th in the first season, quite a respectable finish and this improves in the 2^nd season and only drops marginally in the 3^rd. It is in the 4^th & 5^th seasons after promotion that the drop comes (a pretty steady average of 16^th in both seasons) and this matches up with the findings that no team has been promoted in the 4^th & 5^th season after promotion.

So if you are a Mansfield Town or Newport County fan you have plenty to be positive about, if you’re an AFC Wimbledon fan you’re in the last chance saloon for promotion and if you are a fan of Oxford United, Burton Albion or Torquay United you might just have missed the boat already.

Using Probability Theory & Poisson Distribution to win money!

I’ve been gambling casually on football for the past 8 years or so, and not making a great job of it! I’ve had a few decent returns but I’m almost certainly quite a bit down over the total time I’ve been betting.

Most gamblers will be split into one of a few camps. I do it to make a Saturday afternoon watching Soccer Saturday that bit more interesting (who was it that said it mattered more if there’s money on it?!), there are those that hope to win big and there are those that claim to be able to provide all the answers. I’ve seen a lot of these on twitter who claim to have xx% win rates – doesn’t really help having 90% win rate when you’ve picked a 10 team accumulator does it?!

Football is always a game of randomness and it’s so hard to predict with any great accuracy. My current method of gambling is partly between betting with my knowledge, partly through looking at the odds and seeing the teams with lower odds you’d think are good to include in an accumulator (not a good way of doing it as bookies have full control over the odds) and partly through statistics – things like form/league position/goals scored etc.

Why I’ve never decided to look more in depth at the statistical side I don’t know. Given that it is the area I am involved in I should have looked sooner but never really crossed the two paths of performance analysis and gambling until the past year or so.

After a bit of information gathering on the internet I settled on using Poisson distribution to look at previous scores and primarily the home and away goals scored by each team. (Chris Anderson & David Sally touch on this in their new book The Numbers Game)

So using the data from Football Data I have built a statistical model based on working a few things out and giving me the probabilities on a few things across the top leagues in Europe. The models cover 22 different divisions.

• England (Premier League, Championship, League One, League Two & Conference)
• Scotland (Premier League, Division 1, Division 2 & Division 3)
• France (Ligue 1 & Ligue 2)
• Germany (Bundesliga & Bundesliga 2)
• Spain (Primera Liga & Segunda Division)
• Italy (Serie A & Serie B)
• Holland (Eredivisie)
• Belgium (Jupiler Liga)
• Portugal (Primiera Liga)
• Turkey (Super Liga)
• Greece (Super League)

From all of these divisions the model takes into account the home goals scored and conceded and away goals scored and conceded (depending on where each team is playing) and using Poisson distribution and probability theory I can find probabilities of each of the following.

• Home Win
• Draw
• Away Win

• Home Win or Draw (Double chance results)
• Away Win or Draw (Double chance results)

• Predicted Score

• Both Teams to Score
• Both Teams NOT to Score

• Expected Goals Under/Over
• 0.5
• 1.5
• 2.5
• 3.5
• 4.5

Am I expecting to become a millionaire? No. But I am hoping that the model will greatly help me in my casual betting and so far I’m quite happy with it (I only began using it at the very end of last season and it needs a lot more testing – the hardest part has been waiting until the leagues start back up again!)

It is my aim to use this blog to provide a few updates of how it’s going and look to integrate other things into it, I’m mostly interested in how accurate it is at predicting the H/D/A and scores – the things with the highest odds and least likely to be predictable. For example if we pick a random game from the 1^st week of the Premier League Season – West Brom vs Southampton

• West Brom to Win – 50.28%
• Draw – 23.98%
• Southampton to Win – 25.74%

• Expected Score - West Brom 2 Southampton 1

Not clear cut by any means and just 1 very small example of what the sheet provides.

Hopefully I’ll provide updates on a regular enough basis to be interesting but not turn this blog into how I lost all my money gambling!

Do Goalkeepers Raise Their Game Against Big Clubs

Goalkeepers are traditionally very hard to analyse. There are understandably less statistics produced due to goalkeepers generally having less actions during a game, although the modern day goalkeeper needs to have added excellent distribution to his repertoire to effectively play as a sweeper. The limitations of the available statistics is probably for another post but the obvious one is whether a shot should be expected to be saved. Paul on his Different Game blog has already explored some excellent work on goalkeepers. 

Joe Hart was in unbelievable form during Manchester City's title winning season

From a throwaway line I heard on Match of the Day some time ago, I decided to investigate the impact playing against better teams had on keepers. In essence are performances like John Ruddy away at Liverpool for Norwich in securing a point reflective of their performance over the season or are they “one offs”.

Using the MCFC analytics data, which is now unfortunately a season out of date (here’s hoping they continue to release updates on a better than annual basis although the project does seem to have died a death) we have access to a full seasons worth of data using statistics provided by Opta. By working out how many shots each team face on average we can see whether the % they save is higher or lower against the ‘big 6’ (in this case I have used Man Utd, Man City, Chelsea, Arsenal, Spurs & Liverpool – sorry Everton fans!!)

In all 44 Goalkeepers played during the 2011/12 season for the 20 Premier League Teams and the first thing I looked at was how shots & goals affected the final league position.

It’s clear from the graph below there is a reasonable correlation between the number of shots faced and where teams finished (R Square is 0.6484). It’s common sense that the more shots teams have at you EVENTUALLY one is likely to go in, so conceding less shots initially should result in conceding less goals. Using Shots Conceded is effectively half of the Total Shots Ratio Model in reverse as with that model the percentage which you outshoot your opponent has a high relationship with finishing position – it means that if you concede less shots you are much more likely to be able to outshoot your opponent and ultimately finish higher.

Graph showing total shots faced vs final league position for 2011/12

The second graphic shows the number of goals actually conceded and is an even better fit, with an R Square of 0.7027. There are some outliers here (Sunderland conceded as many goals as Arsenal but there were 10 places between them – it’s OK not conceding many but you do actually have to shoot and score sometimes!)

Graph showing total goals conceded vs final league position for 2011/12

So knowing this I go back to my original point and whether some goalkeepers are able to raise their game against the big boys.

I chose a cut off of 1350 minutes (15 games) to look at, which gave me 22 goalkeepers. The gap between Thomas Sorenson (1350 minutes played) and the next 3 keepers in total minutes (3 at 720 minutes) was reasonably large anyway and a good place to cut.

First to look at is the total number of shots on target (excluding penalties) each goalkeeper faced. We can see from the simple table below which keepers immediately stand out.

Minutes per shot on Target

Minutes per Goal Conceded

Minutes per Goal conceded vs Minutes per Shot Faced

Immediately Joe Hart stands out as much better than the rest (remember, this is 2011/12 data so his dip in form last season is not taken into account) whereas at the other end Adam Bogdan and Jussi Jääskeläinen are both in the bottom 4 keepers in terms of the lowest number of minutes per goal conceded. This goes some way to explaining why Bolton were relegated.

So where it was taking just 2.5 shots on target to beat Paul Robinson it was taking twice this many to score past David de Gea. Not bad for a keeper who supposedly had a poor first season!

David de Gea is now proving to be the top class Goalkeeper he was signed to be

I’ve now compared the figures against the previously mentioned big 6 against the other teams. For this I have removed Sorenson, Jaaskelainen, Begovic and Bogdan due to them not playing enough games against the better teams. I set a minimum of playing at least 2/3rds of each of the games against the big 6 and the other 13 to give a reasonable representation.

Minutes per shot faced - Big 6 vs Other 13

In all cases the big 6 shot on target more frequently than the other 13. This was pretty much to be expected as you would expect the better teams to produce goal scoring opportunities more frequently than lesser teams.

Shots per goal conceded - Bg 6 vs Other 13

This is where it gets interesting, there is no clear distinction showing keepers raising their game against better teams. Michel Vorm, Simon Mignolet and Mark Schwarzer all have excellent records against the big 6 and along with Paddy Kenny all actually conceded less goals per minute against the better teams than the others in the league.

Joe Hart is also in there, and although his record is good in both cases it’s better against the better teams, maybe a lack of concentration in the less important games due to being so well protected?

David de Gea and Brad Friedel come off worst, de Gea’s phenomenal numbers against the other 13 teams (Man Utd always seem to have an exceptional record against teams lower down the table) was always likely to drop against the better teams, but it remains as one of the better ones in the league. Brad Friedel shows a big drop and maybe the 5-1 home hammering Tottenham took by Man City (remember Dzeko’s 4 goals??) from only 9 shots on target skews the data and shows the fragility of how one result can affect this.

Unfortunately these statistics only show the tip of the iceberg. Is there a reason for keepers performing better against better teams? They are likely to have more defenders in the box making the shooting opportunity not as clear, maybe they take shots from further out. All this is speculation and until further detailed information is available speculation is all we have.

Another goal flying in past Paul Robinson

Video Analysis - Iago Aspas

While the football season has mainly ended for the major leagues around Europe, I’ve decided to spend some time getting up to speed on the major analysis platforms used. I’ve used many of these before (Prozone, Sportscode, NACSport, Dartfish etc) but there are more and more coming onto the market (such as S20APP).

 

The first one I decided to look at is LongoMatch. This is a free piece of tagging software available for Mac & PC and while it is stripped down compared to a lot of more comprehensive models I have to say it’s excellent for the job it does. You can set teams, key indicators and you have up to 16 customisable tags which you can then mark against teams, players and time. It’s easy to use and although when I’ve used it previously it had a few teething problems (such as when you did something out of the order it expected the system would crash and you’d lose everything since your last save) that seems to have been fixed now and I didn’t come across any problems.

 

 

 

So the example I used, while only loosely what I would call analysis focussed on Iago Aspas, Liverpool’s new £7m signing from Celta Vigo (although this hasn't yet been completed it's expected to be in the next few days). I already had a game downloaded (Celta Vigo vs. Athletic Bilbao from 03/05/13) and setting the key indicators to match the key Statszone indicators from the iPhone app. Once I had gone through the game and tagged the relevant points it is easy to pull whichever ones you need to into a playlist and export this into an mp4 file. I then used iMovie to cut this into a presentation using some of the graphics from the Statszone app (this can be done in LongoMatch – I just found this easier to do in iMovie, just personal preference)

 

Just to clarify, I’d consider this to be the very tip of analysis, it’s only 1 game and anybody can look good in isolation and although the graphics help with some context it is pretty standalone. With more games however, LongoMatch can be used to build up a library which allows for much more comprehensive video analysis over a season.