EPL Week 25 Bets

As in weeks before, my model gives slightly higher odds to a few of the long shots:

• Arsenal to lose to Everton: 13% versus 28%
  
  
• Sunderland to beat Chelsea: 18% versus 30%
  
  
• Man United to lose to Aston Villa: 9% versus 20%
  

In addition, these two matches look interesting:

• West Brom to lose to Wigan.
  
  The odds makers have this at 22%, whereas it is 36% in my model. I can't see why the odds makers would not see this as a close match. Perhaps they are reading more into the small differences in home and away records than I am.
  
  
• Bolton to lose to Wolves.
  
  The odds makers have this also at 22%, whereas it is 32% in my model. Actually, one model has it at only 28%, but 32% comes from the model that should be the most accurate.
  

My model also says Stoke have a better chance than expected, but given recent changes in Liverpool during the transfer window, I think it best to stay away from that one.

Review of EPL Week 24-mid Bets

More of the same in my mid-week bets: I predicted that two matches — Blackpool v Man United and Liverpool v Fulham — would be closer than the odds makers, an observer would probably say they were closer, but in the end, the expected team won.

The Blackpool v Man United game was particularly painful. Blackpool was up 2-0 at half time. It seemed very likely that an upset would occur. Indeed, the odds of a Man United comeback at half time were 15 to 1. But they did come back. Just as they did the in their previous match, and won the game.

Liverpool v Fulham was a close match that ended up 1-0 on an own-goal by Fulham. That is surprising for Fulham, which has had better than expected defense all season.

It's important to keep reminding yourself: "in statistics as in art, never fall in love with your model." That said, we still need more data points to see anything statistical significant.

Inferring Expected Goals for EPL Week 25

In previous posts, I talked a bit about how a mathematical model can be used to predict the odds of a win, lose, or draw in a given match. However, the process can also be reversed: we can take a set of odds and infer the parameters of the model.

Why would we want to do this? Because we can then learn the parameters that the odds compilers have in mind when they set the odds for the match. And in the case of a Poisson model (which is standard), this parameter is the expected number of goals.

Hence, we can find out the odds compilers estimate of the expected number of goals to be scored by each team. This is extremely useful information to have when choosing a fantasy team. Picking the top goal scorers from teams that are likely to score goals and defenders from teams that are unlikely to concede them is a sure fire way to score points.

I've gone ahead and done that using the odds for this week in the EPL. Here are the results:

• 2.31 — Manchester United
  
• 1.72 — Arsenal
  
• 1.62 — Chelsea
  
• 1.48 — Bolton
  
• 1.47 — West Brom
  
• 1.47 — Liverpool
  
• 1.38 — Manchester City
  
• 1.34 — Tottenham
  
• 1.25 — Fulham
  
• 1.21 — Blackpool
  
• 1.05 — West Ham
  
• 1.00 — Blackburn
  
• 0.97 — Newcastle
  
• 0.84 — Birmingham
  
• 0.78 — Wigan
  
• 0.75 — Wolves
  
• 0.70 — Stoke
  
• 0.70 — Sunderland
  
• 0.53 — Everton
  
• 0.53 — Aston Villa
  

Like any mathematical technique, this one has limitations. So let me point out potential places where this could go wrong:

• The odds compilers might not be assuming a Poisson distribution. This isn't a serious concern, in my opinion, since a Poisson model is very standard and fits these data very well.
  
  
• The odds compilers may not be assuming that goals scored by each team are (probabilistically) independent events. I have no reason to think this is not the safest assumption, however.
  
  
• The "over-round" may not be distributed evenly between win, lose, and draw. This seems like a potential concern. However, as these odds had an over-round of only 0.075, it won't affect these estimated parameters much in any case.
  

EPL Week 25 Fantasy Team

This week, Manchester United are by far the most likely team to score. The next best, quite some way down, is Arsenal. In my model, Bolton and Blackpool are the next closest after Arsenal, whereas the bookies have Bolton and Chelsea next. Of course, I will put my money where my mouth is and use my model for choosing a team.

On the defensive side, the most likely clean sheet is from Manchester City, followed by Chelsea, and then Manchester United. After that, it becomes crowded. Arsenal, Newcastle, Tottenham, Bolton, and Fulham are the next best choices and are all roughly equal to one another. The same applies to Birmingham, Liverpool, Stoke, Blackburn, and West Brom, which are only a tiny bit worse.

In general, there are lots of good choices on defense this week. Given that, it makes sense to choose all offensive players from the top 4 offensive teams. So I will have no defenders from Manchester United or Arsenal.

After we take defenders from Manchester City and Chelsea, we need only one more defender from any of that long list of good defensive choices above. I went with Martin Kelly from Liverpool as he has been playing well recently and delivering a good number of fantasy points per game.

I chose the same offensive players as last week from Manchester United and Arsenal — one striker and one midfielder each. That leaves one striker and one midfielder from either Bolton or Blackpool. From Bolton, the clear choice is Johan Elmander. For the final spot, I'll take the trendy Charlie Adam. His fantasy points per game over the season are less than, say, Stewart Holden from Bolton. But my sense is that Adam is in better form at the moment.

Here is the full team selection:

• Keeper: Hart (MAC)
  
• Defenders: Cole (CHE), Terry (CHE), Zabaleta (MAC), Kelly (LIV)
  
• Midfield: Nani (MAU), Fabregas (ARS), Adam (BPL)
  
• Forwards: Berbatov (MAU), van Persie (ARS), Elmander (BOL)
  

Looking Forward to Barcelona v Arsenal

Without doubt, Barcelona v Arsenal in the Champions League will be one of the most anticipated match ups this year. Negative Football is predicting a decisive win for Barcelona. Not only does Barcelona unquestionably have the best offense in the world, as we noted in our last post, they also have the best defense in the world. Arsenal is in the top 10 in both categories as well, but Barcelona has a substantial advantage in both defense and offense.

The primary threat on Arsenal's offense at the moment comes from Robin van Persie. The job of containing him will fall to Puyol, Pique, and Busquets on Barcelona. Of course, these competitors have faced each other before: in the world cup final. As you undoubtedly remember, Netherlands failed to score. Indeed, van Persie had only two shots, neither of which was on target.

Using the model discussed in our last post, we can predict the probabilities of a Barcelona win, draw, or loss. These are 51%, 21%, and 28%, respectively. The odds compilers have nearly the same predictions, with only a 5% shift from lose to draw. Certainly, these are nowhere near the numbers you'd get with, say, Manchester United versus Wigan, where the favorite would have an 80% chance of winning. But Barcelona are clearly the favorites by a substantial margin.

Best Defenses

In my previous post, I talked about offense and defensive ratings of Chelsea. In this post, I'd like to explain how these are computed (in rough terms) and use the same scores for other teams to give a list of the best defenses.

Offense and Defense Ratings

As described before, each team gets two numbers, their offense and their defense. The number of goals scored by a team are assumed to be given by a Poisson distribution with parameter offense (that team) - defense (their opposition). For each game, there are two numbers of goals, one from each team, to explain. We choose offense and defense numbers for each time so as to make all of these outcomes as likely as possible. (This is called "maximum likelihood estimation".)

Top 5 Defenses in the EPL

Applying this to all the matches in the English Premier League, we get the following scores:

1. Manchester City — 0.517
   
2. Chelsea — 0.501
   
3. Sunderland — 0.382
   
4. Manchester United — 0.348
   
5. Arsenal 0.318
   
6. Honorable mention: Fulham — 0.257
   

There is one surprise on this list (to my mind, anyway): Manchester United has slipped. Earlier in the season, their defense was about the same as Manchester City. Their performance against Blackpool (conceding two goals) can't have helped. It will be interesting to see if move back up over the coming weeks.

Top 5 Defenses in the World

If we do the same to teams in the Champion's League competition, and then extrapolate to the rest of the teams in each league, we get the following scores:

1. Barcelona — 0.52
   
2. Borussia Dortmund — 0.61
   
3. Chelsea — 0.64
   
4. Manchester United — 0.76
   
5. Sunderland — 0.77
   

(Note that these numbers are from slightly earlier this season, which is why Manchester City does not appear here.)

Hopefully, no one is surprised to see Barcelona at the top of this list. Negative Football has talked at length about Barcelona's great defense before. They are without doubt the best defense in the world.

Chelsea's Rough Patch (Part I)

Chelsea seem to have finally come out of the mid-season rough patch, where they slipped from top of the table to 5th place. However, the reasons for that decline are still unclear. Some have, for example, attributed it to the firing of Ray Wilkins. Recently, ESPN reported that Ancelotti attributed the decline to their pursuit of "sexy football".

(To be fair to Ancelotti, he does not say this exactly in the article. Nowhere does the word "sexy" appear in a direct quite. And Ancelotti merely says that they might have done better by parking the bus when they were in poor form rather than continuing to play their game the usual way. However, Chelsea's usual game is not sexy either.)

While the root cause of Chelsea's rough patch remains a mystery, the statistics are pretty clear about what went wrong on the field:

	Offense	Defense
Season (So Far)	1.692	0.501
Rough Patch	1.761	-0.024

The offensive number tells you how many goals Chelsea would score during an average match. This number stayed essentially unchanged during the rough patch. (A difference of .07 is just noise.) Chelsea have had the 3rd best offense so far this season (after Arsenal and Manchester United), and during the rough patch they had the 4th best. So not much changed offensively in either absolute or relative terms.

The defensive number tells you how many goals Chelsea would prevent the other team from scoring, that is, how much their defense reduces the average goals scored by their opponents. A drop of 0.5 (half a goal) is NOT just noise. In relative terms, Chelsea has the best defense over the season so far, but during their rough patch, they slipped to 13th in defense, worse than the stalwart defenses of Wigan, Blackpool, and Arsenal.

The lesson: Chelsea started losing matches because their defense fell apart.

EPL Week 24-mid Bets

There are three matches being played in the middle of this week, each of which was postponed from earlier in the season. On two of them, I see some deviation between my models and the odds compilers:

• Blackpool to win or draw, versus Manchester United.
  
  Odds compilers have an MAU win at 75% versus 42% for me.
  
  
• Fulham to win, versus Liverpool.
  
  Odds compilers have a Fulham win at 17%, whereas I have it at 31%.
  

My sense of what is going through the odds compilers minds is their estimate of the current form of each of these teams. Manchester United just won big, while Blackpool lost. Liverpool seemed to have found confidence again under Dalglish, while Fulham just had a frustrating defeat.

My estimates, on the other hand, are based on the results of these teams over the season, with no special weight given to the last game. For example, Manchester United have 2 wins and 8 draws when playing away. Blackpool are capable of scoring goals. And though they concede many as well, Manchester United play more conservatively away (they play for draws, it seems). So at least from the data, a Manchester United win does not seem 75% certain.

It will be interesting to see the results over the next two days. The BPL vs MAU match is another of the sort that I did poorly on last weekend, so I will be especially interested to see if that run continues.

Review of EPL Week 24 Bets

There were many surprising results this week in the EPL. Liverpool, Sunderland, and Aston Villa all played better than the odds compilers were expecting. But let's focus on the matches I highlighted before:

• Wigan to win. (loss)
  
• Birmingham to win (or draw). (loss)
  
  The odds compilers had these at ~15% chance each, whereas I had them at ~25%. However, both teams got beat badly. Very badly. I also rated the odds of a draw for Birmingham at 30% versus 8% for the odds compilers. That part looks especially bad for my model.
  
  It will be important to focus on matches like these in the remaining weeks because these get at the heart of "how random" these outcomes really are. And of course, that issue is at the heart of modeling them mathematically.
  
  
• Bolton to win. (loss)
  
  This match was also a surprise. The odds compilers were expecting a draw, whereas my models showed that as unlikely. However, I had this as a fairly even match, whereas it was not in practice. Even the commentators I listened to were surprised by the way this matched leaned so heavily toward Chelsea.
  
  I'm not sure what we can learn from this one.
  
  
• West Ham to win. (loss)
  
  To be fair to myself, this match was very close. West Ham was ahead for most of the match, and the final goal for Everton came in injury time. I generally prefer to try to learn from the results of matches (rather than arguing what might have been), but I do think it is fair to say that this was a much more even match than the odds compilers expected. My models showed it much closer to even (though still an advantage for Everton), so that part at least looks good.
  
  
• Newcastle to win. (loss)
  
  Like that above, this match changed outcome in the last few minutes. Newcastle was ahead until an injury time equalizer from Spurs. The odds makers had Tottenham as more likely to win, at ~45%, whereas I had Newcastle to win at ~45%. So one could try to argue that my model was more accurate, as this ended up fluctuating between a Newcastle win and a draw, but I'm not going to pursue that further.
  

Overall, a bad week for my models in terms of outcomes, though there may be some positives in the details. Still too early to learn anything with statistical significance.

Review of EPL Week 24 Fantasy Team

This week's fantasy team did very well. On soccerlens, I won my group with 62 points. The next highest was 52.

Here are the high points:

• Not only did Berbatov (11 points) have a hat trick, but also Rooney had two assists giving him 8 points.
  
  
• Arsenal also scored many goals as expected. Here, van Persie (captain, 18 points) had the hat trick, while Fabregas (6 points) had an assist.
  
  
• Chelsea kept a clean sheet, giving 6 points to Cech and Cole.
  

The only substantial low point was Everton's unexpectedly poor performance. Both failing to score or assist and failing to keep a clean sheet gave only 1 point each to Baines and Coleman.

Of course, with so many games, you can't expect everything to go right. Indeed, I don't expect to get this lucky very often.

EPL Week 24 Bets

Two changes in methodology compared to last week. First, I've continued to improve my model (via techniques I will describe at a later date). Second, I also included my wife's model. Unlike me, she actually trained as a statistician. By comparing our two approaches, I get some amount of sensitivity estimation.

First, all of these long shots seem under valued:

• Wigam to win. 7% versus 26% in my model.
  
  
• West Ham to win. 15% versus 31% in my model.
  
  
• Birmingham to win or draw. 25% versus 56% in my model.
  

This may, of course, be evidence of the (alleged) general tendency of book makers to lower the odds (i.e., up the payout) on long shots because betters tend not to pick them. It may also be evidence of a general tendency to underestimate the amount of randomness in football matches. Or it could be that my models are just plain wrong. Time will tell...

Here is one more match that appears to be a long shot, according to the bookmakers:

• Bolton to win. 18% versus 33% in my model.
  
  The bookmakers have a draw with Chelsea at 62%. This at least reflects the fact that the quality of these teams is much closer than in the long shot matches above. However, a 62% chance of draw seems far too certain. My model has this at only 27%. Either team to win is more likely in my model.
  

Finally, there is one match where a draw seems more likely than the odds would suggest:

• NEW vs TOT to draw. 30% versus... 25% in my model.
  
  Actually, what my model suggests is a Newcastle win is more likely than the odds suggest. However, my model does not account for all the injuries at Newcastle. In particular, the loss of Andy Carroll reduces the odds of a Newcastle win. But on the other hand, this makes a draw more likely, so that is the position I've taken.
  

UPDATE: Fixed some of the numbers above after fixing a bug in my model.

EPL Week 24 Fantasy Team

Feeding this week's games to my models, here is what I notice:

• Newcastle looks surprisingly likely to score.
  
  And with Andy Carroll out, goals are likely to come from Kevin Nolan or Joey Barton.
  
  
• Everton looks surprisingly unlikely to concede.
  
  This is, of course, because they are playing West Ham United. Everton have a pair of fullbacks, in Leighton Baines and Seamus Coleman, who are both attacking threats. And while Negative Football does not endorse such behavior, it is great for scoring fantasy points.
  
  
• Arsenal and Manchester United are both the most likely to score and least likely to concede.
  
  The problem with both teams is guessing where those goals will come from. There are simply too many scoring options in both teams' midfields to find a safe bet. On the other hand, there are some options at striker. Van Persie has been in incredible form recently, and he is Arsenal's sole striker, so he seems the safest option. From Manchester United, I picked Berbatov, even though he has been somewhat up-and-down, since Rooney's form is still questionable.
  
  That leaves a choice in defense from each team. From Manchester United, I went with Vidic simply because he is almost certain to play. From Arsenal, somewhat surprisingly, I chose Szczezny, mostly because none of the defenders look promising.
  
  
• Manchester City is also likely to score and nearly the least likely to concede.
  
  Manchester City presents similar problems in that they have lots of scoring options. However, Tevez up front is certainly the safest bet. In the midfield, Adam Johnson's form in the last game was incredible, but Silva is also a good choice since he appears to be healthy again and is likely to get at least an assist.
  
  
• Stoke is nearly as unlikely to concede as those above.
  
  Fulham is not far behind this, but Stoke looks even better. Huth is an easy choice here.
  

All of that adds up to the following lineup, a 4-3-3 formation.

• Keeper: Szczezny (ARS)
  
• Defense: Baines (EVE), Coleman (EVE), Vidic (MAU), Huth (STO)
  
• Midfield: Nolan (NEW), Barton (NEW), Silva (MAC)
  
• Strikers: van Persie (ARS), Berbatov (MAU), Tevez (MAC)
  

UPDATE: Ran some improved models. Here are the interesting changes:

• Manchester City no longer looks so likely to score. On the other hand, they are facing Darren Bent, who will be out to prove something. Perhaps best to stay away from this one.
  
  
• Chelsea looks unlikely to concede. Added to defense. This also frees up Manchester United for all offense.
  
  
• Blackburn look surprisingly likely to score.
  

That leads me to following lineup, another 4-3-3 formation.

• Keeper: Céch (CHE)
  
• Defense: Baines (EVE), Coleman (EVE), Cole (CHE), Bardsley (SUN)
  
• Midfield: Nolan (NEW), Fabregas (ARS), Pedersen (BLA)
  
• Strikers: van Persie (ARS), Berbatov (MAU), Rooney (MAU)
  

Bagchi on a Great Defender

Rob Bagchi writes, in his post at the Guardian, about how great defenders are frequently overlooked when awards are handed out, despite their supreme importance. Manchester United's Nemanja Vidic was a case in point during last weekend's game against Tottenham:

Vidic is held in similar regard by managers who know that if they could take one player from Old Trafford who would destabilise Manchester United's chances while bolstering their own, the Serbia centre-half fits the bill. And yet he was passed over on Sunday for Sky's man of the match award despite giving a masterclass of power, pace and positional awareness at White Hart Lane that held Tottenham at bay.

Cheers, Rob!

P.S. Need I mention that Vidic is on the author's fantasy football team? No, you knew that already.

Mancini on Great Defense

After Manchester City's 4-2 win over Leceister:

It is good for fans, but I prefer a 4-1, a 1-0, or a 2-0. If you don't concede a lot of goals, at the end of the season, you are sure to win something.

Cheers, Roberto!

Review of EPL Week 23 Bets

Here's how the bets went this week. We recommended upsets (draw or underdog winner) in 4 matches:

1. West Brom vs. Blackpool: 3 - 2 (loss)
2. Stoke vs. Bolton: 2 - 0 (loss)
3. Wigan vs. Fulham: 1 - 1 (win)
4. Sunderland vs. Newcastle: 1 - 1 (win)

These bets were all 45-60% odds, so winning two was expected. In the long term, we should expect to win 3 more often than the odds would predict, if in fact my models are more accurate. However, we can't say much from just 4 data points.

One reason for optimism, though, is that the WBA vs. BPL match seemed much closer than the odds predicted (which matches my model). It honestly seemed like it could have gone either way. WBA got the winner very late in the game, and even then BPL had a great chance to equalize again, though it did not come to pass.

That said, let's not give in to wishful thinking here. It's the actual results that matter. And we need more data points to say anything conclusive.

Our model also had things to say about two other matches:

5. Man City vs. Wolves: 4 - 3

The effect of Dzeko on this game was interesting. He did not score, but he removed pressure on Tevez, who then scored twice. City scored 4 all together, which is in line with what the odds say (as Chelsea and Arsenal actually scored fewer). City conceding 3 was not expected. They made some rare defensive errors.

6. Liverpool vs. Everton: 2 - 2

This match was very even overall and could have gone either way. That is in line with my model and not with the odds.

What can we learn from this week's results?

First, based on Liverpool's continued poor results, we get evidence for the null hypothesis that managers are not all that important. On the other hand, Inter's improved performance with Leonardo as manager is evidence of the opposite. On the other other hand, Inter are perhaps getting healthier, and of course, the models suggested they should have been winning these matches anyway.

Second, based on Everton's good performance even without Cahill, we get evidence for the null hypothesis that individual players are not all that important. On the other hand, the impact of Dzeko at City is evidence of the opposite.

As with our models, we will need more data to say anything definitive. We will look back at this later in the season.

Negative Football's Top 25

Dissatisfied with Tommy's Top 25, I decided to make my own. Unlike Tommy's list, though, this one is not compiled using a "secret formula" (i.e., Tommy's personal opinion). I will describe the exact formula below.

But first, here is the list:

1. Barcelona (ESP)


2. Manchester United (ENG)


3. Arsenal (ENG)


4. Real Madrid (ESP)


5. Chelsea (ENG)


6. Manchester City (ENG)


7. Borussia Dortmund (GER)


8. Tottenham Hotspur (ENG)


9. PSV Eindhoven (NED)


10. Villarreal (ESP)


11. Bayer Leverkusen (GER)


12. Bayern Munich (GER)


13. TSG Hoffenheim (GER)


14. Valencia (ESP)


15. Atlético Madrid (ESP)


16. Mainz (GER)


17. AC Milan (ITA)


18. Ajax Amsterdam (NED)


19. Espanyol (ESP)


20. Internazionale (ITA)


21. Twente Enschede (NED)


22. Juventus (ITA)


23. Heerenveen (NED)


24. Palermo (ITA)


25. Napoli (ITA)

Here's how this list was produced.

First, I computed the offensive and defensive power of each team in terms of goals per game scored and conceded, respectively. Then, I computed an overal power by combing the two scores as 60% offense minus 40% defense. (The minus is simply because fewer goals conceded means more power not less.)

Next, I gave each power an adjustment based on which league it is from. In other words, scoring 4 goals per game in Spain might not mean the same thing as 4 goals per game in England.

Here are the adjustments I used:

• ENG: 0.95


• ESP: 0.5


• GER: 0.35


• ITA: 0.15


• NED: 0.0


• FRA: -0.85

These adjustments were produced by finding a best-fit, log-linear (Poisson) model to the results from Champion's League matches, where teams from different leagues play one another. The resulting values do not seem unreasonable.

Finally, I limited each league to the number of Champion's League spots it has. This was needed so that England doesn't dominate the list. But it's also true that the modeling above uses only data from teams in the Champion's League this year, so there is reason to suspect that it is not accurate for the lower teams in each league.

The Great Defense of Barcelona

Barcelona are perhaps the team most associated with "positive" football today. In contrast, their incredible defense is rarely mentioned.

It will be harder to hide this week, however, as both of Barcelona's center-backs were chosen as part of the FIFA World XI — the best 11 players in the world, covering all positions. Barcelona's right full-back, Dani Alves, was chosen last year. Hence, it is easy to argue that Barcelona has the best defensive backs in football today.

However, Barcelona's defense does not end at its defenders. Watching them play, it is nearly impossible not to notice the relentless pressing of the forwards and midfielders. They put constant pressure on whomever has the ball. As a result, they often win back possession in the opponent's half of pitch, while the opposing players are running away from their own goal, a situation likely to produce good scoring opportunities.

Of course, Barcelona is best known for a style of play that focuses on possession. This too is often associated with a "positive" approach. But of course, it is just as important for defense as offense since the opponents cannot score without the ball.

To see the strength of Barcelona's defense, one need only look at their results in the Spanish Premiera Division. Out of 18 games played so far, they have kept 10 clean sheets. In 7 games, they conceded one goal. Only once have they conceded two goals in a game.

UPDATE: Corrected an error above. Barcelona lost 0-2 to Hercules in the second match of the season.

Some more stats on Barcelona's defense. This season they are averaging just 0.56 conceded goals per game (gpg). They have conceded one fewer goal than Mourinho's Real Madrid — and Mourinho is well-known for his defensive focus — even if we exclude the 5 goals conceded to Barcelona! Barcelona are doing much better than Manchester City, the best defense in the English Premier League, at 0.73 gpg. They are doing much better than Genoa, the best defense in Italy, at 0.83 gpg. The closest team in the top leagues is Borussia Dortmund, who are averaging 0.59 gpg conceded.

It's interesting to note that the next best defensive team in the Bundesliga is averaging 1.12 gpg conceded. Borussia Dortmund's advantage in defense is much larger in percentage terms than their advantage in offense: they have conceded 10 goals, while the next best is 19, and they have scored 39, while the next best is 35. The second place team Mainz has the second best defense (naturally!), conceding 19 goals, and while their offense is not second best, they have still scored 30.

Sacchi on Great Defense

Arrigo Sacchi's AC Milan team, considered one of the best of all time, was well-known for its beautiful, "positive" football. Readers of this blog should not be surprised to learn that Sacchi's success was also due to great defense. As Jonathan Wilson writes in Inverting the Pyramid:

His achievements [at Fiorentina] got him the manager's job at Parma, then in Serie C1. He won promotion in a first season in which they conceded just fourteen goals in thirty four matches — his attacking principles were always predicated on a sound defense.

Cheers, Arrigo!

Off Topic: The Bias for Home-Grown Players

Like the cult of "positive football", another bias I do not understand is the one in favor of growing your own players. Many seem to feel that teams who use players that came up through their youth system are morally superior to those who trade for players.

Not only do I not understand the argument in favor this, it seems quite reprehensible to me. Why should teams be showing preference to players that come through their youth system as opposed to those from elsewhere? It certainly isn't a players fault if they grow up in a different city or even a different country, which will limit the likelihood that they come through the youth system of <<your favorite team>>.

Similarly, why should there be a bias against players who develop later in life? The example of Edin Dzeko is apt. Not only did he grow up in another country (Bosnia) but in a system where his attributes were seen as a hinderance rather than a help. It wasn't until he secured a transfer to another league that got a real chance to prove his worth and develop.

My best guess is that people like growing players because it tends to correlate with stability: a player that comes up through the youth system of your team is likely to stay at your team for a long time. But then, let's concentrate on stability and not on growing players. And stability simply means that teams should try to acquire players that are just starting out at the top level and keep them for a long time. This can be done without any particular preference for home-grown players.

For teams, growing players is really an investment opportunity. If the youth team can turn out a strong percentage of players who go on to play at the top levels (either at that team or elsewhere), then the team will make money.

However, growing players is not even the only investment opportunity for teams. Having great scouts is another. Sir Alex has demonstrated this, trading for the likes of Wayne Rooney, Christiano Ronaldo, Chicharito, etc. while they were still early in their careers.

But of course, there is no particular reason why this type of investment has to go together with running a top-level football club. There could just as well be clubs that focus in developing players and have no professional team at all. Similarly, there could be professional teams without youth clubs. They could focus on making money by selling tickets and winning football matches. Indeed, if that is the only way the club makes money, they are, if anything, more likely to focus on results.

EPL Week 23 Bets

According to my simple models, there are four EPL games next week where the odds are backward, that is, games where the book makers think one side is more likely to win but my models say the other side is more likely. Here they are:

1. West Brom vs. Blackpool: -125 +240 +350 (order: WBA draw BPL)
2. Stoke vs. Bolton: +115 +230 +230
3. Wigan vs. Fulham: +150 +180 +220
4. Sunderland vs. Newcastle: +120 +225 +225

There are two others that seem backward, but seem too risky for other reasons.

5. Man City vs. Wolves has the same odds as Chelsea vs. Blackburn at -350 +400 +1000. The book makers give worse odds to Arsenal over West Ham at -200 +300 +550. My model has Arsenal and Chelsea wins as more likely than a Man City win, although all three are likely.

I'm skipping out on this bet, though, because of a new variable: the arrival of Dzeko for Man City. He introduces a new attacking threat, so it's quite possible that Man City will start scoring more goals, which would make a Man City win more likely.

6. Liverpool vs Everton at -115 +330 +230 has Liverpool as the likely winner (53% chance of win). My models have the teams about equal with Everton ahead under some settings.

I'm skipping out on this bet because Liverpool are changing managers. It's hard to know how the team will react to the replacement of Hodgson by Danglish. If they react like Inter did to Leonardo this week, then a Liverpool win may very well be likely.

Predicting Everton performance is difficult in general. They seem to play better against "big teams" than anyone would guess. Is Liverpool currently a big team? Will Everton step up? I'm not sure.

UPDATE: Newcastle midfielder Tiote was just suspended for 3 games. That makes a bet in favor of Newcastle look less like a good idea.

Tommy Smyth on Great Defense

Tommy Smyth (of ESPN's Soccernet Press Pass) on Man City's "negative football":

Is it entertaining? No. But if you want entertainment, go to the movies. This is football.

Cheers, Tommy!

Arsenal with Defense is Invincible

Arsenal is a team associated with "positive football". They have little love for defensive minded teams, as their ranting at Manchester City after the draw last week clearly demonstrated.

The greatest Arsenal team were the "Invincibles" who went unbeaten in the 2003-4 Premier League season. They are held up as the example to which all teams (not just Arsenal) should aspire.

Yet, Arsenal fans rarely mention the Invincibles defense, which was perhaps the key to its success. Indeed, as the Wikipedia article mentions, "former Arsenal manager George Graham pinpointed [their] success to defensive improvements".

The Invincibles allowed only 26 goals that season, the best in the Premier League. The next best team (Chelsea) allowed 30. (By the way, not only did the best defense (Arsenal) finish 1st and the 2nd best (Chelsea) finish 2nd, the 3rd best (Man U) finished 3rd, the 4th best (Liverpool) finished 4th, and the 5th best (Newcastle) finished 5th. Notice a pattern here?)

Looking back at their defense, we can see that it included some of the best defenders of the time: Ashley Cole, the left back, has often been rated the best in the world at that position. Sol Campbell, a center back, was named the best at his position in the Premier League that season (and other seasons). Kolo Toure, the other center back, is similarly great, and in fact, is now the center piece of Manchester City's dominant defense, which the Arsenal fans deride.

Arsenal fans have a strange blind spot regarding defensive prowess, not only of their own teams but others as well. Barcelona, the best team in the world at the moment, is often held up as a model of positive football. Yet they seem to never point out that Barcelona also has perhaps the defense in the world, with a back four of Alves, Puyols, Pique, and Abidal.

Zonal Marking on Great Defense

Michael Cox (of Zonal Marking) says, in his recent ESPN article about Manchester united:

... like most teams on long unbeaten runs, United's strength is its defense.

Cheers, Michael!

Man City: the "Blue Wall"

During the game between Manchester City and Arsenal, ESPN commentator Steve McManaman commented that Arsenal's shots were "bouncing off a wall" in the Man City defense.

Perhaps it is time for some fanciful titles for great defensive (association) football teams, like those in American football. I like McManaman's analogy, so the top defensive team this year, Manchester City, will be henceforth known to me as the "Blue Wall".

In Defense of Defense

Listening to some football fans, you'd think the winner of a football match were decided by judges rating the two sides on artistic impression and technical difficulty. But football is not gymnastics. In point of fact, the winner is simply the team that puts the most bulges in the old onion bag. As in most sports, every team needs to focus both on scoring goals and preventing the other team from scoring.

The old adage is that "offense wins games but defense wins championships". This is as true in football as it is any other sport. Indeed, simple statistics show that preventing one more goal per game will earn you 10-20% more wins than scoring one more goal per game.

Yet, many soccer fans not only ignore the defensive part of the game but openly deride it. Defensive minded teams are called "negative" and "boring".

The idea that defense is boring seems somewhat peculiar to (association) football.

In baseball, the central player in the defense -- the pitcher -- receives at least as much attention as the top hitters. The Cy Young award, given out to the best pitcher, is hugely anticipated each year. Even relief pitchers (especially closers) receive much of the glory. The mystique surrounding Mariano Rivera being a perfect example.

In basketball, similarly, the best teams of all time include many teams known for their defense, like the Detroit Pistons, the New York Knicks, and the San Antonio Spurs. The Celtics dominant player for many years, Bill Russell, is today primarily lauded for his defense.

Perhaps the sport where good defense is most appreciated is American football. There, we find many of the best teams of all time are primarily known for their defense. They even acquired fanciful nicknames like the "Orange Crush", the "Big Blue Wrecking Crew", the "Purple People Eaters", and the "Steel Curtain".

I still vividly remember watching the Washington Huskies play during their national championship season in 1995. Like any fans, we applauded great offensive achievements, but the loudest applause of the games would come after a string of great defensive plays. When the total rushing yards of the opponents would become negative (which it did quite often), the crowd would go bananas.

Of course, (association) football is not American football, nor should it be. However, defense is critical to winning trophies in both sports, so it remains odd that defense is so poorly appreciated by (association) football fans.

Here at Negative Football, we will see if we can do our small part to help fix that....