MCoA's cludgy team ratings, expressed in a likewise cludgy attack % better than league + defense % better than league:

+.97 Manchester City
+.63 Manchester United
+.51 Liverpool
+.40 Everton
+.39 Tottenham Hotspur
+.35 Chelsea
+.23 Arsenal
+.05 Swansea City
-.00 Fulham
-.03 West Bromwich Albion
-.11 West Ham United
-.13 Southampton
-.15 Stoke City
-.25 Queens Park Rangers
-.27 Norwich City
-.34 Newcastle United
-.40 Wigan Athletic
-.47 Aston Villa
-.60 Sunderland
-.78 Reading

Mine are quite similar.

MCI 89.2
MUN 59.9
LIV 39.2
EVE 32.8
TOT 31.9
CHE 30.4
ARS 24.5
SWA -4.3
FUL -5.0
STO -6.9
WBA -9.2
SOU -16.3
WHM -17.1
QPR -17.4
NEW -22.0
NOR -24.4
WIG -29.7
AVL -42.3
SUN -46.4
RDG -69.7

Stop stealing my stuff! :)

Because it's so close to midseason, and because I got bored, I didn't put together a strength of schedule adjustment. The main team I'd expect to be affected by a schedule strength adjustment are Tottenham, whose extra games have been against league-worst Reading and league-almost-worst Aston Villa.

The Spurs have the second easiest schedule by my metrics behind City. Liverpool and Swansea are next.

Most difficult are Sunderland, Stoke and Reading.

Obviously, these are impacted by the fact the good teams don't play themselves and vice versa. I haven't yet worked out a SOS metric excluding games against the team in question.

Toughest schedule for a top 7 team (and there is a very clear top 7 in my metrics) is Arsenal, quite a bit more difficult than Chelsea, Everton and Man U (who are all basically tied).

Their goalkeepers are ####### awful.
Yeah, I bet that's it. This method doesn't account for shot-saving. Southampton hasn't allowed many more quality opportunities than the league average, but they've let tons of those balls into the net.

A ton of that was early though and a result of playing some ridiculously good competition as a promoted team (City, united, Arsenal, Everton in first 6). The last 10, they've allowed only 10 goals, three of them is the disgrace at Stoke.

Southampton has quietly been a near mid-table club most of the year.

I'd say that when you have players with a track record, their actual scoring rate is a better predictor than "shots on target" or "big chances" or anything else.

Goal conversion hasn't been proven as a skill though, at least year-over-year. It's the BABIP of soccer to some extent. Accuracy seems to have some stickiness, but conversion seems to be as much about luck as anything else.

At least, that's where we stand right now.

Goal conversion hasn't been proven as a skill though, at least year-over-year. It's the BABIP of soccer to some extent. Accuracy seems to have some stickiness, but conversion seems to be as much about luck as anything else.

At least, that's where we stand right now.

Hasn't Suarez been converting many more chances this year than he did last year? That's the immediate example that springs to mind for me.

Who knows what goes through that guy's head. It's a fascinating question--what does Roman want?

Interestingly, I've never had much doubt about this. To me, Roman wants what Jerry Jones wants and what George Steinbrenner wanted (at least most of the time): glory, but glory on his terms. He wants Chelsea to win, of course, but he wants it on his terms, and if that means Torres and Sheva that means Torres and Sheva. And like Jones and Big Stein, Roman--it seems--thinks he knows more about his sport than he actually does.

Goal conversion hasn't been proven as a skill though, at least year-over-year. It's the BABIP of soccer to some extent. Accuracy seems to have some stickiness, but conversion seems to be as much about luck as anything else.

I get what you're saying and it's appealing on one level. But there are certain players who just never convert their chances. Maybe they spend a career being "unlucky". I doubt that; I've watched such players and they never get better. Maybe it's just the ability to be calm. Maybe it's a vision thing. But I believe there's something going on beyond luck.

People who think there is no difference between having Emile Heskey and Robin Van Persie when a chance is to be converted are out of their mind.

What the Chef said. Much pithier.

The sad thing is that half the English people I know would probably take Heskey.

Goal conversion hasn't been proven as a skill though, at least year-over-year. It's the BABIP of soccer to some extent. Accuracy seems to have some stickiness, but conversion seems to be as much about luck as anything else.

I third what Chef said. I simply cannot accept this. For one thing, you can improve conversion rates by coaching.

Coming back to the strength of schedule issue, do you attempt to adjust for home/away or is it strictly based on quality of opponent regardless of the venue?

I'm sure there are real and important differences in player skill at shot-on-target conversion, and it's crazy to think there aren't. (Just like with pitcher BABIP).

The issue is, like with pitcher BABIP, the skill isn't obviously recognizable from the numbers even in pretty large samples. I figure this is because the goal-scoring skill isn't just the ability to avoid having the keeper save the ball, it's also the ability to put a shot on target, and once the ball is on target, keeper skill and random chance are major causes of the outcome. (For pitchers, the goal isn't to prevent hits on balls in play, it's to prevent good contact. When non-HR contact is made, fielder skill, park effects, and random chance play major parts in the outcome.)

I mean, Manchester City are converting shots on target at a rate below league average. Their leading shooters are Tevez, Aguero, Dzeko, Toure, and Silva. Are those guys really poor finishers? I think it's usually a better practice to regress half-season conversion numbers back toward the league average quite a bit.

Coming back to the strength of schedule issue, do you attempt to adjust for home/away or is it strictly based on quality of opponent regardless of the venue?

I need to go back and adjust for opponent as well as home/away - so home/away is already in there. The home field effect is huge. This season, home sides have outscored away sides 333-264, with commensurate shots / shots on target numbers.

The home field effect is huge. This season, home sides have outscored away sides 333-264, with commensurate shots / shots on target numbers.

Any sense of how this compares with previous seasons? I'd been wondering if we'd see any statistically significant shrinking of the home/away disparity following the Premier League's standardization of pitch dimensions for this season.

I'm sure there are real and important differences in player skill at shot-on-target conversion, and it's crazy to think there aren't. (Just like with pitcher BABIP).

The issue is, like with pitcher BABIP, the skill isn't obviously recognizable from the numbers even in pretty large samples. I figure this is because the goal-scoring skill isn't just the ability to avoid having the keeper save the ball, it's also the ability to put a shot on target, and once the ball is on target, keeper skill and random chance are major causes of the outcome. (For pitchers, the goal isn't to prevent hits on balls in play, it's to prevent good contact. When non-HR contact is made, fielder skill, park effects, and random chance play major parts in the outcome.)

I mean, Manchester City are converting shots on target at a rate below league average. Their leading shooters are Tevez, Aguero, Dzeko, Toure, and Silva. Are those guys really poor finishers? I think it's usually a better practice to regress half-season conversion numbers back toward the league average quite a bit.

Exactly. No one is saying there isn't skill involved. However, when the entire league pretty much regresses to the same rate, it's hard to see how skill has much influence.

The good players get shots and get them on target. Once they are on target, it's pretty much up to luck and factors out of their control as to whether they score or not. At least that's what the current data shows.

It may be difficult to swallow, but BABIP was difficult to swallow as well. Doesn't make it incorrect.

Also of note is RVP. Would you guess that last season he converted at a bit below average rates for a forward once his penalties were removed? He was basically league average.

Hasn't Suarez been converting many more chances this year than he did last year? That's the immediate example that springs to mind for me.

Yep. Suarez is converting 37.5% of his SOT this season after converting just 23.4% last season. His percentage of shots in the box has actually decreased quite a bit, so there is really no explanation for the change other than regression to the norm.

Any sense of how this compares with previous seasons? I'd been wondering if we'd see any statistically significant shrinking of the home/away disparity following the Premier League's standardization of pitch dimensions for this season.

The past decade has seen about a 16% home field advantage judging by goals alone. This year, that number is running at 11.5%, so it does seem a bit smaller, although probably not enough to raise any flags.

I don't have the data in front of me but I believe the advantage was pretty random over the last ten years, moving up and down 6-7% per season.

EDIT: Arrggh. For some reason, I removed the year from my data set. I will put together another set with the year here in a few days and let you know.

Coming back to the strength of schedule issue, do you attempt to adjust for home/away or is it strictly based on quality of opponent regardless of the venue?

Oh and I know you weren't asking me, but my numbers do include venue location.

The good players get shots and get them on target. Once they are on target, it's pretty much up to luck and factors out of their control as to whether they score or not.

A huge factor in whether a goal is scored or not is the part of the target that is hit. A good forward, given time, will try to pick a spot that the keeper can't reach. Wouldn't you classify this as a factor within the control of the forward? (no snark intended, interested in the data and the research)

A huge factor in whether a goal is scored or not is the part of the target that is hit. A good forward, given time, will try to pick a spot that the keeper can't reach. Wouldn't you classify this as a factor within the control of the forward? (no snark intended, interested in the data and the research)

Intuitively, I agree. But, intuitively, it would seem it would be more difficult to get hits when making contact off Greg Maddux than off Greg Hibbard. Unfortunately, the data doesn't suggest that.

If you notice, I continue to use words such as "suggest" to illustrate how primitive soccer data is at this point. It may turn out there most certainly is an impact but the current crude data sets do not easily expose one.

People who think there is no difference between having Emile Heskey and Robin Van Persie when a chance is to be converted are out of their mind.

If Heskey gets it on target he has about as much chance of scoring as RVP, especially over the full range of an EPL season which includes big chances, half chances, and out of left field shots that somehow make it in.

I third what Chef said. I simply cannot accept this. For one thing, you can improve conversion rates by coaching.

You are going to have to show me some data for a statement as profound as this. Otherwise, one would thinkEPL teams with the best coaches would dominate conversion rates on a yearly basis. They don't.

You are going to have to show me some data for a statement as profound as this. Otherwise, one would thinkEPL teams with the best coaches would dominate conversion rates on a yearly basis. They don't.

Well, to give a simple example of what I mean, in a one on one situation where the keeper is advancing, if the forward hits the keeper with a shot, that's an accurate shot on target, but it's not converted into a goal. You can, however, teach a forward to lob the goalkeeper in that situation. An accurate lob is, in that scenario, a certain goal, because there's no one protecting the net - the keeper is helpless - and luck or other factors are irrelevant. Those factors are out of play because the player has been taught a way to take the keeper out of the equation.

It comes back to the same point I made in my other post you answered: a good forward has a better chance of picking a spot that the keeper can't reach and thus eliminating random chance, and therefore not all forwards who can hit the target are equal. There is certainly some element of randomness, but your premise seems to me to be too absolute.

Intuitively, I agree. But, intuitively, it would seem it would be more difficult to get hits when making contact off Greg Maddux than off Greg Hibbard. Unfortunately, the data doesn't suggest that.

I don't want to turn this into a DIPS discussion, but just to be clear - there are definite and highly important differences between pitchers in their abilities to prevent hits on balls in play. BABIP skill made Tom Seaver a Hall of Famer, it made Dave Stieb an ace, it made Sid Fernandez a professional ballplayer, and it made Andy Messersmith rich. If you look at Tango's fun DIPS spreadsheet, you see that there are far, far more players several SDs removed from the league mean than you'd see if it were truly random.

However, because of all the confounding factors, it takes a couple of seasons of play to get good data on BABIP skill for pitchers. Single-season FIP is a better measure of pitcher skill than single season ERA because it takes so long for BABIP to find its true level. On the sabermetric side, it took a bunch of different studies before the extent of BABIP skill could be identified.

My intuition is that shot-on-target conversion is like this. Factors of random chance, keeper skill, quality of opportunity provided by teammates will obfuscate the skill in even a relatively large sample. A few percentage points of skill in shot-on-target conversion would be worth a lot of money. So it can be a very real and important skill that you can't identify in a season of data.

Coming back to the strength of schedule issue, do you attempt to adjust for home/away or is it strictly based on quality of opponent regardless of the venue?

Oh and I know you weren't asking me, but my numbers do include venue location.

In case I misunderstood the question - I'm not making adjustments for the particular venue, it's just a generic HFA factor.

I've only been working off the two years of data in the FFS database, so I figured that any individualized park factor based on just 1.5 seasons of data would probably be more noise than signal. But I realize that G/GA for home and away seasons dating back for quite a few years are available on ESPN's website, so I perhaps should try to add that in at some point. I played with those numbers earlier in this thread (or in the last one) and the main thing I found was an unsurprisingly huge HFA at the Britannia. I think that Fulham had the smallest HFA. That could just be random, though.

Worse than North Korea.

FWIW, in hockey the exact same effect is seen with scoring rate (which is called shooting percentage there). Over like 5 seasons you'll start to see who actually shoots better or worse than usual. Some guys can sustain high shooting percentages by being crease crashers, but there's not much of that in soccer. On the one hand, hockey has many more shots per game (and more games per season) so you'd expect it to stabilize sooner, but on the other, something like 92% of all shots are saved, so you're going to see lots of randomness.

If Torres really is Abramovich's personal project, what are the chances that Roman decides to walk away now that his last two personal projects (Torres and Sheva) have done nothing but cost him money and earn him the opprobrium of the supporters?

Just got around to watching that match on BEIN late last night. Torres was just dreadful. Watching Ba create 3 chances in 10 minutes, while Torres was mostly invisible during a match where Chelsea just completely dominated the ball for a good 60 out of the first 80 minutes was just damning. It's bad enough that he hasn't been in great form this year, but his inability to deal with any competition whether it a legend like Drogba or just an inform striker like Ba should mean his exit. I say that they try to sell him for a Pato-like discount to Atletico the minute Falcoa walks out that door. This is bad for the squad and just painful for him. I don't hate the guy, but like Randy Johnson in NY this was just not a good fit.

Agree with the Steinbrenner comparisons, Roman wants to have the glory. I would love for him to leave the player acquisition to the experts at Chelsea who have been better than good especially during the last year + (Mata, Hazard, Luiz, Lukaku, Oscar who are all young and good), but I'm not confident when he decides to set his eyes on a guy. At least he passed on Walcott as that guy.

So, more time wasting in the morning, leading to more cluttering of the page. Sorry if this is annoying. I'm enjoying myself, though.

Anyway, I added in a strength of schedule adjustment to the ratings. I'm not sure to what degree the changes are because of that or because of other minor tweaks I added - I now run the attack and defense team quality adjustments separately, rather than producing a single +/- figure for every game. So that might also be the culprit in the small changes.

Stoke City gets a big boost, which is surely a schedule effect - the only clubs they've played twice this season are Liverpool and Manchester City. Likewise Newcastle, who have doubled up on Manchester United and Everton. Chelsea's drop in the ratings may be a schedule effect (they have only the one extra home game against QPR), but it could also be an effect of spreadsheet tweaks. QPR by contrast actually move down a bit, so who knows. I expected a bigger drop for Spurs.

EDIT: There's got to be a bug in the spreadsheet. This output is flawed somehow. Will fix at some point at re-post.

Great discussion going on. It seems to me that part of shot on goal conversion is quality of service. Not all shots on goal are equal, and not all passes to the shooter are either. I think we can all agree that a long ball from Pirlo or a cross from Leighton Baines is significantly better than from most players. That has to be taken into account somehow.

Oh and I know you weren't asking me, but my numbers do include venue location.

Grover: I was asking you and MCoA, so thanks for the reply. What I was trying to get at regarding SoS adjustment was, for example, whether you would weight playing Manchester United away differently than playing Manchester United at home. Actually, United might be a bad example since they're awfully good everywhere. Better would be clubs like Stoke and Fulham that have an established pattern of huge home/away splits that goes back several seasons.

How common are direct goals from free kicks? My intuition is not common at all.

How common are direct goals from free kicks? My intuition is not common at all.

I'd be amazed if the conversion rate is much better than about 1%. It seems to be a very, very difficult skill just to get the ball on target and force a save.

Okay, I found some data and analysis here and here. Consider me amazed. I think I may have been colored by watching Tottenham more than any other team, because apparently they actually are truly lousy at free kicks in addition to it being very hard to score that way in general.

I was just about to post that. Perhaps I missed it but it appears that counts all goals from direct kicks--including ones that deflect, and Zack was wondering how many go straight into the net. I don't think 1% is that crazy a guesstimate.

I'd be amazed if the conversion rate is much better than about 1%. It seems to be a very, very difficult skill just to get the ball on target and force a save.

Seems to be about 3%. SSS alert.

Would have guessed 5% myself.

475 seems well-stated to me. Part of the debate may be that some players simply don't hit the target very often. That might not get picked up in the stats MCOA is talking about.

Speaking of conversion rates, do you have the data for corner kicks? Seems to me that it takes an awful lot of corner kicks to score one goal.

Edit to add: An example of a player who began his career very wasteful of chances but learned quickly is Thierry Henry. Maybe it was SSS when he was a teenager, or maybe he was taught something, IDK.

So, I think I fixed the minor problem in the spreadsheet. So these are my cludgy team ratings. I've also listed team ratings by actual goals scored / goals allowed (% better than league) as a comparison. Adjusted for home/away and strength of schedule.

Based on Expected Goals / Expected Goals Allowed
1) +.97 (1.57 - 0.60) - Manchester City
2) +.64 (1.57 - 0.93) - Manchester United
3) +.42 (1.24 - 0.82) - Liverpool
4) +.37 (1.36 - 0.99) - Everton
5) +.36 (1.22 - 0.86) - Tottenham Hotspur
6) +.26 (1.14 - 0.88) - Arsenal
7) +.21 (1.11 - 0.90) - Chelsea
8) -.03 (0.77 - 0.80) - Stoke City
9) -.04 (0.94 - 0.98) - West Bromwich Albion
10) -.08 (1.02 - 1.10) - Swansea City
11) -.12 (0.92 - 1.04) - Southampton
12) -.16 (0.81 - 0.97) - West Ham United
13) -.18 (1.02 - 1.20) - Fulham
14) -.20 (1.12 - 1.32) - Newcastle United
15) -.22 (0.89 - 1.11) - Norwich City
16) -.27 (0.84 - 1.11) - Queens Park Rangers
17) -.36 (0.81 - 1.17) - Wigan Athletic
18) -.41 (0.72 - 1.13) - Aston Villa
19) -.46 (0.72 - 1.18) - Sunderland
20) -.80 (0.67 - 1.47) - Reading

Based on Goals Scored / Goals Allowed
1) +.81 (1.86 - 1.05) - Manchester United
2) +.67 (1.34 - 0.67) - Manchester City
3) +.65 (1.34 - 0.70) - Chelsea
4) +.60 (1.33 - 0.78) - Arsenal
5) +.45 (1.32 - 0.87) - Tottenham Hotspur
6) +.20 (1.18 - 0.98) - Liverpool
7) +.17 (1.18 - 1.01) - Everton
8) +.05 (0.69 - 0.64) - Stoke City
9) -.03 (0.99 - 1.02) - Swansea City
10) -.04 (0.81 - 0.85) - West Ham United
11) -.05 (0.97 - 1.12) - West Bromwich Albion
12) -.18 (0.73 - 0.91) - Sunderland
13) -.21 (1.00 - 1.20) - Newcastle United
14) -.21 (0.94 - 1.15) - Norwich City
15) -.26 (1.07 - 1.33) - Fulham
16) -.35 (1.02 - 1.38) - Southampton
17) -.49 (0.83 - 1.32) - Reading
18) -.52 (0.72 - 1.24) - Wigan Athletic
19) -.58 (0.61 - 1.19) - Queens Park Rangers
20) -.67 (0.61 - 1.28) - Aston Villa

It'll be interesting to see if either is a slightly better predictor of team results in the second half.

I note also Liverpool is pretty good at scoring and preventing goals, and their xG/xGA underperformance is equally as bad on the defensive side as on the offensive.

Spurs show there are no hard feelings by sending Jake Livermore to QPR on loan, reportedly. Will Wesley Sneijder move to Turkey? We will soon find out!

Turning from the statty to the emotive, it seems there's been a minor conflagration over ticket prices going on in England lately. Germans evidently think the English are nuts for paying more than €15 for a ticket. I wonder what the English, who are up in arms over £62 tickets, would think of NFL prices.

So what are Rangers F.C. complaining about? That the bottom division will be expanded from 10 to 18 which won't enable them to be promoted?

I wonder what the English, who are up in arms over £62 tickets, would think of NFL prices.

To be fair, if there were only eight home Bundesliga or EPL games per season, they'd be paying a lot more than £62 or €15 or whatever.

To be fair, if there were only eight home Bundesliga or EPL games per season, they'd be paying a lot more than £62 or €15 or whatever.

Judging by the NFL's attendance problems, maybe not...

Judging by the NFL's attendance problems, maybe not...

If the data in this chart is accurate (and yes, I know that's tickets sold, not butts-in-seats) then every league in the world would love to have the NFL's attendance problems. No team below 75% capacity, every team above 54,000 per game and five above 75,000 per game.

(and yes, I know that's tickets sold, not butts-in-seats)

Their attendance last year was the lowest since 2001 and, more importantly, there were a LOT of now-shows. Not great for a sport with only 8 home games a year. I don't think it's because people don't like football, it's just that watching football at home or at a bar is a much better experience than going to a game anymore. Why spend hundred bucks to go to a game, fight the traffic getting in, get gouged on parking, get gouged on food and drink, freeze your ass for 4 hours and then fight traffic to get home when you can get NFL ticket, invite your friends over and go from game to game and bore everyone with details about your fantasy team from the comfort of your Lazyboy?

Does anyone know what percentage of NFL tickets are season tickets? I would imagine the ease of reselling tickets makes attendance (# of tickets sold) lower. I think for a lot of people going to a NFL game it is all about the tailgate - meeting up with all your friends, drink some beers and gorge on food. That you can make a family outing makes it more palatable to wives and significant others rather than spending all Sunday in a bar or some guy's man-cave. In my opinion you go to a sporting event to experience the atmosphere. If you want to watch the game stay at home and have it on your HDTV.

I've only been to one European soccer match and had seats along one end line. I had no idea what was going on the far side of the field. It was exhausting - everyone was standing throughout the match.

It was exhausting - everyone was standing throughout the match.

Lucky. I WISH they'd let you stand at WHL. I think sitting down when the game is getting tense just drains the energy from the home crowd.

It was definitely a great experience and the section was full of energy. I couldn't understand a single chant/song the crowd had going even though it was English (it was Upton Park). I hadn't stood through a game since college and had been walking around London for three days already - my feet hurt.

flip