Also, it will probably be forgotten in the Bale hysteria, but Hugo Lloris is such a good keeper. That save on Taylor was as good as Bale's winning strike. I don't think Taylor imagined a keeper would be on top of him so quickly.

That was nuts. Lloris was so calm. Taylor gets put through. Lloris has a quick peek to see where the runner (Carroll) coming from his right is. He knows if he charges out now, Taylor only has to square it for Carroll for a simple tap-in. So he waits. Taylor drops his head and shapes to bend it into the far corner. In the time it takes him to do that, Lloris covers about five yards and smothers the shot as it's barely left Taylor's boot. Sensational.

Also, Benny is awesome even when he doesn't play:

@AssouEkotto: Fergie what????? Naaaaa Not N'E MORE .... well done spurs lol

What would say the chances are that Villas-Boas was totally tripping on X tonight?

Exhibit A: Gareth Bale
Exhibit B: Scott Parker
Exhibit C: Bale again

What would say the chances are that Villas-Boas was totally tripping on X tonight?

He's just happy that he finally found players willing to celebrate with him on the sideline (to be fair, the Parker hug looked only semi-consensual).

Question for the stats guys here. What do you do with missed calls? Suppose, for example, the keeper tips the ball wide, and we can see it on replay, but the ref misses it and rules goal kick. Is that a SoT or not?

No idea how Opta rules that. I will try and determine next game I watch. My guess is that would not be a SOT.

So at this point where does Bale rank in terms of quality of season. I'm not asking if he's the best player in the world but has he been the MVP of soccer in 2012-13 so to speak? I suppose Messi's performance is so brilliant that it's probably Trout-esque but while Tottenham are good, they are not Barcelona good. Without Bale Tottenham would be...well hell, a lot further down the table while Barcelona would probably still be atop La Liga.

Am I just buying into EPL/recent events hype or has Bale been THAT good? Because to me the way he carries that team seems like he has been that good.

It's not Bale's fault, but it would be extremely difficult for anyone to win a worldwide "MVP" without participating in the Champions League or a major continental competition. There are certain similarities between Bale's 2012-13 and Falcao's 2011-12. In addition to a "Europa League discount", each of them was/is disadvantaged by the fact that the Ballon d'Or remains a calendar year competition.

Bale is likely to win both the writers' and players' Footballer/Player of the Year award in England, though (as Van Persie did last season).

No idea how Opta rules that. I will try and determine next game I watch. My guess is that would not be a SOT.

If it's like every other sport I'm familiar with, it's a random distribution. And varies by home ground if the stringer is based locally.

He's just happy that he finally found players willing to celebrate with him on the sideline (to be fair, the Parker hug looked only semi-consensual).

Damn straight. Do you have any idea how much work went into the celebratification Powerpoint presentation?

Bale is likely to win both the writers' and players' Footballer/Player of the Year award in England, though (as Van Persie did last season).

The van Persie analogue is a good one, I think. However, I have to say that as much fun as it's been watching Bale take his game to the next level, he's only really been at that level for about six weeks. He was having a fine season before that, of course, but this Roy of the Rovers business is a more recent development. Let's see if he keeps it up over the next couple months, because unless somebody else steps up for Spurs that's when some more Bale magic is going to be needed. Here is Tottenham's fixture list for the next couple months:

Arsenal
Inter
@Liverpool
@Inter
Fulham
@Swansea
Everton
@Chelsea
Man City

That's some pretty tough wheelin'. Spurs could very conceivably play quite well over that stretch and only come away with about 7 points from 21 in the league and a who knows what result against Inter.

Jonathan Wilson on everyone's favorite Welshman

It's all down to Lloris. Well, not really, but the addition of Lloris to Spurs has fascinated Wilson all year. I think this is the 3rd or 4th article he's written about how Lloris has changed the shape of the team.

It's all down to Lloris. Well, not really, but the addition of Lloris to Spurs has fascinated Wilson all year. I think this is the 3rd or 4th article he's written about how Lloris has changed the shape of the team.

I think he's pretty clearly Tottenham's best signing of the season now. Which is saying something when the competition is Dembele and Vertonghen.

Chelsea scored 47 and conceded 32 during AVB's 27 league games in charge last season.

Through 27 games this season, Spurs have scored 47 and conceded 32.

FC Shootyburg off to a rough start in the DFB Pokal quarters.

Wowza. Elfmeter for FC Shooty in Nachspielzeit.

And Freiburg convert.

Going to extra time with Mainz down to ten men.

You missed an opportunity to mix in Ausgleichstreffer and Verlängerung in your last post.

That would have been showing off.

Quite an impact substitution of "The Great" Santini there. When he came on in the 84th minute, they were 2-0 down. He scored within two minutes and then drew the penalty that allowed them to equalize.

Pique was so late on that tackle of C Ronaldo that it would have been more fitting in the next Clasico.

Penalty converted, and Madrid lead on aggregate.

Schlusspfiff.

Freiburg are through to the semis.

Barcelona are really being whiny tonight. Hard to watch.

"The Great" Santini

That's a great movie, sports fans.

That's some pretty tough wheelin'. Spurs could very conceivably play quite well over that stretch and only come away with about 7 points from 21 in the league and a who knows what result against Inter.

Yeah, the more I have looked at the schedule the more I need to back track off my earlier prediction regarding UCL qualification. Both Chelsea and the Spurs have a hell of a run while Arsenal will be favored to win in all but maybe four games the rest of the way. Plus, the Spurs offense has really hit a snag without Defoe and seem 98% reliant upon Bale magic. That can work for awhile, but at some point they are going to need contributions form someone else.

I still think the Spurs qualify but I am not as confident as I once was.

Genuinely humiliating evening for Barcelona.

Now 0-3.

Will be interesting to see if there is any reaction in the rematch in the league, and how much the team is changed before the second leg against Milan.

Will be interesting to see if there is any reaction in the rematch in the league, and how much the team is changed before the second leg against Milan.

The popular explanation seems to be that fielding Cesc and Iniesta together on the left is a major bungle, which is what Barca did against Milan as well. This is, I believe, the exact same team (save Valdes) that had its pants pulled down by Max Allegri last week. Lesson learned now?

The steadfast refusal to buy a real center back probably doesn't help either.

I think that Roura believes that he lacks authority to change the team. Whether that is true or not, I have no idea, but I find it hard to believe that he doesn't see the weaknesses that are apparent to everyone else. The managerial situation is clearly a bit of a mess.

All that said, credit to both Milan and Madrid for taking full advantage of the situation.

Wow, haven't watched the game yet, but what a result. Not surprised about the win, but the scoreline is certainly shocking.

Freiburg are through to the semis.

You mean Halbfinale.

Here's hoping Dortmund pull their Real impersonation and take out Bayern in the Pokal tomorrow. After today's debacle, I need some faith returned and all that. Today was tough, but completely unsurprising. Barca's been slipping throughout this stupid "Messi has scored in X games in a row" shenanigans. I fear the Milan return fixture in ways seemingly impossible when the draw was announced, but that's football.

So how did I spend my evening? I made a moderately poorly-modeled projection of the rest of the Premier League season. Something like Cool Standings or BPro projected standings for the Premier League.

-I based it off my team ratings I've been posting here, using 2/3 team quality based on expected goals, 1/3 team quality based on actual goals. (I chose those weights entirely arbitrarily).
-This is where the modeling gets moderately poor. If you fudge the normal distribution a little this way and a little that way, you can get a curve that looks a lot like actual goal scoring, so I did that to model goal-scoring. (I can write up the model in more detail if anyone isn't bored already.)
-One problem here, beside the fact that this curve is non-identical to the actual incidence of goal scoring, is that goals scored and goals allowed in a game do not appear to be independent, but have a tendency to converge. My model treats them independently, because it would have been much harder otherwise. (I basically broke Excel as it is.)
-So I did not model that, and so I'm getting a few too many results, a few too few draws. (I should be getting ~30% draws, I'm getting ~25%).

That all said, here are my results. Nothing particularly shocking, but I like putting numbers to things.



-I'm not showing things that happened less than 5 times in 10000 runs - there three Tottenham and two Chelsea titles, once Swansea got relegated, and once Stoke made the champions league.
-This system thinks there are non-zero chances of an upset in either the top-4 race or the relegation battle. In 37% of seasons, either at least one club from outside the top 5 made the champions league, or one club from outside the bottom four was relegated.
-Or you could say, two-thirds of the time, the entire middle of the table ended up still safely in the middle of the table.
-Points don't like up exactly with W/D/L because of rounding issues.

There's a site I use for other sports that does a similar-ulation. I don't necessarily endorse their methodology because it is obviously based primarily around goal differential, though Man City is ranked above Man U, so there must be some adjustments. Does Man City have a tougher SOS so far? The GD is at least regressed (to the mean, which is almost certainly wrong for the premier league). The advantage is the same methodology can be applied to pretty much any sport. The coolest bit is seeing how the percentages change after every game, plus the graphs are pretty. Especially this one right now.

Anyway, they have pretty much the same results, the most noteworthy difference being Tottenham vs. Chelsea, I'm assuming because Spurs have better shots ratios but worse goal ratios. Or maybe it's SOS. Here's your and their results side-by-side, in your order, and their rank at the end.



Of course, teams importing the entire population of France in mid-season calls the whole exercise into question.

I can't tell if it's taking the other games into account or not, but if you play with the thing at the bottom of that page, it says Tottenham's estimated 67% chance of being in the top 4 swings to 81% with a win or 50% with a loss against Arsenal.

David Beckham is scheduled to start tomorrow for Paris Saint-Germain against Marseille in the French Cup. The match is available on ESPN3 at 2:55 PM EST.

-This is where the modeling gets moderately poor. If you fudge the normal distribution a little this way and a little that way, you can get a curve that looks a lot like actual goal scoring, so I did that to model goal-scoring. (I can write up the model in more detail if anyone isn't bored already.)

At the risk of boring everyone to death, I would actually be somewhat interested in this <.<

and once Stoke made the champions league.

Man, I hope this is the reality which actually occurs, just to screw with everyone.

Not that the Becks signing is anything like a publicity stunt, but his first start is also being streamed on YouTube, Daily Motion and psg.fr . . .

Not that the Becks signing is anything like a publicity stunt, but his first start is also being streamed on YouTube, Daily Motion and psg.fr . . .

Even The Guardian will be streaming it. Honestly, I do not give a ####. Wake me up if PSG get Zidane out of retirement. Boro and Chelsea or Dortmund and Bayern are better options to me anyway.

Balotelli and Italy just seem like they're such the perfect match ...

AC Milan star and enigmatic striker Mario Balotelli has reportedly commissioned a sculpture of his infamous goal pose from the 2012 European Championships.
Now it seems as though that moment will be captured and preserved in history forever, with Balotelli teaming up with a local Italian artist to sculpt the celebration—muscles and everything—to express intimidation toward fellow opponents.

Per Football Italia:

“I received the commission, but have never actually met Balotelli, so I am working from photographs,” local artist Livio Scarpella told Il Giornale di Brescianewspaper. “I presented various sketches, but Mario wanted to be immortalised in the pose after a goal: muscles in evidence and an expression to challenge the opponents.

“I imagined him as an athlete from ancient times. The statue will be between the classic and pop style, in platinum and coloured bronze with the eyes made of precious stones.”

This guy is getting a statue for free

Suck on that, Balotelli.

I like #1384 because Balotelli is actually enigmatic, whereas it is usually a code word for "is not from here" in north american sports press.

Grantland's football guy sexes up Gareth Bale

-This is where the modeling gets moderately poor. If you fudge the normal distribution a little this way and a little that way, you can get a curve that looks a lot like actual goal scoring, so I did that to model goal-scoring. (I can write up the model in more detail if anyone isn't bored already.)

You shouldn't do that, use the Poisson distribution ?^k*exp(-?)/k!. The ? is the scoring rate (goals per game) and the number of goals you're interested in is k, and it will give you the probability of that many goals being scored in a game (divide by the sum for all k, or more practically the ones before it becomes close to zero, or just plug it into Excel, it's a built-in function there).

Strictly speaking it is not exactly correct as goals can't be scored at arbitrary times (each attack takes some time), but it should be close enough for most purposes, especially if you're not interested in intra-match probabilities.

EDIT: Opera killed my lambdas!!!

EDIT2: Added the division by the total to get a probability :-)

SC - Yeah, Poisson would be better. But to get the goals as an output from a random number, as I was looking through it, I'd need an inverse poisson function, and I'm working in Excel which doesn't have that function. I looked up how to write an inverse Poisson in VB before realizing I didn't want to do that. Treating goals scored as a truncated normal distribution is not the right way to do it, statistically, but it comes pretty close to looking like reality, and it uses an existing function in excel.

EDIT: Reading your post over, it looks like you're saying I wouldn't need an inverse poisson. That'd be good. I'll look at it later.

Arjun - basically, goals scored look a lot like a truncated normal distribution. I already have a spreadsheet with team quality as a percentage of league average in goal scoring and goal prevention, as well as an estimate of home field advantage for GS/GA. So I created an expected goals for / expected goals against number for every remaining game. I then modeled goals scored using an inverted normal variation function - use expected goals as the mean, use the observed SD of the truncated normal distribution, input a random number, find that location on the curve, and round it to a goals scored number. If it spits back a negative goals scored number, try again with a new random number. I also modeled possible variance in team quality by running each simulation with team quality +/- a random amount, up to about 20% of league average. That's not terribly clear - does it make sense?

That's not terribly clear - does it make sense?

Yeah, I think I follow, thanks. What's the formula you used for expected goals for and expected goals against (at the beginning, I mean. I think you posted this somewhere earlier, but I must have missed it)?

EDIT: Opera killed my lambdas!!!

Oh, okay, your post makes much more sense now, I was really thrown by the question marks :)

But to get the goals as an output from a random number, as I was looking through it, I'd need an inverse poisson function, and I'm working in Excel which doesn't have that function.

Use MATCH, just make a column with the cumulative poisson-probabilities for goals from 0 to 10, then MATCH(<your random number>, A1:A11, 1). It returns the position of the last row that was less than your random number (the position in the range, not the spreadsheet), subtract 1 to get goals.

Thanks SC!

What's the formula you used for expected goals for and expected goals against (at the beginning, I mean. I think you posted this somewhere earlier, but I must have missed it)?

For every game, I compare goals scored and expected goals scored (based on shots on target, shots in box, and big chances) to expected goals scored based on opponent quality and home/road. So for each game I get team attack and defense ratings of X percentage better or worse than league average. Add all those up for the season to get projected team quality. For the season projection, I estimated team quality with 2/3 the expected goals based rating and 1/3 the actual goals based rating.

EDIT: dupe.

I hope the Bayern - Dortmund match will be better then yesterday's game. Will anyone besides Chelsea and Boro fans watch that FA cup match?

Now that I think about it, you have to add a 0.0000 in the row before the poisson with 0 goals, because that's the high limit and MATCH needs a lower bound too. I guess it doesn't hurt to add 1.0000 at the end either, for those games that would have gone to 11+ goals.

SC - Yeah, Poisson would be better. But to get the goals as an output from a random number, as I was looking through it, I'd need an inverse poisson function, and I'm working in Excel which doesn't have that function.

In addition, if you have the (free) data analysis plug-in you can generate random numbers from a number of distributions, including the poisson. Although I don't know if the output format is workable for your needs.

Matt: thanks!

Nathan Ake getting the start for Chelsea. In the Sky studio, Ray Wilkins busy explaining how he never wanted Torres at Chelsea.

I think Bayern-Dortmund has become my favorite game to watch from a neutral perspective. They both play such a fun brand of football. And yes, I was drinking PBR and grooming my mustache when I wrote this.

Bayern has control of the game so far. I think collectively Bayern may be faster on the break than Madrid.

Dortmund far too sloppy with their passes so far. Conceding possession far too easily.