Man United 2012/13, TSR and other statistics

As a culture that celebrates the individual, we inevitably seek out marks of personal excellence. Some games, like baseball and cricket (this is one Yank who loves cricket), are aggregations of individual acts, and thus are ripe for analysis through individual statistics. But football is the ultimate team game, and it’s in team stats where football numbers really come into their own. As we’ve noted, perhaps too often, individual stats need context; the team numbers are the context itself. Go back and take a look at the earlier articles in this series, and you’ll see that to understand an individual’s performance we’ve repeatedly referred to team tactics, which find themselves embodied in team numbers.

So in a sense we’ve done much of the work of this section already. We know that team pass completion percentages reflect the kinds of passes players are asked to make; we’ve seen that team interceptions depend largely on a side’s defensive philosophy; we can figure out that team possession percentages are a product of both style and competence.

The purpose, then, of the next couple of articles is to look at which team stats tell us most about team performance. Because although it’s nice to know that your team is third in the league in interceptions, what really matters is how well your team is doing.

 

Total Shots Ratio (TSR) and Shots on Target Ratio (SoTR)
Among the basic stats, our two best measures of team performance are shots and shots on target. Because the point of the game is to score goals while preventing your opponent from scoring, shooting stats will naturally tell us a lot about how well a team is doing. The numbers to look for are 1) Total Shots Ratio (TSR), which is the ratio of your total shots to your opponents’ total shots; 2) Shots on Target Ratio (SoTR), which is the ratio of your shots on target to your opponents’ shots on target. It’s been shown that these numbers consistently track team success, although not by any means definitively for each season. Over 50% is a positive, under 50% is a negative.

Here are the current numbers for TSR:

Liverpool 69.0
Man City 66.4
Tottenham 65.2
Man United 64.2
Chelsea 63.3
Arsenal 59.9
Southampton 59.3
Everton 53.0
West Ham 50.7
Bournemouth 45.8
Crystal Palace 45.6
Watford 44.0
Swansea 43.8
Stoke 43.6
Leicester 43.5
West Brom 42.2
Middlesbrough 39.2
Hull 36.9
Sunderland 35.2
Burnley 33.4

And here for SoTR:

Tottenham 68.2
Chelsea 67.6
Liverpool 67.0
Man City 64.0
Southampton 62.4
Everton 58.1
Arsenal 58.0
Man United 56.1
Bournemouth 48.2
Watford 45.2
Swansea 44.6
Crystal Palace 43.6
Stoke 42.9
Leicester 41.6
West Ham 41.0
West Brom 39.4
Hull 37.1
Middlesbrough 37.1
Burnley 35.0
Sunderland 31.2

You’ll see that for the most part the better teams are at the top, the middling teams in the middle, and the strugglers at the bottom. But you probably noted some discrepancies: Burnley and West Brom are significantly lower in TSR and SoTR than they are in the table; Crystal Palace, Swansea and Southampton are significantly higher. Let’s take Burnley and Southampton as our examples. In TSR and SoTR, the Clarets are very close to the bottom, and the Saints are right around the top six – yet the league table shows them only a point apart in 11th-13th place. Does this mean the numbers have it wrong?

No – it means that Burnley’s results are better than their performance and Southampton’s are worse. We know this happens in individual games: you play poorly and win, or play well and lose. But over time we assume it all evens out, so we say ‘the table never lies’. But sometimes it does. Since we know that over time TSR and SoTR track success, we know that teams with Burnley’s numbers will usually be fighting relegation and teams with Southampton’s numbers will be competing for the Europa League.

This has two simple but vital ramifications: 1) teams whose results are significantly better than their TSR and SoTR will tend to drop down the table; 2) teams whose results are worse than their TSR and SoTR will tend to rise.

The textbook example is Swansea City. In 2014/15, under new manager Garry Monk, they finished eighth, with an excellent total of 56 points. But their TSR pegged them in 17th place, their SoTR in 15th. In other words, their results greatly over-represented their performance. They weren’t an eighth-place team, they were a team in close to relegation form. Based on results, naturally, Monk was given a new contract in the summer. So what happened next season? They quickly found themselves in a relegation battle, because that was roughly the level at which they’d been performing all along. Monk was sacked, and a succession of managers have been unable to raise the performance level significantly.

Now ask yourself: which side has dropped the farthest in the table this year? Leicester City, of course. And Leicester were the side whose results last year most over-represented their performance. Leicester were ninth in TSR, fifth in SoTR. Their title win was a statistical anomaly. This year they haven’t been anywhere close to those figures; in fact, their place in the table right now is just about equal to their current TSR and SoTR.

And speaking of statistical anomalies, surely you remember that other recent long-shot winner, the Europa League-level side that won the title? No? Well, it was Manchester United in 2012/13, Sir Alex Ferguson’s last year. United were seventh in TSR, sixth in SoTR, yet, like Leicester, they won the title with ease. We all assumed it was a typical United title team. It wasn’t at all — it was about as great a statistical upset as Leicester. Which means, of course, they were due to drop the next year. In 2013/14, their TSR was almost the same as the previous year, their SoTR a little lower. So wee Davey, if you’re listening, it wasn’t all your fault. (Not that United fans will ever be convinced.)

What about the other way – teams whose results weren’t as good as their performance? The best recent example is Liverpool last season. They finished eighth, but were second in TSR, third in SoTR. So you’d have expected them to climb the table the next year, and they most certainly have. Their TSR and SoTR rankings are about the same, but this time they accurately reflect performance.

This tells us that TSR and SoTR may be rough measures, but they’re reliable. It turns out that 38 games, while in most cases sufficient for results to match performance, won’t always tell the tale exactly. So occasionally, the table will lie. That’s a good thing, because we need more Leicesters and more Ferguson farewells. At the same time, you need to look at the underlying numbers to get the true performance level. Of course, if your team is doing much better than expected, you might not want to look at TSR or SoTR. But eventually the numbers win out. Unless…

Unless you’re Tony Pulis. Because the steelworker’s son from Pillgwenlli is the man who broke the bank at Monte Carlo. In every single full Premier League season he’s managed, his sides have finished higher in the table than their TSR and SoTR would indicate. Last year, West Brom were 20th in TSR and 19th in SoTR, yet never in danger of relegation. We all know about Pulisball: defend deep, use the long ball, score on set-pieces. Depending on how you look at it, he’s either exposed the flaw in the numbers or got more out of his teams than any other Premier League manager. That doesn’t mean he’d win the league with Chelsea, but if you’re a bottom-half team and want to laugh at fate, call for the cap and trainers.

One more thing about TSR and SoTR: a significant discrepancy in your numbers may indicate a particular source of success or failure. West Ham have a solid 50% TSR, but a poor 41.0% SoTR; Manchester United have a very strong 64.2% TSR, a less impressive 56.1% SoTR .That means these sides aren’t getting enough of their shots on target and/or their opponents are getting more than their share. This may be an issue of shot quality – are the opponents getting shots from significantly better places? I’ll leave it to you to find the teams whose SoTR is stronger than their TSR, and who may therefore have superior shot quality. We’ll revisit this idea when we get to Expected Goals.
Conversion Rates
Another way to understand your team’s performance is conversion rate, the percentage of your shots and/or your opponents’ shots that go in. Since conversion rates are obviously linked to shots and shots on target, they’ll provide significant information in this regard. They’ll also tell you, at least in part, where a problem may lie.

Remember what we said about conversion rates earlier in this series: among individuals, they tend to vary widely from year to year. The variance isn’t as wide for entire teams, but it doesn’t need to be. Last year Premier League teams averaged 484 shots for the season, so a mere two-point conversion rate change means on average of around 10 goals, and 10 goals can make a big difference in where you finish.

Let’s go back to last season and Leicester City. The numbers tell us that the Foxes led the league in conversion rate, at 13.0%. This isn’t a particularly high rate, historically speaking: Liverpool in their Suarez season were up at 15.5%. But their main rivals for the title, Arsenal and Tottenham, were down at 11.4% and 10.4% respectively. (Spurs under Pochettino are a bit of a special case, because they tend to shoot on sight.) Remember that a two-point swing averages nearly 10 goals. So their conversion rate played a significant role in their title.

And that last United title team, the other big anomaly? They had an abnormally high conversion rate, 16.8%, which has only been exceeded once in the 11 full seasons these stats have been available. The record of 17.1% is held by 2007/8 Aston Villa, a side that actually took fewer shots than their opponents but finished sixth.

But, as noted, Leicester’s 13.0%, while quite good, wasn’t exceptional. What was exceptional was their opponents’ conversion rate, 7.0%, the fifth-lowest in the 11-year stat period. Arsenal and Tottenham were a bit closer here, at 8.0% and 8.3% respectively. But in the second half of the season Leicester were almost impenetrable, and combined with their attacking conversion rate it was enough to push the Foxes to the top, despite their moderate TSR and SoTR.

Let’s look at Burnley again, one of this year’s overachievers. Their conversion rate is 11.4%, just about the league average. But the conversion rate of their opponents is a quite low 7.9%. The Clarets sit back, defend sturdily, and have Tom Heaton in goal.

What about Southampton, our sample underachievers? Southampton’s opponents are converting a roughly average 11.4% of their shots. Actually it’s more complex than that; we’ll look at this stat again in later articles. But what’s quite obvious is their struggle at the other end. They’ve converted an extremely low 6.6% of their shots, which if it lasts would be the fifth-lowest recorded in the 11-year stat period.

The good news: there’s still nearly half a season remaining, and if your numbers are at historic lows, the odds are they’ll rise. It may be already starting. Before last weekend’s game, Southampton’s conversion rate was down at 6.1%. But against Leicester they scored with two of their 19 shots on target, the penalty excluded.

Conversion rates, too, bring up the question of shot quality. If your conversion rate is low, maybe the places you’re shooting from aren’t the best. And so we move closer and closer to xG. Next week a piece looking at shots from a different angle, and after that we’ll dive in.

Peter Goldstein