Stats More Like It: Sheffield Wednesday; Championship challenge accepted

Stats More Like It: Sheffield Wednesday; Championship challenge accepted

After too long an absence, Stats More Like It is back to take a good, thorough look at how our Wednesday’s doing according to the numbers.

Last season we upset the odds and statistical wisdom to win promotion. What do the stats say about how we’re doing this season?

This season, I’ll first look at what number of points we need to win before the end of the season, based on statistics of the last 20 seasons of Championship/Division II* football, which I recently  posted excerpts from on Twitter (1, 2, 3, 4, 5; *=I’m aware it wasn’t called Championship until the 2004-05 season, but for ease of use I’ll be using Championship throughout the article, even when refering to before 2004-05).

Then I’ll look at the same sort of graphs used last season in League 1 to see how we’re doing compared to the division’s other teams on a range of different parameters. I’m relying on the stats supplied by ESPN Soccernet, as these seem to be the best sort of stats available at Championship level. OPTA might introduce their far more comprehensive set of statistics for the Championship next season (according to this tweet), though, another reason I’m hoping we survive!  Until then, though, Soccernet’s stats it is. If by any chance you know of anywhere carrying better/more statistics on the Championship, please let me know.

Badge Owl POST

Last 20 seasons of Championship

Using Statto.com’s brilliant custom table generator I’ve managed to find the tables at February 9 of all the past 20 seasons of the Championship as well as the end of the season tables.

The results are interesting:

  • In only 2 of the 20 seasons have the team in 21st had as many points per game as Barnsley (21st) do this season (1.10). So it’s no wonder if we all have the sense that it’s a tight relegation battle.
  • In 18 of the 20 seasons the points per game of the team in 21st had increased ever so slightly from February 9 to  the end of the season. The difference isn’t big (the improvement in form amounts to an additional point in the points total at the end of the season), but it suggests that the current points per game of the team in 21st isn’t likely to deteriorate towards the end of the season.
  • Barnsley’s 1.1 points per game is 50 points in a full season, so the safety target this season is likely to be 51 or 52 points, depending on whether we add in the additional point of the average improvement in the points per game of the team in 21st have seen historically.
  • 51
  • Only twice have 52 points been needed to beat the team in 22nd: 1995-96 and 2007-08. In the latter, Wednesday would’ve been relegated but for an emphatic 4-1 win against Norwich on the last day of the season, ultimately finishing as high as 16th. Fingers crossed we won’t have to go through something like that again!
  • On average, only 1 in 3 of the teams in a relegation spot on February 9 managed to secure survival at the end of the season. Only once (1997-98) did all 3 teams in a relegation spot at this point manage to save themselves.
  • Despite a lot of talk about “making a push for the play-offs” from a host of teams even deep into mid-table, historically only 16 % – 3 in every 20 – of teams outside the play-offs on February 9 managed to finish 6th or higher at the end of the season. On average, only 1 team manage that feat.
  • We aren’t likely to be it: Even if we kept up our promotion-winning form of the last 10 matches (21 points) until the end of the season, winning  67 points, it still wouldn’t be enough: The lowest a team has won 6th by was 70 points (Watford, 2007-08).
  • While achievements like Reading’s last season – winning the Championship from a relegation position early in the season, and still only 8th on February 9 – will inspire and be remembered, they are very rare: Other than Reading, only Sunderland (2006-07) have managed to win promotion after being outside the play-offs on February 9. Once every 10 years isn’t a trend.
  • Incidentally, no one has “done a reverse Reading” and dropped from a promotion spot to outside the play-offs in the past 20 seasons, and 73 % – 3 in every 4 – of teams in top 2 now went on to win promotion.
  • In general, there isn’t as much movement in the table from now until the end of the season, as tales like Reading’s and Sunderland’s might suggest: The highest jump in the table was actually our dear neighbours, who in a crazy 1995-96 season climbed 14 steps to finish in 9th (Millwall went the other way that season: From 11th in February to relegation as 22nd).
  • On average, teams generally move 2 positions from now until the end of the season.

In short: It’ll take a bit more than the usual to survive this season, which we should be odds on to do, but forget about the play-offs.

Badge Owl POST

2012-13 Championship

If, like me, you’re into football statistics, you can do a lot worse than following Ben Mayhew on Twitter (@experimental361) or check his website at www.experimental361.com. In fact he’s probably covering  the Championship a lot better than I can in the following, using what I would consider more reliable data (scraping Press Association statistics from matches), and doing cool graphics like these with them. He also uses his stats to predict matches, with varying degrees of success, but this past Saturday he predicted we had a big advantage from headers and going forward on our right – only for Llera to score a headed goal from a right-wing cross and before that we netted a corner from the right.

Anyway, what are Soccernet’s numbers showing us?

Shots and goals

(disclaimer: Dale Johnson from ESPN Soccernet have explained that, for some odd reason, their statistics don’t count goals as a shot on goal, which I was not aware of at the time of writing this article. It shouldn’t drastically alter the numbers, nor my conclusions – and a quick re-calculation shows that to not be the case – but I’ll revise the shots on goal statistic in the next stats article I do)

First up we have the number of shots on goal – that is, on target – per match. The statistic gives you an idea about how many chances have been created. After the team name, in brackets, is the current league position of the team. A team ranked higher in this statistic than in the league table can be said to have something of an unfulfilled potential.

1112-feb11_shots-on-goal-per-matchWe’re 8th in that statistic, suggesting chance creation hasn’t been an issue for us, but having scored just 1.22 goals a match, with only 4 teams in the division worse, you can patently see the lack of a seasoned goalscorer we’ve all been talking about most of the season. Dave Jones knows that only too well: Over the course of the season he’s brought in 4 different Premier League strikers. While neither Bothroyd nor Sidibe worked out, Leroy Lita and Connor Wickham are two very promising loan signings indeed. Other than those two, 6 recognised strikers have played for Wednesday this season, scoring a total of just 9 goals. 2 defenders – Miguel Llera and Reda Johnson – have scored more than that (10 goals), saying everything about Wednesday’s troubles up front.

Notice also that Bolton – 1st – and Blackpool – 3rd – have plenty of shots on goal, but sitting in 16th and 14th in the league table respectively they, like Wednesday for large parts of the season, are not turning those chances into points. The most striking feature is perhaps that there isn’t greater variation among the teams in this statistic – in shots taken (not just on goal/on target) – there’s a lot more variation in the Premier League than in the Football League. Another indication the Championship is a very tight division, where small differences in performance can potentially reap great rewards.

1112-feb11_shots-on-goal-shots-percentage

 

Shots on goal as a percentage of total shots taken tells us something about the accuracy of shooting. I don’t have Premier League numbers to compare to, but again, the striking thing is how closely together the values are for the teams in the Championship (from Huddersfield’s 39 % to Leicester’s 54 %). It’s not an easy measure to interpret, as there are a couple of possible interpretations: 1) A high percentage means the shots taken by a team are in better positions, and thus more likely to be on target, 2) A high percentage could also mean shots are taken from all sorts of positions, but that the team’s players have good shooting technique and still manage to hit the ball on target (or, as a sort of Jermaine Johnson Law, maybe a lot of the shots they do manage to get on target are tame efforts going straight into the hands of the goalkeeper).

1112-feb11_goals-shots-on-goal-percentage

If we want to get a good sense of whether a team are being wasteful or not, and perhaps if they are over- or underachieving, it makes more sense to look at how many shots on goal they need for every goal they score. As you can see from the graph above, every 4th shot of Wednesday’s has been a goal. Compared to the rest of the division, we’ve needed a lot of shots on goal to score: Only Ipswich, Bolton and – perhaps surprisingly – Hull have needed more shots to score than we have.

As I said, having a low score in this statistic suggests the team is underachieving. On the other hand, a team with a low score could be overachieving. It makes sense with Crystal Palace (44 %), who have scored from nearly every other shot on goal, and who have dropped from 1st in early December to 5th two months later. If Wednesday had Palace’s efficiency in front of goal, we’d have scored 64 goals – more than any other team in the division.

To put our score – 26 % – into perspective, we had 41 % last season in League 1 and when we finished 15th in League 1 39 % (the average of League 1 in 10-11 and 11-12 and this year’s Championship is near-identical, so it’s not because the Championship as a whole is worse or better than League 1).

Overall, then, the numbers seem to confirm what we’ve all been saying most of the season: If we had a proper goalscorer, we’d be doing a lot better.

Badge Owl POST

Saves and goals conceded

To paraphrase a golf-saying: Attacking might be funny, but defending is money. So let’s look at how we’re doing defensively. First up is a measure that’s designed to look at how “frail” we are defensively: How many times have other teams either scored or forced a save from a team’s ‘keeper?

1112-feb11_frailty-saves-made-goals-conceded-per-match

Considering our position, we’re doing surprisingly well, being 4th, and with the highest ranked team outside top 10 in the league table other than us being Huddersfield in 9th. There’s quite a bit of difference between best and worst in this statistic: Peterborough (8) are breached nearly twice as many times as Leicester (4.5). And where you are in this statistic  fits quite well where you are in the league table, so maybe there’s something to that botched up golf saying from before?

This next statistic might on the surface be overly complex and perhaps a bit weird, but bear with me: It shows goals conceded by a team as a percentage of the saves the ‘keepers of that team has made. The lower the score the better: 25 % means the team makes 4 saves for every goal conceded, whereas 50 % means they only make 2 saves for every goal conceded.

1112-feb11_goals-conceded-saves-made-percentage How are we doing then? Well, not that well – only Burnley and Huddersfield are worse than us. As we saw before, we’ve been doing very well in restricting chances – and a lot better than being 19th in the league table would suggest – but we have a weakness here: When opponents do breach our defence, Chris Kirkland only makes 5 saves for every 2 goals he’s conceded. This will raise an eyebrow or two, seeing as there’s general agreement that Kirkland has done very well for us, and he’s certainly made a lot of high-profile close range or one-on-one saves.

Before too many fingers are pointed Kirkland’s way, it’s worth to remember how we fared in this statistic in the past 2 seasons: In 10-11, we were at 39 % – only 2 were worse than us – and in winning promotion last year, only 3 teams were worse than our 37 %. So have we had bad ‘keepers these past 3 seasons? Well, that is one way of looking at it, and to those still clamouring for Lee Grant as no. 1, his score in 2007-08 was 16 %, 27 % in 2008-09 and 31 % when we were relegated in 2009-10. But if there are systematic problems in defence – say someone’s slow in perception, positions himself badly, charges too rashly – that could also be an explanation. It pains me I can’t back this up with a statistic, but certainly early on this season, we seemed to open up our defence like the Red Sea for Moses on a regular basis; something the amount of one-on-one saves Kirkland’s made also hints at.

Last season, I speculated that “The Bywater Effect” mightn’t be about saves made, but chances restricted, and the same could be true for Kirkland this season. Had Kirkland’s score been 30 % rather than 40 %, we’d have conceded 12 fewer goals than we have. Having lost 8 matches by a single goal, it would’ve made a difference.

As a bit of fun, if we combine saves made, goals conceded and shots taken we get a crude measure of how action-packed a team’s match is. There’s not a lot of variation, although Warnock’s ‘lovely’ Leeds side have 12.8 events per match compared to Hull’s 17.3:

1112-feb11_action-saves-made-goals-conceded-shots-per-match

Fouls

Part of Gary Megson’s legacy at Wednesday is a team of combative players: Wednesday committed the most fouls of any team in League 1 last season (we were only 12th in that statistic in 10-11). It’s interesting to see if that change in style persists during Dave Jones’ first full season in charge:

1112-feb11_fouls-committed

Short answer: Yes. Only Bolton commit more fouls than we do. “Dirty Leeds”, incidentally, are 4th (and we also have a worse disciplinary record than they do).

A season of two halves

The above statistics are interesting enough seen in isolation, but borrowing the old football cliché, our’s has in many ways been a season of two halves: Before and after Barnsley away (or, as some have put it, before and after Stuart Gray (who joined as a coach on December 19)).

In our first 21 matches in the Championship, we picked up just 15 points, and were three points (and a worse goal difference) adrift of safety (and 8 points adrift of 20th). In our last 10 matches, we’ve picked up a whopping 21 points (and we’re just 8 points of 8th) – that sort of form would’ve seen us safely in 2nd had it been sustained for the entire season. In many ways, the two parts of the season couldn’t be more different. But if that leads us to think the season has changed completely, we’re in for a surprise:

1112-feb11_first-21-vs-last-10

What is extraordinary here is that there’s very little actual change in all but two measures: Goals conceded and points won. In fact we actually have fewer shots and shots on target in the last 10 than in the first 21 (but we score a bit more often). It’s rare that there’s one factor that stands out like that. Again it fits with what a lot have observed about us on the pitch: That we seem more solid at the back. Not just conceding less, which we evidently do, but not allowing the opposition to get into as many dangerous positions as before. As clear evidence for the “Stuart Gray factor” as I’ve seen.

Players individually

Moving on from looking at the stats for the 24 Championship teams, it’s just as interesting to look at the individual players. Thankfully the data I collect from Soccernet allow that (brief explanation: I use Excel to automatically retrieve the “Player stats” pages of every Championship team, and then add together the various statistics for the players to get an aggregated number for each team and then play around with those).

I sadly do not have minutes played for each player (if anyone knows of a website showing these for Championship players, do let me know). But I have number of games started and number of subs appearances, and I’ve combined the two. I’ve weighted subs appearances as 30 % of a game started (most substitutes seem to be introduced around the 60th-65th minute). It’s far from an ideal way to do it, but in the circumstances I’d say it’s the best I could do.

Badge Owl POST

Goalkeepers

Glove Face Free at Last

Free at last!

Looking at ‘keepers, and that goals conceded as a percentage of saves made statistic discussed before, who is the fairest in the land, or at least in the Championship?

Well, none other than “Glove Face” aka. Hull’s Eldin Jakupovic: He’s made 30 saves in 5 matches played, but other than the 2 goals against Wednesday, he’s only conceded 1 in the other 4, meaning his percentage is a mere 10 %.

At the under end of the spectrum, Huddersfield’s veteran Ian Bennett has made just one substitute appearance this season, but conceded three goals without making a single save.

If we look at ‘keepers with at least 10 starts, none other than the aforementioned Lee Grant comes out worst of the lot: 43 %, closely followed by Huddersfield’s apparently highly rated Alex Smithies (42 %), and…Chris Kirkland (40 %).

Best of the lot is Barnsley’s Ben Alnwick (21 %), with Brighton’s Kuszczak (22 %) and Charlton’s Hamer (23 %) close behind. From these stats Hamer appears to be the division’s most effective goalkeeper, as he makes more saves a match (6.2) than all but one of the ‘keepers with at least 10 league starts.

Outfield players

As I’ve mentioned, OPTA sadly do not do stats for the Championship, so there aren’t any good defensive measures for individual players like tackles made, interceptions made etc. The closest, which is far, is fouls committed.

Reda arms up

Take no prisoners

In Wednesday’s squad, among those who’ve started at least 5 matches, you’d never guess who’s committed the most fouls per match. It’s Rodri (1.91).

It’s less surprising who’s 2nd: The lovable Reda Johnson (1.8), whose defending can have an oafish quality to it.

Among all Championship players with at least 5 starts, Reda Johnson is booked every other game, more than any other player (Michael Brown is 5th in that list).

Semedo’s actually just 6th in this list, despite having been the undisputed “foul king” last season in League 1. Perhaps an indication that he hasn’t managed to get as close to his opponents in the Championship as he did in League 1?

It’s also no surprise that Michail Antonio, surely the Wednesday player most often challenging his direct opponent with dribbles, suffers the most fouls per match (1.58). Madine – so often going down easily looking for fouls – suffers just 1.44 fouls per match, suggesting that, in addition to being annoying for us as fans to watch, that sort of playing style hasn’t been very effective for him.

In the entire division, among players with at least 5 starts, the most surprising thing is Leeds’ Michael Brown is only 10th (2.07) in the fouls committed statistic (though Bechio (2.13) is 7th), whereas Blackburn’s Ruben Rochina gets fouled more than three times a match on average.

Antonio scores

Trying to help

Attacking-wise, Michail Antonio’s 5th in the division’s assists table with 8 (Chris Eagles (12) and Thomas Ince (10) are 1st and 2nd), and in many ways, in preparing goals, we’re a one man team: Jermaine Johnson, with 3 assists, is the next Wednesday player on that list.

If you combine assists and goals scored, you get a sense of a player’s worth to the team going forward. Thomas Ince (27), Glenn Murray (26) and Matej Vydra (24) – they’ve all been involved in around half of all the goals scored by their respective teams. Our MVP is most definitely Antonio, who leads both our scoring and assisting chart: 14 goals and assists combined, meaning he’s been involved in 2 of every 5 league goals we’ve scored this season. Jermaine Johnson and he have been involved in half of the goals we’ve scored, and have 40 % of our shots and shots on target. So we’re very reliant on especially Antonio, whose goals per game rate is about average.

Speaking of goals per game, three extraordinary super subs lead the way: Jason Scotland, Kevin Phillips and Frazier Campbell have scored 5 goals between them from a combined 7 substitute appearances.

Leroy Lita (60 %; 38 % at Birmingham) and Reda Johnson (56 %) are among the division’s most effective in percentage of shots that end up as goals, though if we just look at players with at least 20 shots to their name, Vydra (66 %), Becchio (63 %), Murray (51 %) and Jordan Rhodes (40 %) have some intimidating strike rates, especially considering the number of games they’ve played.

COG a1

Doing well

Among those with 20 shots or more, Derby’s Will Hughes (70 %) shares top spot in shots on target as a percentage of shots taken with an interesting name: Jay Bothroyd. In comparison, Antonio’s 47 % is close to the division’s average, beaten – perhaps surprisingly – by Jermaine Johnson’s 50 %. Llera makes up for all his wayward free-kicks with those penalty box salmon leaps, and has 52 %, while Chris O’Grady’s doing well with 55 %.

If we look at shots per match, it’ll come as no surprise that Jermaine Johnson’s our highest ranked player in 20th with 2.33 (Chris Eagles (3.79), Ruben Rochina (3.51) and Craig Davies (3.01) lead the way. Among Wednesday’s players with fewer than 20 shots, Lita (2.17) proves he’s not afraid to have a go.

If anyone feels like looking at the stats themselves, I’ve arranged the data I’ve used for the player analysis in this Google Spreadsheet (click on the arrows in the column headings to sort the data): https://docs.google.com/spreadsheet/ccc?key=0AlreCxmmHxr_dEVFTlBZSW5DZHphZjdtM0NNY2pMalE&usp=sharing

I’d like to thank you for having read this far, and I hope you’ve found at least bits of it interesting. As always, if you know of any good sources for data, have seen statistical analysis of football you’ve found interesting, or have any comments or suggestions, do not hesitate to let me know.

Badge Owl POST
Peter

Owls Alive
TWITTER: @OwlsAlive or @ploehmann


Leave a Comment

Your email address will not be published. Required fields are marked *