League 1 weekend review, 15 Oct 2016

This is my first attempt at a more comprehensive review of a round of matches, as I’ve felt for a while that I could do more to make the various post-match numbers and graphics I chuck out more coherent. In here you’ll find the match timelines, E Ratings and probabilities, plus a few more bits and pieces to string the whole thing together.

Round summary

I’ll kick things off with a new graphic. It’s explained in full here but in a nutshell I’ve dropped each match onto a grid based on:

  1. How surprising the outcome of the match was, based on pre-match expectations (further left = more surprising, further right = less surprising);
  2. How “deserved” the outcome was, based on how likely the balance of chances created during the match would be to produce that result (higher = more deserved, lower = less deserved).

The point of doing this is to compare games by how surprising (or not) they were, both in terms of how the teams have been performing so far and how they performed on the day:

l1-outcomes-2016-10-15

The games clustered in the top right were ones in which the favourite duly delivered and in convincing fashion, looking strong on paper and comfortably out-creating their opponents on the day. Bolton and Sheffield United‘s fixtures were the furthest over to the right and are two of the three highest, so they clearly dominated as expected.

In the bottom right, Scunthorpe were fancied to triumph at home to MK Dons but seem to have made heavy work of it – we’ll see what went on a bit further down.

Even more surprising was the stalemate at Wimbledon, with only a one in five chance of a draw based on events in that match suggesting that either they or Swindon fared far better than the other. Again we’ll see what that was all about in the next section.

If you want to read a bit more about how the E Ratings model generated the probabilities on the horizontal axis, the previews for these games can be found here.

Match-by-match summary

Now let’s look at each match in a bit more detail using the match timelines, which show the combined quality of the chances that each team created and how these increased during the game. They’re explained in more detail here:

2016-10-15-bolton-oldham

The E Ratings flagged this as one of the likeliest home wins and Bolton duly delivered. Oldham have struggled up front this season and this match was no exception, with their hosts carving out far better chances.
2016-10-15-bristol-r-gillingham

Rovers look to have edged a relatively close game here, which was likely to generate goals due to both teams looking vulnerable at the back this season. Gillingham were frustratingly flat for long periods here, not managing many attempts at goal besides an early flurry and a bright second half spell which saw them temporarily take the lead.
2016-10-15-charlton-coventry

This was assessed as likely to be an even game and with 20 minutes to go that’s how it looked. However Charlton found another gear at that point and virtually doubled their total output in an energetic finish, while Coventry were unable to much of genuine menace throughout.
2016-10-15-fleetwood-peterborough

Peterborough usually guarantee goals at both ends due to their combination of potent attack and leaky defence, but only one of those two was in evidence here. Fleetwood stunned their visitors early on and then had by far the better of the second half, putting the game beyond the out-of-sorts Posh’s reach shortly after the hour mark.
2016-10-15-northampton-millwall

This one looked tough to call beforehand as despite their horrible recent results Millwall have continued to perform well. They looked the better side here even before Northampton’s red card early in the second half, yet still had to survive a late scare before redoubling their lead late on and racking up an overdue victory.
2016-10-15-oxford-bradford

The model flagged this as an unlikely source of goals and its blushes were saved when an early Bradford penalty struck the bar. A late Oxford wonder-strike from a free kick rewarded their far more fluid second half performance, with a quiet Bantams’ attack again raising questions about its capability to power a promotion challenge.
2016-10-15-rochdale-bury

A Rochdale penalty was one of few exciting moments in the first half here and they added a second from one of several good second half chances. Bury picked up two red cards late on but the damage had already been done by then and they’ll surely be disappointed with such a flat performance given their energetic attacking displays this season.
2016-10-15-scunthorpe-mk-dons

Despite their lowly league position the E Ratings maintain that MK Dons remain a decent side and they gave Scunthorpe plenty to think about here. The hosts barely threatened for over 50 minutes here and were trailing before Paddy Madden came off the bench to turn the game on its head. This reinforces the data’s assessment of Iron as being clinical rather than dominant, but either way they took another step towards promotion here.
2016-10-15-sheff-utd-port-vale

Another match flagged as a very likely home win by the E Ratings saw Sheffield United dominate from start to finish. Despite sitting fourth in the table prior to this match, Vale were – and still are – fourth lowest in the ratings table and never really got going here.
2016-10-15-southend-chesterfield

After a dull start this looks to have been a relatively comfortable victory for Southend. They look to have been deservedly ahead at half time, with Chesterfield only threatening sporadically throughout and enduring long spells without even a sniff of the Shrimpers’ goal.
2016-10-15-walsall-shrewsbury

Both teams took a while to get going here before each netting from their first shots of the match. An even game then played out, with a red card for Shrewsbury not long after the hour mark ultimately costing them the points when Walsall grabbed a late winner.
2016-10-15-wimbledon-swindon

As was hinted at in the first graphic, a draw is a surprising result when one team creates an extra goal’s worth of chances than the other, although Wimbledon were mostly anonymous before their penalty was saved late in the first half. While out-created overall and perhaps fortunate to leave London with a point, Swindon were no slouches and could even have found the net themselves.

Ratings update

Let’s now look at how this round of matches has influenced the E Ratings table. The ratings are explained in detail here if you’re unfamiliar with them, but put simply they track team strength over time by measuring the quality of chances created and allowed:

l1-ratings-2016-10-15

Bradford remain the top-ranked side, although their defence is contributing far more to this than their attack, which let them down this weekend. Millwall remain in second but there’s plenty of movement below them at the moment.

Rochdale are the biggest risers, having well and truly overcome their dire start. Dale’s performances never looked that bad anyway, so it was no surprise to see them recover.

It’s perhaps a bit controversial to see Scunthorpe drop despite them extending their lead at the top of League 1 with another win, but the E Ratings calculations prioritise performances ahead of results and thus penalised them slightly for their fortunate-looking home win this weekend.

There’s little change at the bottom, with the five lowest-rated sides looking set to remain that way for the time being.

Predicting the rest of the season

While far from terrible at predicting individual matches, the E Ratings are at their best when making longer-term forecasts. After every round of matches I use them to simulate the rest of the season thousands of times to come up with probability forecasts for where each team will finish the campaign:

l1-probabilities-2016-10-15

Bradford‘s strong ratings are still seen as stronger currency than Scunthorpe‘s four-point lead at the top of the table, although as similarly-attired Northampton and Burnley showed last year, it is possible to sustain a promotion challenge with superior shot conversion in the way that the Iron are doing.

Sheffield United look likeliest to challenge the current duopoly, as the only other side with a better than 50:50 chance of finishing in the top six as it stands. Overall there are eight teams with a one-in-three or higher probability of a top six placing – including two newly-promoted teams in Oxford and Bristol Rovers – so it may be a while yet before the promotion race stabilises.

At the bottom things continue to look bleak for Shrewsbury, whose odds of survival remain worse than one in five. There are plenty of other candidates for the drop – as you’d expect in a division with four relegation places – so as for the play-off race it may yet be a while before fates begin to look more certain.

What’s changed?

As the graph above only gives the latest snapshot, I wanted to show what effect the latest round of matches has had on the bigger picture. Below I’ve added a few simple graphics to show how the promotion and relegation contests have changed since the previous round of games.

First of all, let’s look at the automatic promotion race. The filled green bars are the current probabilities (and should match the greenest bars above), with the hollow bars showing how each team’s chances looked after the last previous of games.

l1-promotion-race-2016-10-15

Bradford’s late defeat at Oxford has dented their prospects and the five clubs beneath them – all of whom won this weekend – have benefited. Millwall have edged above Bolton but the chasing pack as a whole has lost ground to Sheffield United after the Blades’ dominant win over Port Vale.

Now let’s look at the relegation battle:

l1-relegation-battle-2016-10-15

Southend’s win over Chesterfield has seen them leapfrog Oldham in the battle for survival and a similar set of results in midweek could be sufficient to see them clamber over the Spireites as well.

Defeat hasn’t made a dire situation much worse for Shrewsbury, but Coventry and Gillingham have both seen their prospects more heavily damaged by their respective reverses this weekend.