24-25 season,  midseason

Mid-season review – Guildford, Manchester, Nottingham, and Sheffield

Concluding the mid-season review posts, let’s take a look at the remaining four teams.

Guildford Flames

Guildford had an inconsistent season last year, ending with a poor 44% win ratio, which given that they challenged for the title the previous season was surprising. This year they’re not back to that 22-23 form, but they have improved on 23-24 with a win percentage of 55.56% for the first half of the season. One might expect an improvement across all stats, but really the only one of note is Goals For, where they are scoring 3.3 goals per game vs 2.96. Elsewhere the stats are similar if not slightly worse, for example they are conceding roughly the same (3.19 vs 3.06), having fewer shots (50.22 vs 52.72), allowing more shots (51.26 vs 46.70) and have similar save percentages and special team figures.

So, in some respects they’re not doing much different to last year, but they are doing it slightly more effectively. I did a correlation for each team on a number of metrics against goals scored or goals conceded. For example, how much of a correlation is there between the number of shots on target and the number of goals scored for each team? Quite a lot you’d think, but actually it varies wildly. I’ve plotted this one below, and as we’ll discuss later, I have no fucking idea what’s going on at Sheffield. If you take the averages across the league and correlate the shots on target with goals scored, then there is a reasonably high correlation of 0.71. But if you do that for the games for each team then individually they do not show such a high correlation. So averages correlate well but not individual scores.

We’ll come on to Sheffield later, but for most teams there is a mild correlation in the 0.3 to 0.4 range. So generally more shots on target, more goals, but not to a great extent. Glasgow are the standouts here. To return to Guildford then, their correlations are pretty much like this across the board, which is not a surprise as nearly all their stats sit at around the league average. Which is to say, they are a solid mid-table team. I would be concerned though if I were a Guildford fan that some of the issues from last season remain bubbling away ready to resurface.

Manchester Storm

Storm are the flipside to Guildford in a way, doing worse than last season with a win percentage of 37% vs 51%. They’re scoring more (3.11 vs 2.63) but conceding more also (4.11 vs 2.85). So whereas their average goal differential last season was -0.22, this season it is -1. Their PP was rubbish last year at around 16% and it’s rubbish this year at 14%.

Last season they had the lowest shot conversion rate, this season they operate at around the league average, taking 10.77 shots on target to score (the average is 10.32). However, as the goals conceded stat suggests, they are generous in front of their own net, with teams requiring only 8.62 on target shots to score (league average is 10.19). It’s not that they are giving up lots of chances, their high and medium chances against are lower than the league average, but they are allowing these to become goals at a higher rate. Their Goals Saved Against Expected average is -1.78, with only Fife being worse. Too many shots that shouldn’t be goals are ending up in the back of the net.

What Storm and Guildford illustrate I think is that small differences in the stats can have large differences in actual games won. I can’t see anything in the data to suggest that Manchester or Guildford will have a markedly different second half of the season to the first.

Nottingham Panthers

As I mentioned at the start of the season, the tragic events of last year meant that we should view the data from that season as a blip, and we should expect Nottingham to have a much improved year. After a wobbly start, that has proven to be the case, and their win percentage now stands at 59.26% compared with last year’s 38.89. There’s not much point in doing a comparison with last season on any metric – all of them are way better. They sit comfortably amongst the usual main players of Sheffield, Belfast and the Devils in terms of offensive stats, and have the best defensive stats apart from Cardiff conceding only 2.37 goals on average, and an Expected Goals Against average of 2.17 (the lowest in the league). A team reborn.

Sheffield Steelers

Last year the Steelers were all conquering on the stats, they topped every metric. This year the competition is tougher so they are not as dominant. They top Goals For per game, averaging 3.93, which is a touch down on last year’s 4.22. But the increased competition is evidenced by other team pipping them in other metrics, for instance Nottingham, Cardiff and Belfast all concede fewer goals than the Steelers’ 2.56 (which is still pretty impressive), Belfast have more shots on target (39.41 vs 33.89), Cardiff have a better Save %age (93.41 vs 91.33) and Nottingham a more miserly XGA of 2.17 vs 2.28. However, although Sheffield don’t dominate these stats as they did last year, they are strong in all of them, and could be leaders by season’s end.

At the start of the season I couldn’t really find any areas to suggest improvement and that their difficulty would be in maintaining their form. I also suggested that Greenfield had an excellent season but netminder’s averages often tend to regress to the norm, so they could do with cover there. When Greenfield was out for a few games, this lack of depth was exposed, but now that he’s back, his save %age is back up at a healthy .934, so his form doesn’t seem an issue. With the CHL backlog out of the way, I expect the Steelers to have a slightly stronger second half of the season.

They have given us some unusual results, and this is born out in a deeper dive into their analytics. They can lose away to Fife (one of only two victories for the Flyers) and then pound Guildford 9 – 0. I mentioned above that I had looked at correlations of different metrics such as overall shots, shots on target, high chance shots, etc for each team to goals scored. While the average of these across the league correlates well, when you do the correlation at a game by game basis the fluctuation smoothes out a lot of that correlation. But even with this caveat, the Steelers are just fucking odd, their correlations look like they were cooked up by a racoon on meth. Let’s look at overall number of shots for, shots on target, high chance shots and medium chance shots. You’d think there’d be some positive, if fairly low correlation with the number of goals scored, right? Not in the wacky world of Sheffield, take a look at these correlations:

Both number of shots and shots on target are negatively correlated with actual number of goals. So the fewer shots they take, the more they score. You can see this fluctuation in games – one week they take 60 shots and score 8 and the next they take 75 and score 2. This is why averages are more effective for this kind of correlation, but over 27 games you’d expect some sort of positive pattern. Other teams have weak correlations but no-one has this kind of pattern.

I’m not sure what it means though. Perhaps it indicates that Sheffield have not found a consistent style of play in the first half, but they have good players, so stuff happens. Their recent run of 7 victories also look to have a more sane pattern, so that may suggest they have found their groove now. Based on last year this could be ominous for the rest of the league if so.

(Image courtesy Sheffield Steelers)

Leave a Reply

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