24-25 season,  league

Shot creation

Each gamesheet the EIHL provides gives us a tally of the High, Medium and Low danger shots, based on where those shots took place. Obviously you want to create lots of high danger shots and not give up many. A high danger shot would be something like a player in the slot with a clear sight of goal, an odd man rush, etc. So, let’s take a look at how meaningful these figures are.

To take an example of when they are not very meaningful, look at Cardiff’s 6-0 away win at Manchester. They must have created a lot of high danger chances, right? Nope, zero recorded. Okay then, they must have been bombarding the net with medium danger shots? Again, hard no – zero medium shots. They scored 6 goals from 34 low danger shots. I didn’t see the game but can only assume Manchester were picking the puck up and depositing it in their own net.

To take an example at the other end of the predictive scale, we can look at Sheffield’s 5-2 win over Coventry. Here Sheffield had 7 high danger shots to Coventry’s 2, and they scored on 3 of those (Blaze converted one of their high chances). The number of high chances here determined the outcome of the game.

Let’s see how teams are faring so far this season on High, Medium and Low chances created in the following stacked graph:

Belfast have the highest average of High Danger shots per game (4.46), followed, perhaps surprisingly by Nottingham (4.27). For the Panthers this is potentially good news – they are creating chances and that might indicate that their results are not quite a fair reflection of their play. If so, this may even out and their win ratio will increase.

But of course it’s not just about scoring goals, it’s about keeping them out. So, let’s look at the High, Medium and Low danger shots against each team (the lower the better here, obviously).

The lowest team in terms of giving up High Danger shots is (again, perhaps surprisingly), Dundee. Belfast do pretty well on this metric also however, conceding only 2.29 high chances on average per game, with the league average being 2.98.

But, as our first example shows, you can win without High Danger shots. We can look at some correlations around these types of chances, and the results are not quite what I expected.

ItemsPearson Coefficient
Win %age vs High For0.699
Win %age vs Med For0.725
Win %age vs High Against-0.261
Win %age vs Med Against-0.527
Av Goals per game vs High For0.803
Av Goals per game vs Med For0.874
Av Goals conceded vs High Against0.389
Av Goals conceded vs Med Against0.752
Av Goals vs Low For0.03
Correlations using High and Medium danger chances

What’s interesting in this table is that Medium danger chances are more highly correlated with outcomes than high danger ones. So, they are a stronger predictor of overall win percentage, and average goals scored per game than the high chances. This is likely a factor of there being more of them – 3.01 High Chances vs 14.48 Medium on average across the league. But it’s not just the size of the number as the final entry shows – Low chances are, as we would expect, very poorly correlated against the actual number of goals scored. These are the rubbish chances that often aren’t even intended as a shot. It’s even more of a marked difference when we look at goals conceded against the chances given up. The correlation for High danger chances against goals conceded is only 0.389, compared with 0.752 for Medium chances. Similarly, we might expect a strong negative correlation for win percentage and the High chances against (ie the more good chances you give up, the less likely you are to win), but it is only -0.261, compared with -0.527 for Medium.

It might also be the case, that like Corsi, this is not so much a scoring chance stat as a proxy for possession and offensive zone pressure. If you are creating lots of medium chances that means your team is generally dominating the game overall.

If Medium chances are then a bigger indicator, let’s see how the teams fare on just that stat:

This time, Sheffield come out on top by some distance, so we shall see if that is a significant factor in determining the league overall.

Leave a Reply

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