‘Imitation is the sincerest form of flattery’
Charles Caleb Colton
This post will hopefully demonstrate that looking at a league table during the season may not be the best method in judging how well a team is performing, and will perform, come the end of the season. This is hardly a revolutionary position, and is undoubtedly a feeling that can be appreciated by coaches and fans alike.
Before I discuss what I have done I would like to say that the methodology was first proposed by Simon Gleave (twitter @SimonGleave), Head of Analysis at Infostrada, in this blog post. I have decided to replicate his analysis to hockey for a number of reasons. I suspect that the majority of people who read my blog do so as they are involved/interested in hockey and thus will be unaware of his work. As well as making my readers aware of his work, it may also help those who do work in hockey as it is really simple to conduct. Finally, I did this as I knew immediately after reading his post I had to apply it to hockey to satisfy my own curiosity with regards to my team’s performance!
Simon proposed what he calls the ISG coefficient, a value calculated by comparing the results this season with those achieved in exactly the same fixtures as last season e.g. If team a played team b at home this season, we compare the points total from their game this season from when team a played team b at home last season and do this for all matches played so far.
To quickly and crudely test the assumption that teams maintain a certain level of performance that manifests in their end of season points total, I correlated the England Hockey League (EHL) West Conference’s 2011-12 teams points totals with their 2010-11 points (R2 = 0.80) and EHL West’s 2010-11 teams points totals with their 2009-10 points (R2 = 0.86) (see graphs below). As can be seen, there is a positive relationship with points in one season with points in the next season.
Having done this, I then proceeded to calculate the ISG coefficient for each team currently in the EHL West Conference. Simon suggests that promoted teams should replace the relegated teams, thus when looking back at last season matches, I replaced Lichfield with the University of Exeter and Bournville with Firebrands.
*Update: I incorrectly treated University of Exeter as a team who were promoted into EHL West, when they actually came down from the league above. Replacing them with Lichfield then will have distorted their ISG coefficient and points projection, and project them as finishing lower than where they will most likely finish.
The projected points is calculated by adding the current points totals, with the number of points that would be gained if the remaining matches for a given team have the exact same outcome as last season.
The below image shows the current points, ISG coefficient, and projected points for all EHL West Conference teams.
What can we take away having calculated the ISG coefficient and the projected points totals?
Team Bath: Despite the worst ISG coefficient, they are currently 4th (just 4 points off 1st) and still projected to finish top, and so it would be naïve to write them off at this stage. Nevertheless, this is perhaps clouded by the fact that Bath had an exceptional season last year, winning all but two matches (lost one and drew one). Producing the kind of form they had last season will be very difficult, and though they will finish near the top, I think it is more realistic that Exeter and Cardiff & UWIC will finish top.
Cardiff & UWIC: Currently joint 1st, and finished 2nd last year. +4 ISG coefficient and a projection of 2nd place seems reasonable, though only if Guildford are unable to maintain their current form. Nevertheless, if Cardiff & UWIC are able to repeat last seasons results, and Bath are not (as mentioned earlier) they will win the league. As I am the analyst for Cardiff & UWIC I know we are in a very strong position, with the results of a 1/2 season analysis report I produced revealing that we have, on average, more 25 entries, circle entries, penalty corners, shots and goals than our opposition per match.
Given that there are just 10 teams in the league, and that just 8 points currently separates 1st and 7th, care must be taken to not read too much into the end of season projections. This should not deter others from applying this to their own hockey league in search for a better perspective on a teams performance during the season however.