Wednesday, 28 December 2016


What are Man-Games Lost?
Man-games lost (MGL) to injury is normally understood to be the total number of times a team has had a player or players unavailable to dress for a game, i.e. a single player missing 10 games builds up 10 MGL; 2 players each missing 5 games also builds up 10 MGL.

MGL is the injury statistic most commonly referred to the media and figures are recorded and published in some form on a pseudo-official basis by the vast majority of NHL teams. However...

What are the drawbacks?
Teams' "official" MGL figures are not published on an aggregated basis by the league, nor is there necessarily any consistency in how teams compile and record the figures.

For example, some teams count "retired" players still under contract (more on this later), some count non-NHLers who are injured at the start of the season and can't be re-assigned until healthy, some have very hazy distinctions between healthy and injured scratches.

The most obvious drawback of using raw MGL figures is that they do not account for differences in value of the absent players. A game missed by Sidney Crosby of the Penguins is indistinguishable from a game missed by Gord Facepuncher of the Oilers.

What is CHIP?
Using a player’s cap hit to weight MGL is one way to approximate value.

Cap Hit of Injured Players (CHIP) represents the per-game cap charge of a player missing a game through injury/illness, i.e. annual cap charge* divided by 82.

(*Strictly, I use Average Annual Value (AAV) here, which includes the full amount of potential performance bonuses.)

Clearly, this is not perfect as a measure of value by any means - it won't take you long to think of examples of bloated UFA deals for ageing players or young bargains still on entry-level contracts (even allowing for inclusion of performance bonuses). However, cap hit is simple, stable, comparable across positions and lends itself to snappy, food-based acronyms.

Cap hit above replacement level would arguably be an easy improvement, but again, the focus here is on simplicity.

Why not weight by something else?
I often do - posts on my Twitter feed regularly include an alternative weighting by time-on-ice (TOI), but it's impractical to keep updating the viz here on two bases (and TOI has its own drawbacks - messed up by small/zero samples, less stable over the season and less comparable across positions).

Attempts at single rating metrics such as GVT, WAR and others could also be used as weightings, of course, but are not as immediately intuitive and are prone to small sample volatility within a single season.

I'm definitely not claiming CHIP to be infallible or superior to alternatives.

Which players are included/excluded in your figures?
Players still under contract but either officially “retired” or out indefinitely with no prospect of a return in future are generally excluded from the figures from the start of the season (though remain separately identifiable in team charts).

Minor league players who are often on teams’ IR (non-roster) lists at the start of each season are excluded, but anyone who gets injured while on the active roster during the season is included.

Absences due to illness are included, but healthy scratches, suspensions, and absences due to “personal reasons” are excluded.

Some subjectivity in all of this, obviously, but I try to be consistent.

My team is always the most injured, right?

How frequently do you update the figures?
Usually weekly, life-permitting. Fastest updates will by through the Twitter feed (@NHLInjuryViz) - normally done on a Monday. Viz on this site might typically take a day or two longer to be refreshed.

Where is your data from?
Can you share your raw data?
Yes. Just ask nicely via @NHLInjuryViz and something can normally be arranged.

No comments:

Post a Comment