clock menu more-arrow no yes mobile

Filed under:

The PK Subban/Matt Pfeffer fallout tells us a lot about how the NHL attitude to analytics still refuses to change

Matt Pfeffer losing his job in Montreal, and the reaction to it, shows us that analytics still have a very tenuous place in the minds of many inside the NHL-to the point some are almost wilfully ignorant out of principle now.

By pucktistic
Eric Bolte-USA TODAY Sports

By now, you've no doubt heard about the latest twist in the tale of PK Subban and the Montreal Canadiens.

Even weeks after he left the Habs, the Subban Trade (the capitals almost seem to be implied) is still the very hottest of topics inside NHL circles. Up until last week, it was following the usual path.

We've had the angry reactions, we've had the smear campaigns, we've had the fighting talk about how trading a franchise cornerpoint "makes the team better" from the front office. We've even had the slow rationalisation of media trying to talk themselves into it and the counter-takes...perhaps the most interesting of which being this one on the French section of Habs EOTP, which confronts questions about Shea Weber's declining performance by...wondering if it could be Roman Josi's fault.

However, this Subban trade has thrown a new wrinkle into the mix - the fact that teams are seemingly not only going against the advice of the experts they've hired to actually analyse these trades, but now, actively punishing those who argue against management decisions ased on their statistical evidence.

Habs' data analyst Matt Pfeffer was released by the Canadiens last week, in a move that has been strongly linked to the fact that he, as a consultant to the Habs management, not only argued against the Subban trade but apparently did so so vehemently that someone in the top echelon decided that it should see him lose his job, barely a year after being hired.

TSN Montreal's Eric Engels explicitly linked Pfeffer's argument against the Subban trade to his canning by the Canadiens last Wednesday, saying the following:

A source told Sportsnet that Matt Pfeffer, who was hired as an analytics consultant at the beginning of the 2015-16 season, made an impassioned and elaborate presentation to management to dissuade them from following through on this trade. Ignoring Pfeffer’s advice only served to reinforce the notion that Bergevin was following different criteria in his evaluation of both players, said the source, who also suggested Pfeffer’s vehemence on the matter might have ultimately cost him the job (he was told on Wednesday that his contract won’t be renewed).

This was countered by Pfeffer himself then revealing that the presentation wasn't actually face-to-face after all in his own statement, which raises the question why something like this was actually leaked from the Habs, who are notoriously difficult to penetrate when it comes to leaks to media.

However, what we can infer for sure is that the Subban trade was made by Bergevin and the Habs against the advice of the analytical section of his own organisation...and leaves at least the strong possibility that arguing too vehemently against it meant that Pfeffer was considered "expendable" by the organisation.

Now, we know about the continuing battle between fancystats and the eye test. We can talk about the actual proven effectiveness of analytics until we're blue in the face and nobody will budge either way. It's also important to say that the rhetoric in this discussion is flawed on both sides, with unnecessary insults and dismissal being thrown around (if you don't believe me, just take a look at the way those in support of analytics are referred to in the comments of that stats piece).

However, the Pfeffer case is highly interesting because it appears to mark a new level in the analytics battle. Up until now, we've seen the arguments fly back and forth between those in favour and those against it, and it's not a generalisation to say that those who are most fervently in the "anti" camp are...shall we say, more traditional in their hockey outlook, and hold to a certain narrative when it comes to what's important.

To this point, though, although we've seen plenty of dismissive talk, we've rarely seen quite the climate of almost wilful ignorance of what analytics shows as that shown by the Montreal front office and their defenders in the Subban saga.

We've seen GMs and "hockey people" express scepticism over the value of analytics, we've seen arguments against them...but now we're seeing people's livelihoods reportedly being affected by how strongly/vehemently they're analytics supporters or not.

Let's take the rhetoric around another big trade this offseason, too...that of Edmonton's trade of Taylor Hall for Adam Larsson. This is another trade that common sense and analytical arguments both seem to say is a bad idea - one that has seen the Edmonton media falling back on the weakest of arguments in an attempt to justify it to their readers. Compare these two tweets, for example:

Two weeks later, in respones to repeated questioning whether or not Adam Larsson will be the offensive defenseman the Oilers seemingly need, the same writer:

This is a little strange, given the same writer appears to not be able to tell what an offensive defenseman actually is:

Here is, clearly, a man who hasn't actually looked into the analytical evidence that says Travis Hamonic is...er...far more offensive than Larsson is.

This is the thing - at this point there appears to be a significant section of the NHL media that is not only sceptical of analytics - they're almost proud of wilfully ignoring the evidence that they present when making an argument-an argument that is easily available given the simplest of research. The example above is merely one in hundreds. You can probably find your own examples.

However, the firing of Matt Pfeffer seems to mark a new high-water mark in the debate...one where actual team management personnel are stepping beyond the carefully-worded talk of it being "an aid, not the be-all and end-all" and at least one team are now willing to at least give the impression they're now openly disdainful of what their own hired staff are telling them, and that presenting "inconvenient truths" is the quickest way to get yourself put out of an analytics job in the NHL.

That is a dangerous new development - it's one that demonstrates more than anything that even in 2016, the distrust of analytics is such that some NHL staff will openly ignore every indication it gives in order to pursue their own path. It's also potentially set a dangerous precedent - one that will see analytics staff across the NHL likely more fearful of presenting the "wrong" data to their team management. After all, if one GM will fire you for presenting data that doesn't agree with their opinion, who's to say another won't?

What the Canadiens have done with the Pfeffer affair is introduced a new element into the analytics debate...one that may see analytics professionals more reluctant to contribute to their team for fear of presenting the "wrong" opinion.

That will be a ripple effect that could ultimately harm teams across the league in their evaluation of future moves, and lead to a regression in the way hockey is evaluated and analysed.

That's good for nobody - fans or team staff.

After all, refusal to embrace progress only means that you risk being left behind.