One of the main purposes of statistics is to predict likely outcomes and make informed decisions based on probabilities. Smoking doesn't guarantee a shorter life, but statistics show it makes it more likely. Investing your savings in the stock market instead of a lottery ticket isn't guaranteed to the better fiscal decision, but it's likely to be. At the end of the day, there are always going to be some smokers who inexplicably live to 85 years old as well as those handful of scratch-off addicts who manage to win millions of dollars. Statistics aren't foolproof, but they are very, very good at predicting the future. The vast majority of smokers will deal with health issues. People investing money will almost always come away richer than people playing the lottery. Hockey analytics are the same. They're the best tools we have to make predictions and evaluations, but they aren't perfect.
In the last 6-12 months, you can find a handful of articles on Blueshirt Banter that discussed how Kevin Klein's scoring surge was unsustainable and insisted that trading Kevin Klein would be the right move. We were wrong. But only kind of.
We were wrong - specifically, I was wrong - to say that Kevin Klein's scoring output last season would not carry over to this season. For the second-straight season, Klein has scored eight-plus goals despite only potting 17 goals in the first 433 games of his NHL career. But we were wrong only in the sense that a blackjack player was "wrong" to stand at 18 when the next card dealt was a 3. It was the right judgment based on all available evidence, and yet an unlikely outcome occurred. After shooting 11.8% last season despite being a career 3.9% shooter, Klein has laughed in the face of regression and has somehow improved upon that number this season; he currently owns a 15.8% shooting percentage.
Does this mean that the "eye test" failed us, or that we need to change our understanding of statistics? Of course not. The highest Brian Leetch ever shot in a single season was 9%. Nick Lidstrom bested at 11.1% in the shortened 94-95 season. Al MacInnis maxed out at 10.2%. What I wrote two Decembers ago still holds true. We have to either accept that 30-year-old Kevin Klein magically became head-and-shoulders better than the best offensive defensemen in NHL history, or we have to accept that these numbers are a fluke. It should be pretty clear which one is the realistic answer. That Kevin Klein has extended his high conversion rate this long is a statistical improbability. It still is very unlikely that this continues.
Where we got it wrong on a more meaningful level is in regards to the rest of Klein's game and what it has meant for the Rangers. Many advocated for the team to trade Klein during the offseason, when his value was presumably at the highest it would ever be thanks to the previously mentioned scoring fluke. Instead, Klein has played some of the best hockey of his career this season.
Via Corsica.Hockey, this chart shows Klein's rolling 25-game Corsi For % since the 2011-2012 season. It's a bit hard to read, so here's the gist. In 25-game spurts, it shows the percentage of shot attempts Kevin Klein's team has at five-on-five when he is on the ice. The number is adjusted for the score. Earlier in his career, he was consistently well below 50% (the dotted line), which is not good. Since joining the Rangers, he's been inconsistent at best, with some great spurts but also some putrid ones. This season, he's been consistently around the 50% mark, which is very good considering that the Rangers are poor in possession overall. This is highlighted in how his numbers look when made relative to the rest of the team.
In essence, the puck is in the offensive end a lot more and in the defensive end a lot less when Kevin Klein is on the ice than when he isn't.
Klein's improved play has been particularly important to the Rangers this season. The rock-solid Dan Boyle from last season is now a third-pairing power play specialist, while Dan Girardi has gone from below-average to abysmal. Klein won't move mountains, but he's not going to drag anyone down, either. It's not a coincidence that Ryan McDonagh is a paltry 42% in shot attempts and -7.19% relative to the team when paired with Dan Girardi, whereas he is 53% and +8.71% respectively with Klein.
So we were wrong in our prediction of how Klein would play this season and how (not so) disposable he would be to the Rangers. But again, only kind of. We were flat-out wrong in our projection of Klein's 15-16 season. Maybe there are a few things to learn from this, but overall I believe we were wrong purely in result; not in process. For defensemen in their early 30s, what you see is what you get. Probably 99 out of 100 players either stagnate at this age or even begin to decline. Klein is the one guy who managed to buck the trend and reach a higher level than he's ever performed at. Over the previous five seasons, Klein was a strong 3rd-pairing defenseman. This season, he has morphed into a really solid second-pairing defenseman who can even slip onto the top unit without derailing everything. If you claim to have seen this coming, you are either a blind optimist, a wizard, or a liar.
We were only sort of wrong in petitioning for a Klein trade. His emerging importance to this team is less about him and more about everything crumbling around him. Reality is that a contending team should not be this dependent on Klein, even with his improved play. With a proper first-pairing defenseman instead of Dan Girardi, Kevin Klein would be a $2.9M luxury the Rangers don't particularly need. McIlrath has proven success alongside Yandle, and Boyle could play sheltered minutes on the third pairing. Klein could have been traded for some big assets that the Rangers needed then and need even more after the Staal trade. That cap space would also go a long ways towards re-signing Keith Yandle. The Rangers were right to keep Klein, but only because getting it wrong elsewhere has forced an unnecessary dependency on him.
A few questions emerge. First, will Vigneault finally acknowledge that Girardi does not belong on the top pairing? Klein isn't the ideal partner for McDonagh, but he's a major upgrade. The other is, what should we do differently now that Klein has (sort of) proven all the analytical statistics wrong? The answer is absolutely nothing. There's a prevalent fallacy that people in analytics are arrogant and refuse to admit they are ever wrong. This is not true at all. Here's an entire self-written article exposing how we were wrong! We start to get cranky and irritated when people insist that this somehow proves analytics to be unreliable. Again, statistics aren't perfect, but they are right far more often than not. If you're at the blackjack table and a three is dealt after you stay on 18, you're not going to start hitting on 18 in future hands. Kevin Klein's successful season is like a warm, sunny weekend when the weatherman predicted a thunderstorm. You're happy that it happened and you're going to enjoy it. However, you're still packing your umbrella the next time he predicts rain.