The cost of doing business


(Disclaimer: read this post before anything here)

If you browse through the leaders of Relative Corsi or Points per 60 minutes, or any other of my favourite measures to quantify players, it’s not like you’re ending up with a bunch of third lines scrubs up top.

We’ll look at the 2012 season because it’s the most recent 82-game campaign, with a large enough sample size to take away some of the hangers-on that lived as linemates for a full season. Patrice Bergeron and Tyler Seguin at the top. Keep going down, and you’ll find players generally recognized by everybody as good two-way players: Anze Kopitar, Pavel Datsyuk, Joe Thornton, Henrik Zetterberg… most good players wind up with good possession numbers.

There are very few players that hockey fans recognize as “good” that the statistics don’t suggest are. While certain players have skill sets related to defence or shooting that may not be captured entirely in Corsi statistics, they’ll generally pop up near the top of some other list. Who cares if they do. You can tell who good players are by watching them. That was never the issue.

My enjoyment of statistics in the last year or so has come from trying to predict how players will play in the future. With a few more years of data we may have enough to be able to have real possession aging curves, or a proof that perhaps a player’s defensive ability peaks later than his offensive ability. I think statistics have a lot of value less in trying to figure out a player’s worth and more trying to determine a player’s future worth. If you want to rank players by how productive they were to their team, all you need to look at is goals, points and plus-minus. A good Corsi rank and a low PDO doesn’t necessarily mean a player contributed to a team’s success.

Mikhail Grabovski’s a good example of that. He didn’t really contribute much to the Leafs’ playoff appearance last season, though it’s unreasonable to think that he’s anything but a better option for the future than Tyler Bozak. All things being equal; chemistry, likability in the dressing room and so forth, I think we can all agree that Grabovski, who scored more goals and points in the last three seasons and was the best play-driving centre on the team, is simply a better hockey player. Last season, where Bozak was heavily relied on, the Leafs made the playoffs.

I hate going back to that comparison, but the Toronto Maple Leafs are so interesting from an analytical perspective because they’ve almost entirely decided to go with the players that the coach intuitively selected and got rid of better players. I’m not entirely sure that Clarke MacArthur is better than David Clarkson, but I do know that MacArthur is much cheaper and on a much, much less risky contract. The Leafs are a walking market inefficiency, overvaluing certain traits that some teams don’t even bother to pay for.

Statistics are less about ranking players and finding who’s the best player as they are for attempting to forecast how the player will play in the next season or two. If a team has a wider amount of data to use, they can probably draw more historical comparables and have a closer guess to how a player will perform over the life of a contract. Nothing is certain with any one player, but it you provide smart thinking and a rational, data-based approach to every player signing, you’ll probably come up on top more often than not.

I found Craig MacTavish’s comments about David Perron really interesting after the Edmonton Oilers acquired him:

“Our analytics guys … they love this guy,” MacTavish told Oilers insider Bob Stauffer on Oilers Now. “He’s right up there in terms of controlling the play and shots for and against differential. So it will be a good test for our analytics guys. They have him with some of the game’s elite.”

I admit that when he was drafted, I was bearish on Perron even though I wasn’t super into analytics at that point. I don’t think I had ever consciously seen a highly-ranked player with just one year of junior hockey and was a year older than everybody else in the draft pool. I thought it might be a bad idea, and maybe over the long run, I’d be right, but there’s no denying that David Perron has become an excellent hockey player.

Among the game’s elite? That’s a bit of a stretch. He’s usually been a plus-player when facing fairly tough competition. He has second-line calibre offensive talent while playing favourable minutes alongside some of St. Louis’ best players, David Backes, Patrik Berglund and TJ Oshie were his main linemates over the last two seasons, all of whom appear to pull up Perron’s socks.

Now, even if the Oilers primarily used shot statistics to make the trade, and even if Perron was called one of the game’s elite, is it still an excellent trade for the Oilers? The player they gave up, Magnus Paajarvi, is a talented player with some defensive capabilities that some Oilers bloggers wanted to see play a bigger role. From an analytics perspective, it may have been wiser to develop in-house, since Paajarvi will probably wind up with a much cheaper contract than Perron, having very little to show for at the NHL-level. It’s an interesting “win now” move for Edmonton, since they acquired the player who is closer to his peak.

Will Paajarvi get his shot in St. Louis? Maybe, maybe not. They have a lot of offensive talent, but no standouts. Top six jobs should be open competition, and perhaps we can see how Paajarvi performs in that kind of role next season, and compare him to Perron’s performance.

Every trade has an opportunity cost to it. Whether general managers like it or not, they make bets on player performance with every trade and signing, and generally, the managers that think smartly build the best teams, even if not all of their moves pan out. For the Oilers, it’s not about how Perron plays, it’s about how much better Perron will play compared to how Paajarvi would play in the same situation, which is difficult to calculate since not all minutes in an NHL game are equal. For the Leafs, they determined that a slightly-more expensive Grabovski was not a big-enough improvement over Bozak.