Thursday, September 22, 2011

JaredL's Principles For Evaluating Roster Decisions

Now that the long part of the summer is over, hopefully along with all the horrible news stories, training camps are open and preseason games have started, a lot of the focus goes to roster decisions. This includes who should make the team and who should be shipped back down to the minors, what the best line and defense combinations are, who should play on the power play and penalty kill, which lines should get the offensive starts and who should get the tough minutes. In the style of fellow Driving Play author Triumph, who wrote on evaluating prospects, I humbly present my principles for evaluating the decisions made by the coach and front office of your favorite team.

You'll see that as I go on a lot of these principles will be focused versions or combinations of previous ones. I'm trying to convey not only particular ideas but what I feel is a good overall thought process.

Principle #1: It's Complicated!

Hockey is not an easy sport to analyze. A lot of the things we love about the game make it tough - the game flows instead of being chopped up into individual plays, substitution and coaching adjustments happen on the fly, there are many different styles of play, there are a million different hockey skills and most of them can't be isolated or measured and luck is a big enough factor that in a few minutes or even several games just about anything can happen. Other than a penalty shot, nothing is isolated to just one player so teamwork is very important; players must complement each other and be able to work well together. On top of all this, the players get tired much more quickly than in other sports so that puts more limitations on what a coach can do with his lines.

I'm not saying these things make it impossible to analyze, far from it. We can offer a lot of insight on what coaches and GMs should do. We certainly have and will continue to criticize what we feel are bad decisions and at some point we might even get around to offering praise. These problems are very difficult so we should expect even the brightest hockey minds to make mistakes. Also, the level of complexity could make what we on the outside think is a blunder a smart move for reasons we don't know about.

Principle #2: Keep Tradeoffs in Mind

Perhaps I should have titled this there is only so much ice time to go around. One reason for the complexity I discussed above is that there are tradeoffs all over the place. If your coach goes from rolling lines to focusing on matchups then his guys aren't going to be quite as fresh, or at least their rest time won't be as regular. If you give your top line more starts in the offensive zone then that's fewer Ozone starts for the other lines. Your favorite gritty young player earning a spot on the fourth line means someone else is getting dropped to the minors. Putting your second-best scorer on the same line as your top guy lessens your scoring depth. Separating them means your top line won't be as powerful.

Every single roster decision is riddled with tradeoffs. They are unavoidable. You can't only consider the benefits of making a change but must consider the drawbacks as well. More on how to do that below.

Principle #3: Beware of Unintended Consequences

Principle #2 was about there only being so much ice time to go around. This is the other side of that coin - someone has to take up every kind of ice time. In addition to the tradeoffs I talked about above you have to think up a level. If you give your top guys more Ozone faceoffs not only does that mean that there are fewer of those to go around, but they are also not going to be able to take as many Dzone faceoffs so the rest of the team will have to take more of those. You can't have the Sedin line getting 3 faceoffs in the offensive zone for every 1 they take in their own end without the other lines taking worse starts than average - most of it being Malhotra with the ratio reversed. Kane can't get all those easy minutes without someone else picking up the slack and taking on the toughs - in this case Bolland and Hossa.

Principle #4: Be Reasonable!

Especially with matchups and zone starts, there are limits. With matchups, you have very little control in your away games. Beyond that there is the rest issue plus the opponents are only going to put their worst guys out there for so long. You can't give a line nothing but easy minutes, at least in the regular season. You can't give a line nothing but tough minutes, either. The same goes for zone starts. Sometimes two or even three lines are dead tired and you have to put the fresh one out there no matter where the faceoff is.

There is only so far you can go.

Principle #5: Cater to Sensitive Players

In an earlier draft I had a bunch of econ jargon but it was a Catch-22; it would only be understandable if you were already familiar with the concepts. You didn't, but if you feel like you missed out on something here's a link to the wikipedia page on comparative advantage.

So you realize it's complicated, have kept tradeoffs in mind, are considering unintended consequences and are reasonable. How should you make the case that X should get more Ozone starts, with Y picking up the extra Dzone starts or Z should face the toughs with W getting protected minutes? The answer is that you should pander to the sensitive players. In other words, you want to give the players where zone starts or quality of competition matters the most the easier minutes.

Put differently, whichever group has the largest gap in performance between easy and tough competition or after starts in the offensive zone and defensive zone should get the more sheltered time. I'm big on number crunching, but it's still a better way to think about divvying up the time if you just use the eyeball test and your own intuition. The key is to focus on the relative importance of each situation for the players, not in absolutes.

Principle #6: Context Matters

The last few were mostly about line usage, but when evaluating personnel decisions remember to take context into account. Maybe the fourth liner you think was good for nothing and should get the drop actually did well, fourth-line well at least, but faced tough opposition or spent a lot of time in his own zone. It is especially important to look into context stats for fringe players because their starts and especially Qualcomp numbers can be more extreme due to the small sample sizes. According to BTN, of players with 40 or more games Nicklas Grossman had the highest 5-on-5 Corsi QoC last season facing opponents with an average Corsi rate of 1.742. If we drop the games-played requirement to 10 then there were 8 players that faced tougher competition on average than Mr. Grossman, led by Petr Prucha at 3.259.

Principle #7: This Season Isn't Last Season

A lot changes from season to season. In addition to personnel changes, players age and their offseason workouts or time at the Byfuget table can mean big differences in productivity and style from one year to the next. Be especially cautious projecting young players, old players, players coming off injury, players moving to a new team and, I suppose, players who had their linemates/defensive partner leave town. I made this mistake myself in an offhand comment in this article when I said that we should expect Stamkos's stats to take a hit this season. I'm not completely backing off from that statement, look for a post projecting his goal tally in the near future, but I certainly should have thought about it more.


Using the above principles you will not always reach the right conclusion. Sometimes players perform above expectation. That's true for you, your GM and coach as well as your favorite bloggers. Sometimes what we expected of a player is flat out wrong or for some other reason we make bad judgments. While it doesn't guarantee that you get the right answer, the principles above should help your thought process, improve how you look at these sorts of decisions and make you more likely to get it right.

No comments:

Post a Comment