What is a zone start?
A zone start is simply what zone a player begins his shift in on shifts that begin with faceoffs. Unlike the statistics that we've looked at before now, zone starts do not reflect player performance directly, but show how the player was used by the coaching staff.
Zone starts are mainly calculated as the percentage of zone starts a player , with the most common form ignoring all shifts that begin in the neutral zone. The purpose of ignoring neutral zone starts is that they're a zero sum effect. They're by definition, neutral, and what we're looking with zone starts a way to quantify how a player is used either on the offensive or defensive side of the puck.
Let's bring back Player A and Player B with another made up data set:
|Name||Offensive zone starts||Neutral zone starts||Defensive zone starts|
This is how you'll see the raw data for zone starts most of the time. The way you would calculate zone start percentage is the same formula you would use to calculate Fenwick or Corsi percentage:
From this we can see that Player A was used in an offense heavy role. He was sheltered away from defensive responsibility as much as possible. Meanwhile Player B was used in a very defensive role.
What do zone starts tell us?
Looking at our example, we can see that the players were used differently, most likely because the coach has identified that Player A isn't very good defensively, but has offensive value. Player B, however, seems to be someone the coach can rely on to play tough minutes in the defensive zone and not hurt the team.
The main thing that zone starts tell us is how the coaching staff views the player, and how that translates to usage. In single games, you don't really learn much because of mitigating factors. P.K. Subban usually plays a tough minutes defensive role, but when the Canadiens are trailing, Michel Therrien often shifts him into the offensive zone in order to generate more offense at the sacrifice of defense.
Zone starts also give us context for possession numbers. If we say that the Fenwick numbers from yesterday's post were from the same game as the numbers for our zone starts example, all of a sudden Player B's 56.5% Fenwick looks a lot more impressive.
What are the limitations of zone starts?
Zone starts are not an all encompassing usage statistic, and getting a ton of offensive zone starts does not necessarily mean that a player is poor defensively. Max Pacioretty was given more offensive zone starts last year than nearly any forward in the NHL, but he's a very effective defensive player. Michel Therrien and his staff judged that it would be most effective for Pacioretty to carry the defensive load for Desharnais in an exploitation role, generating more offense at the risk of playing inferior players in the defensive zone when Plekanec is on the ice. Whether this is the best use of Pacioretty is up for debate, but examples like this is why careful analysis informed by watching how the lines work is necessary.
Some coaches also don't zone match, so there are times when you don't learn anything from a coaching perspective from looking at zone starts, but you can still use them to an extent to contextualize performance.
It's still debated how much of an effect zone starts have on possession. There have been a few different studies on it, and it seems like we get a new number every time. There are some people who even say there's no effect at all, but that's quite the absurd opinion considering the amount of evidence suggesting otherwise.
The other main criticism is the way in which zone starts are calculated and represented. Some believe that neutral zone starts should be part of the equation. That the stat is usually not said explicitly to be eliminating neutral zone starts can be seen as dishonest.