(It feels as if this is more a "By the numbers" thread, but I understand that 1) icetimes are missing for past seasons, and 2) there is an interest here in trying to estimate them.)
I'm talking about ice time estimates here. Why would you pay any attention to GF% and GA% if you weren't interested in the approx. proportion of the team's minutes the player was playing?
I think GF% can be used just like it is. It tells how many of the team's GF a player was on ice on. All goals counts equally in hockey, and it really doesn't matter in what situation it was scored.
If I remember correctly, only two players (Wayne Gretzky and Mario Lemieux) have had GF% of 50 % or more, since the early 1980s (or if it was 1968). That is very indicating, as those two are regarded as the perhaps best offensive players ever.)
Q: But aren't players who plays a lot of PP are favoured compared to those who don't play PP?
A: Well, that isn't necessarily a problem, as players sort of "earn" their icetime. There are always things factoring in, like PP icetime and ES icetime. A player's production is also affected by his teammates, and by his role on the team and how he is used (for example in a "shut down" role). And that is difficult and almost impossible to "adjust" for.
I think using GF% as it is, is fine (unless you want "per minute" stats). It tells us how much a player have contributed to his team's GF totals.
GA%, however, is completely different and much more difficult. Here the goal of the player is not to get as high number as possible, but to get as low number as possible. And different players have different icetimes, meaning that (as this thread is about) players with large icetime getting higher GA% than those with very little icetime.
While the players with the highest GF% is basically the best offensive players in the league or on a team, the players with the highest GA% are probably not at all the worst defensive players.
My current take is to take GA% in itself as a "curiosity" stat.
And yes, I agree that the biggest use for GA% is to get a picture of how much icetime the player had.
So I agree that GF% and GA% it can be used to estimate icetimes. But I think we need large differences between players on a team in order to tell which one played the most in different situation.
I think the coach/GM generally knows what he does. If a player is good on the PP, he will get his PP time, and if he's not he will not get it. (Yet, this is a difficult topic. There are for example great penalty killers who aren't used at penalty killing, because they are being more valuable when playing ES and PP. And vice versa. But generally, I think overall point production tells how good a player is offensively.)
That's great, but my point is that in this context "game" for Player A is not necessarily the same as "game" for Player B, making the results inherently misleading. If one player plays 20 minutes include 5 on the PP, while the other plays only 12 minutes at ES, then any per-game comparison between them is inherently misleading on the player level, because you're not comparing players on the same basis.
See above.
Yes, the "context game" may vary considerably between different players. When it comes to offensive production, I think one can just use the "total points" (or total points percentage).
You mean that it would for example be interesting to notice that "Player X only had 22 PP points during the season, but when he got the chance he really was effective being third best on his team on a per minute basis"?
Well, I have studied that subject too. I think it does not contribute much of value. For example, the numbers (points) which we divide by "minutes played" are often small. The results don't really meet the "eye test" (or how you say it). A player being great "per minute" one season may be below average the next season. "Randomness" plays a too big part in things.
It is understandable. 5% is the average error, but the error can be 0% for some players and 15% for others (that's about the highest for my estimator).
Exactly.
And as long as you bear that in mind, there's no issue.
Well, if it is considered a "curiosity stat", little harm can be done.
What do you mean by reliable? Something does not have to be 100% accurate to be useful.
I absolutely agree.
Absolutely, SH GF and GA have to be estimated, which can be done fairly well. It's part of what is built into the average error.
Well, from what I've seen, one uses PPGA% to estimate SHGA, and PPGF to estimate PPGA. That is, one uses an estimation (with error rates) to make another estimation.
I myself used a safer method. By looking at gamelogs, one can sometimes deduct which players must have been on the ice during the SH goal. It was a time consuming method, and it also depended on doing correct calculations/deductions. I have been thinking about writing a program (I'm a long time programmer) that goes through all available gamelogs, but at the time decided I would be better off doing other things.
The method would help us narrowing things down, by for sure knowing how many SH goals a player at least was on ice for/against.
I don't disagree that people can and will misuse such estimates. That doesn't mean it's an issue with the estimates, it's an issue with how people use them.
Yes. (It's just that I think they are so frequently being "misused".)
I don't disagree that people don't understand the role than normal variation plays in player statistics. But the solution to that is to stop using statistics, because some people might misuse them. Illustrate how they're misused, when they're misused.
Yes. And here we should also add that context is often overlooked. Both normal variation and things like teammates, roles, etc. affect statistics.
You'll have to be more specific here. What are you surprised isn't common knowledge?
I encounter it on a daily basis. Basically things like the ones mentioned here, like how normal variation works, how small differences between players are often neglectible, how no consideration is being paid(?) for teammates, role on team, etc.
Another example:
If player A have more SH icetime than player B, it does not necessarily mean he's considered a better penalty killer. It might simply be more useful for the team as a whole to use him on other situations instead. (Peter Forsberg might be an example of a player who could play SH well, but it was considered more valuable for the team to let him focus on ES and PP only.)
Overall, I find defensive performance and value is very hard to quantify.
Plus that we have the goalie influence too. Having a prime Roy or Hasek will help a player far more than having a below-average goalie. (I know that for this thread in itself that doesn't matter, because all skaters on a team should be affected basically equally.)
