Sunday, March 15, 2009

Scoring Chances - Consistency

One thing a lot of fans are constantly on about is a team not playing with enough consistency. Matt Fenwick had a great piece on consistency a while back which I will quote at length here:

"I have posted before that whenever a coach or media type says that Local Team X needs "more big _____", that can easily be shortened (and made more accurate!) by deleting the word 'big'. Today's lesson in hockey/sports lingo is that more consistent = better. If the coach says the powerplay needs to be more consistent, then what he means to say (even if it's subconsciously) is that the powerplay needs to be better. If your team needs a more consistent goalie, what they actually need is a better goalie. And if the Flames can only win their 1st round series if they play very consistently, what that actually means is that the can only win if they play very well.And if I play a round of golf with Tiger Woods, I might tie him on the first hole, but when he beats me by 12 or 15 on the round, it won't be because I lacked consistency."

I think that there is a ton of truth to that. Most times when people are clamouring for more consistency from their team they actually just want the players to be better. Nonetheless, I thought that it would be interesting to look at consistency for the Edmonton Oilers through the lens of scoring chances, as counted by Dennis here. Are some players really more consistent than others? If so, in what ways is this the case?

I struggled a little bit in deciding how to present this data, so I'm actually going to present it in two different ways (all at EV). In the first chart we'll see the player followed by the number of games the player was +3 or better in scoring chances, followed by the number of games the player was between +2 and -2, followed by the number of games the player was -3 or worse: decidedly good games, medium games, and decidedly bad games.

(For those unfamiliar with the metric, a player is awarded a chance any time someone on the ice has a chance to score. He is awarded with a “chance for” if someone on his team has a chance and awarded with a “chance against” if someone on the opposing team has a chance to score. The players are all organized according to their jersey numbers.)

Games One to Sixty-Eight, October 12 to March 14
Missing Games 22 at St. Louis, 33 at Vancouver and 40 vs San Jose



I would be prepared to label anyone with more good and bad games than they have medium games, with neither the good nor bad being dominant, "inconsistent." The guys in this category are Gagner, Hemsky, Gilbert, Souray and Horcoff, also known as, most of our best players. I find this fascinating to say the least. Nilsson, Cogliano, Pouliot, Penner and Stortini look to be the most consistently average players. The young guys, the guys that tend to be stereotyped most often as being "inconsistent." The only guy on the team kicking ass and taking names with some consistency is Visnovsky though he does have a few bad games sprinkled in too. It's kind of frightening that he might end the year leading the team in "good games."

The next chart will hopefully provide a little bit more information and it's something that I think people will find interesting anyway. This is simply the player followed by the games which he has outchanced the opposition, followed by ties, followed by games in which he's been outchanced:


Looking at the data this way gives a bit of a different impression but gives some clearance for a guy like Nilsson over a guy like Stortini who may not get totally destroyed but is on the wrong side of the ledger on most nights. Looking at the data this way also makes Gagner, Penner, Visnovsky and Grebeshkov look the most consistenty good, though even then only Penner has more mildly poor games than he does awful ones. Pisani, Strudwick, Moreau, Staios and Smid aren't inconsistent, they just aren't very good.

I think the bottom line is still with Matt's assessment, namely, our top players just aren't good enough. Still, it's interesting that they play more than half of their games dominating or being dominated in roughly equal measures. At the other end, the many of the younger players generally have a narrower range of performance, generally staying within +/-2. It seems to me that consistency really is an issue for some players more than it is for others.

17 comments:

Coach pb9617 said...

Hey Scott - you've been doing some really interesting work here lately, but I have one suggestion - steal Tyler's method of taking screenshots of excel spreadsheets to post inline. Some of the data is tough to decipher at first, I just end up dumping in into excel and separating the columns.

Keep up the good work

Dennis said...

I liked the second way you listed the guys because what I quickly did was eliminate the middle bracket and then do a plus/minus including just the games in which there were clear decisions.

The first things I see are things I've noticed all along; which only makes sense considering I am the one who logged the chances:D

Both Cole and 89 have suffered through a lot of bad luck. 89's on a little bit of a counting stats run now and that's really not a surprise. Cole's counting stats sucked but once again, it looks like bad luck. 71 is indeed the man and I'll say 37's the next Hejda in terms of a guy who'll excel if he goes elsewhere. I've always been a bigger 77 than 37 fan but I dare say that right now Grebs is driving that bus.

Finally, 85 seems like a boom or bust guy and 18 is having a super lucky season.

BTW, I always consider consistency as something that's easily predicted; it might fall as good or bad.

Scott Reynolds said...

steal Tyler's method of taking screenshots of excel spreadsheets to post inline.

I'm pretty inept when it comes to making things like that work. I would love to do this but I don't know how. Do you know where I could look to learn or, if it's not too complicated, explain to me how to do it. Also, thanks for the encouragement.


BTW, I always consider consistency as something that's easily predicted; it might fall as good or bad.

Yeah, I was just looking for which guys have a smaller range of performance and most of the guys with this smaller range are either bad (Moreau, Staios, Pisani...) or consistently mediocre (Cogs, Nilsson...). Fits pretty well with your definition I think. Anyway, I'm glad to get your take since these are really all your numbers anyway.

Bruce said...

I think that I would be prepared to label anyone with more good and bad games than they have medium games with neither the good nor bad being dominant as "inconsistent." The guys in this category are Gagner, Hemsky, Gilbert, Souray and Horcoff, also known as, most of our best players.

Scott: Surely there's an ice time component here as well. If you set the bar at a firm number like +3 or -3, the guys playing 18-24 minutes are far more likely to exceed that number either way than the guys playing 8-12. Moreover, the guys on the fringes are more likely to get the short end of the bench if they're struggling a bit rather than being given more opps to bleed chances, whereas the top guys are going to play regardless.

If you set your threshold at, say, +/- 1 per 5:00 EV TOI, you would iron out some of the observed inconsistency.

JaKing said...

Excellent work. I'd love to see running totals -- game by game -- for scoring chances. It's some amount of work, but makes the stat much more readily available and useful.

As for consistency, there's that darn Penner again, tearing up the pea patch, doing just fine.

I agree that 37 is driving the bus with 77, but 77 is playing some of his best hockey with 37, just as 71 did.

I'd like to see 71 scoring chance stats with 44. They weren't such an excellent pairing as 77-37, which is best Oiler d-man pairing since Pronger was in town.

As for Fenwick's overall point, he's right. There's so much noise about this and that, when really the best team gets the most chances, gives up the least, so it most often wins. Talent rules in the NHL, though we've all been bothered in Edmonton by MacT's odd line combos, which have seemed to stiffle the talent of the various players somewhat.

Whether this is an illusion or not, I don't know. Whether some other coach could have settled on strong and consistent line combs that would have increased the scoring chance plus/minus for each forward is an unknown, though it's the big knock on MacT.

dstaples said...

Good work, Scott, and good point, Bruce.

Coach pb9617 said...

First list

Second list

dstaples said...

BTW, I was "JaKing" My son Jack must have been logged on. And I meant 71-37 was the best defence pairing since Pronger and Co.

Coach pb9617 said...

Lordy does that cast Moreau in a terrible light. He's the worst forward on the team.

If you set your threshold at, say, +/- 1 per 5:00 EV TOI, you would iron out some of the observed inconsistency.

Moreau's added TOI might make him the second worst forward on the team.

He's consistent alright, consistently bad.

Coach pb9617 said...

Heh, and not to bang the drum here, but it also shows the most consistent forward on the team is...Dustin Penner.

Coach pb9617 said...

So the last column in this list is percentage of games in which the player is outchancing or tying the opposition.

The culled numbers from the shot are here broken out by forward and defense:

Penner 68%
Nilsson 63%
Gagner 61%
Horcoff 60%
Cogliano 58%
Cole 57%
Pouliot 55%
Hemsky 54%
Reddox 50%
Brodziak 49%
Stortini 49%
Moreau 43%
Pisani 29%


Grebeshkov 67%
Visnovsky 66%
Souray 52%
Gilbert 51%
Strudwick 49%
Smid 44%
Staios 41%


So, this is interesting. MacTavish's two whipping boys are...well, doing really good work in the scoring chances department. Really good work.

The Captain is doing poor work and it's clear here how awful the Pisani at center experiment really was.

Scott Reynolds said...

If you set your threshold at, say, +/- 1 per 5:00 EV TOI, you would iron out some of the observed inconsistency.

Really good point but I don't have the times for all of the individual games handy and I wanted the broader sample of games. I'll do another one at the end of the season and I'll try to implement your change.

I'd love to see running totals -- game by game -- for scoring chances.

Yeah. It's easy for me to check up on because I have it all in a spreadsheet but it would probably be nice to have a running total published for others. I'll throw one up tomorrow and try to update it after every few games.

The Captain is doing poor work and it's clear here how awful the Pisani at center experiment really was.

Thanks for the links Coach. With this point, it's not just Pisani at center, it's just Pisani. Since returning from injury he's outchanced once and been outchanced seven times with one being a shellacking. He just hasn't been good this year.

Scott Reynolds said...

Coach,

How did you create those images from Excel? They definitely are much easier to read then what I put up using asterisks to create spaces.

Coach pb9617 said...

Scott,

All I did was put the data into excel and then key CTRL-PrintScreen. I used MSPaint and pasted the screen into paint. Once it was in MSPaint I used the select box to get the portion I wanted and cut it into a new MSPaint image. Save that as a jpeg or a gif in paint and there ya go.

Bruce said...

Really good point but I don't have the times for all of the individual games handy and I wanted the broader sample of games.

Scott: I wasn't suggesting you actually do it, I was merely pointing out that this might explain the observed "inconsistency" in the high ice-time guys.

Scott Reynolds said...

Scott: I wasn't suggesting you actually do it, I was merely pointing out that this might explain the observed "inconsistency" in the high ice-time guys.

Yeah, but it really is a good point and if I put in a bit of work I should be able to find the data and use it to put together a better look at this. It's not like there will be hockey to watch in the summer so I should have more time.

Bruce said...

Come to think of it, Dennis does log the TOI per situation right by the scoring chances, doesn't he? I find this useful on an individual game basis, but it might be significantly more so when tracking results over time as you are doing.