Tuesday, March 25, 2008

Difference Makers


A long time ago mudcrutch had a cool idea to look at the impact of players on their teams. He checked a few star players, to see how their teams fared in terms of wins and losses over the years, this in games that they were out of the lineup.

It was interesting stuff, and intuitively there was definitely something to it, though small sample sizes meant a lot of noise. And now, with the increase of relevant stats information available to us, I think it's worth taking a finer look at it.

I'll get back to this myself, eventually. For now I thought I'd just leave some useful links and see what other people on the blogosphere come up with.

So if you wanted to see how the Blackhawks have done at evens when Havlat is out of the lineup, enter this url to see the games that Havlat did not play in. They are listed by NHL.com game number.

This url will churn out the total goals/shots/missed shots/etc stuff for the Blackhawks this season.

This url will churn out the same stats for the hawks during Havlat's first injured stint. Just repeat this for his other two injury stints (dude makes Moreau look like an iron man).

This is 4v4 and 5v5 stuff, if you want to be exact you can find the goalie numbers for the Hawks here, and just delete the rows for the other players.

From there, get your geek on, cut and paste into a spreadsheet, and it should be basic grade school arithmetic from there on out.

Players that have missed a bunch of games are going to show clearer results I'd think. Crosby, Havlat, Ohlund, who else? ... anyone who has missed a big chunk of the season without too many other tangible factors affecting the team at the same time are favourite. And it will be trickier if you're looking at a team like COL, who had so many of their difference makers out at once.

We're not looking for a magic bullet to determine player value here, just following simple reasoning to see what it tells us. And when it comes to applying common sense, there is no such thing as too much.

Hopefully we can spot some obvious, reasonable, and repeatable patterns here, and get a little closer to understanding which hard facts we should be looking at when we try to fairly evaluate players. Especially those players that we don't get to see very often.

Hopefully some of you take a run at it, and tell us what you found in the comments section.

15 Comments:

Blogger Jon G said...

Well from what I found using those stats was that the were as follows:

With Havlat
record = 16-16-2
Goals For = 97 (GF/G = 2.85)
Goals Against = 89 (GA/G = 2.6)

With Out Havlat
Record = 20-16-6
GF = 122 (GF/G = 2.98)
GA = 123 (GA/G = 3.00)

So looking at it they even though both records are 2 games under .500 the score more while Havlat is out but they also give up more goals.

3/26/2008 9:28 am  
Blogger Vic Ferrari said...

Thanks Jon G.

That looks about right. So Havlat is worth about a quarter of a goal per game or so. That's a lot, star player level for sure.

About a 4% increase in the chances of the Hawks to win a particular game with him in there. At some point I'll check that against how the early game lines from the bookies, but it feels right for sure.

It's half of the season more or less, and EV, SH and PP all in there, no? Enough that the bounces must have washed out.

Props, Jon.

I guess the other things I'm wondering is where those results came out of? Mostly PP or mostly at Even Strength? In both cases was it the number of shots that were affected or the shooting%? I think we can agree that if there was a big swing in save%, or if the PK fell apart for a stretch ... that's probably unrelated to Havlat.

My gut feel is that it's probably 2/3rds from even strength, the rest PP where the goal differential went missing. And that it's an equal share of shots and shooting% at evens, and mostly the shooting% on the PP. But that's just a guess. And I'm not even sure where to get that.

3/27/2008 12:56 am  
Blogger speeds said...

I'm guessing you picked Havlat for a reason, Vic?

He's a guy that has been rumored may be on the block in CHI, if the Oilers are going to move Stoll (RFA, 2.2 to qualify) and Torres (2.2 mil 2 more years)anyways, I wonder if they could land Havlat (6 mil, 1 more year) in some kind of trade involving one or both of those 2 players? It would be taking a risk on another "injury prone" player, but the lineup would look pretty decent when healthy:

Penner-Horcoff-Hemsky
Nilsson-Gagner-Havlat
Moreau-Cogliano-Pisani

as your probable top 3 lines, if they are ever healthy at the same time.

3/27/2008 10:55 am  
Blogger YKOil said...

To me Havlat is a perfect example of a ripple effect player. Put him in and the rest of the team tightens up considerably.

No way the Oilers can bring him in though. From what I can tell the team has ~ 500k in Cap space next year and, maybe, an additional 2 million in flex.

3/27/2008 11:19 am  
Blogger Vic Ferrari said...

It was Jon G that picked Havlat, though presumably he was curious about him for the same reason that I am.

I just did a quick and crude check on the effects of Crosby, Havlat and Lidstrom on Corsi (Even strength only, and Lidstrom only missed about 10 games so it's dicey there).

Still, we all know these guys all effect the shooting%, i.e. the quality of chances generated, in a positive way. Though somehow PIT got a lot better at EVshooting% with Crosby out (they were so low to start with though, they had nowhere to go but up, I'm chalking that up to coincidence).

As for the net effect on Corsi:
Crosby: +113
Lidstrom: +123
Havlat: +113

Now that's a crazy small sample size, but he's in the range of the difference makers. He plays tough minutes too since leaving Ottawa, it's not just his points that are lost, there's a ripple sent down the lineup as other guys have to take on more of the responsibility he was carrying.

Health is the obvious issue, and I didn't like the CHI contract at the time, but Havlat has stepped out of the shadow of Alfredsson and Hossa and really delivered when healthy. It's tough to argue against that.

The arrows all seem to be pointing in the right direction, he's a guy that I hope Lowe takes a flier on. I mean, he's a cap team type of player ... he may very well go on long term IR, and that hurts the budget, but not the cap.

I don't think that I've ever heard Lowe or the other Oilers brass mention Havlat though, and they have enough guys off with serious shoulder problems that they're probably gunshy. Still, even with just a bit of evidence at the moment, he looks like a good bet to me.

3/27/2008 11:26 am  
Blogger Vic Ferrari said...

YK

Ah, I hadn't realized that the cap was so tight to the. I guess the only way to make space would be to move Pitkanen for someone cheaper? And the thought of the D without Pitkanen would be a little scary to me.

Or would moving Stoll and Torres (or similar) be enough?

Odd that we both threw in the term ripple effect at the same time, btw. And three years ago if you said we'd both be giving havlat props for that in 2008 ... I would have thought that you were off your nut.

All players grow up eventually, I guess.

3/27/2008 12:02 pm  
Blogger YKOil said...

Thanks for the prompt Vic.

:-)

I took the time to finally post some analysis on the Oilers salary sitch going forward. Way to long to post in a reply though so you will have to head on over for the read. I left some holes in there (mostly unexplained stuff) but hey, it's 3:30 am and I am starting to fade. Fast.

As for Havlat. As impressed as I am surprised. Not all players who agitate for lead dog status actually raise their game to the same level (lead dog that is).

Also, what we hope Nilsson could be. Fewer goals and more assists but basically the same end game - a difference maker with or without the complementary players.

3/28/2008 3:40 am  
Blogger voxel said...

From watching the Hawks this year, I suspect it is Toews, Williams, and Sharp who are the difference makers. The sample size is extremely small...

With Toews:

32-23-6 (61GP)
GF: 190 (GF/G = 3.11)
GA: 177 (GA/G = 2.9)

Without Toews:

6-10-2 (18GP)
GF: 34 (GF/G = 1.88)
GA: 47 (GA/G = 2.61)

Kane's production took a nosedive without Toews. Note: Jason Williams was injured around the same dates as Toews (and longer). Havlat has been out for a good month or two, but the Hawks have been winning without him - 8-6-2 against tough teams (only STL could be considered cream puff) AND with Khabibulin out for half those games.

Rookies can make a difference. Toews is that special rookie.

4/01/2008 12:04 am  
Blogger Matt said...

Vic, I've been mulling the reverse ripple effect w.r.t. Eric Godard, i.e. the whole idea that you can only dress so many forwards you need to shelter, and not only does Godard take a lineup spot from a (likely) better player, but it results in someone who IS dressed perhaps playing more shifts in tougher situations than they ought to be.

Anyway, I ran this stuff for the Flames with and without Godard.

GODARD PLAYS (72 games): Fenwick +45 (+0.6/gm), Corsi + 212 (+2.9/gm)

GODARD SITS (8 games): Fenwick +49 (+6.1/gm), Corsi +100 (+12.5/gm)

The "SITS" sample is pretty small, so I don't want to get *too* worked up about the appearance that the Flames are much better with Big Eric in the PB, but yeesh. Anyway, there's another angle for you on this whole thing.

And at the very least (without having dug too deeply), I think there has to be such a thing as a player who is a negative difference maker: who drives results, but in the wrong direction.

4/03/2008 11:33 am  
Blogger PunjabiOil said...

Time to update the results from the contest trivia you had back in September (over/Under)

I checked my prediction. 6-0. Still could lose depending on tiebreaker.

Cheers

4/05/2008 12:09 am  
Blogger PunjabiOil said...

Actually, Fuck that. I updated the contest results myself.

Vic Ferrari 4-2
Metrognome 6-0 if Minnesota does not get a point on Sunday. 4 point difference tie breaker
Gnash 4-2
Matt 5-1
Black Dog - 6-0 if Minnesota does not get a point. 76 point difference tie breaker
Jonathan 4-2
Julian 5-1 if Minnesota does not get a point on Sunday.
Mattm 4-2
godot10 6-0 if Minnesota does not get a point on sunday. 2 point difference tie breaker
Jon 6-0, 10 point difference in tiebreaker
Dawgbone 4-2
Kurri_17 5-1
Hbomb 5-1
PDO 4-2
PunjabiOil 6-0, 7 point difference tie breaker
Doogie 4-2
gret99zky 4-2
showerhead 3-3 if Minnesota does not get a point on sunday
dd 5-2 if Minnesota does not get a point on Sunday
Rob 5-1 if Minnesota does not get a point on Sunday
Loxy 4-2
riversq 5-1 if Minnesota does not get a point on Sunday
Steve 4-2

CONTENDERS (all with 6-0 records):
Metrgnome, Black Dog, PunjabiOIl, godot10, Jon

If Minnesota gets a point on Sunday, godot10 wins. Otherwise, PunjabiOil wins.

4/05/2008 12:31 am  
Blogger PunjabiOil said...

Original Thread can be viewed here

http://vhockey.blogspot.com/2007_09_01_archive.html

4/05/2008 12:33 am  
Blogger PunjabiOil said...

Minnesota did get a point.

I win.

< /Endthread>

4/07/2008 2:00 am  
Blogger Vic Ferrari said...

Matt:

Yeah, terrific point. Mirtle had a post up early in the season regarding enforcers. Their numbers, collectively, were brutal. And individually their Fenwick numbers were almost all in the red, maybe one or two exceptions. Surely that caught up to all of them as the season wore on.

And we all know that these guys rarely see an own zone draw unless it's against the other team's 4th line, and they rarely see the ice against quality opposition in any game whose result is still up in the air.

Still, I really believe in the value of a tough guy, a guy that the other team's tough guys aren't interested in tangling with.

I remember a game that Hemsky checked Reed Low in the back, nearly knocked him into the players bench iirc.

Laraque went over and had a chat with Low the shift after. He probably told him that Hemmer is a clean player, and that he didn't recognize the number on Low's back. I dunno.

In any case Reed rarely scored a point in the NHL, he is probably a good guy, and is as tough as nails, and he wouldn't be afraid of getting punched by Georges, but he would be afraid of getting embarrassed by him. Because like a lot of these guys, he had a tenuous hold on a roster spot, the Blues have an open Greyhound ticket to Peoria waiting in the office, and AHL money ain't NHL money.

If you're a Blues player, and you run Hemsky later in the game ... Reed will have to fight Georges. And God help you at the next practice, because I kind of doubt that Low will have seen the humour in it. My advice would to just ignore the banter and pretend that those puck whistling past your head and into the back of your legs were accidental.

Plus, come playoff time, Godard isn't going to play anyways.

It also highlights the value of enforcers who can actually play a bit, like Brashear, Simon and Neil. Probably more bang for your buck from those guys than most NHLers.

4/08/2008 11:09 am  
Blogger Eric said...

There are several errors in the data.

Some of the games don't match up actual scores (specifically a few of the OT & SO games).

The data needs to be double checked.

4/11/2008 6:49 pm  

Post a Comment

<< Home