Faceoffs Are Not That Important: Laperriere JUST what the Doctor Ordered

I completely agree — Malhotra would be a great signing. I hope we get him, he definitely would not hurt in any way.

it was a fair amount of work, so i appreciate the high-five

there are general issues with the statistics I used, which is why I went to such large sample sizes — it’s true, that in the course of the season, Malhotra (etc) would take more high-leverage FO’s than Powe’s. But given the huge # of samples (400 is a lot), that basically cancels out noise such as “Powe goes against worse FO guys than Malhotra.” It’s true, but over 400 FOs that will essentially even out (and there’s statistical evidence regarding sample size to support my argument that greater size accounts for the extremes).

The fact that Malhotra wins ~58% of the time is a statement of fact. It has a big enough sample size that you can’t say “Malhotra wins 80% of the time against the average centerman, but because he’s always going against other FO specialists that percentage falls” or the reverse, “Powe wins 48% of the time against average centermen, but far less often against Malhotras.” The statistics of 58% and 48% take into account the extremes, and while I’m sure Powe’s #s would fall a bit if he only faced Malhotra in the FO circle, it wouldn’t be by that much, probably 1 – 2 faceoffs a week difference.

Again, none of this is to say I don’t want a Malhotra type. If you want the best chance to win, may as well get it in all facets of the game, high-pressure FO scenarios included. Really, those are so rare that it doesn’t truly matter who you put in — you’ll probably win ~50% of the time regardless of who your centerman is with 2 minutes left in a 1 goal game. Give or take 10%. Considering the # of times this happens over the course of the season and matters (i.e. you win possession and then force OT/win/get more points than 0, or the other team gains possession and gets more than 0 points), the difference between a Malhotra & a Powe in those scenarios (of which there are probably no more than 10 per season, at the most) will probably mean 1-2 games in the standings at best. Is that enough to pay for 1 guy the price of 2 guys who are equivalent except in FOs??? I don’t know. Depends on how much cap space you have, I suppose. If you do, then sure — why not? If it’s between spending the money there and fulfilling a need elsewhere, the consummate effect of having 2 Darroll Powes @ ~800k vs. having 1 Malhotra at ~1.8m swings vastly in the favor of having 2 Powes.

So to summarize my defense, certainly faceoffs are not meaningless, and my stats (as stated) are imperfect at best. But, that said, they’re good enough to draw conclusions, and they do address the points you’ve raised by virtue of being large in sample size (if you don’t believe in the power of large samples (and I’m not saying you don’t! I don’t mean to imply that), there is plenty of literature on the subject, which I encourage anyone and everyone to read). It’s not disingenuous, it’s fact. Malhotra would be great! I’d love him to death — his defensive prowess and FO ability would absolutely help the team. Given our cap constraints, I’m not sure that spending on a FO specialist would be the best use of money, but given Malhotra’s defensive prowess it would basically be like adding another defenseman — which would be great. Just like adding Lappy was.

Your opinion basically seems to be that I’m calling faceoffs unimportant. I have not said this. Additionally, the fact that all faceoffs are not created equal is taken care of by that large sample size — that’s what I meant by saying the sample size shpiel. Of course they’re not all equal, but if you have 400 or so, it balances out.

Additionally, my next line after the one you quoted was:

Is that enough to pay for 1 guy the price of 2 guys who are equivalent except in FOs??? I don’t know. Depends on how much cap space you have, I suppose. If you do, then sure — why not? If it’s between spending the money there and fulfilling a need elsewhere, the consummate effect of having 2 Darroll Powes @ ~800k vs. having 1 Malhotra at ~1.8m swings vastly in the favor of having 2 Powes.

Suffice it to say, I specifically say that if you have enough cap space, “then sure — why not?”

However, I stand by my argument that I would rather have 2 Powes than 1 Malhotra (remember, these are just names representing statistics, not necessarily referring to the actual players), because the upswing of having 2 Darrell Powes wins-wise will be greater than the upswing of having 1 Malhotra. That’s the crux of my argument.

Also, the plural of anecdote is not data. There are guaranteed to be situations such as the one you are saying, and I have already said that:

Again, none of this is to say I don’t want a Malhotra type. If you want the best chance to win, may as well get it in all facets of the game, high-pressure FO scenarios included.

That said, if the push is to sign a $2m FO specialist when it comes at the expense of having 2 almost-as-good players, the sum of whom is greater than the 1 $2m FO specialist, call me crazy, but I’d gladly take the 2 Powes over the 1 Malhotra (to use our Player X / Player Y effective names).

When I say “Faceoffs are not that important,” I say that as in I’d rather have the 2 “eh” faceoff guys than the 1 “woah” faceoff guy. The summation of the additions of the non-FO skills of 2 guys would more than make up for the 1 guy who is good at FOs. 2 Powes > 1 Malhotra. That’s all.

It’s true, 15 seconds in a 5 on 3 is crucial. Is that worth $800k? I don’t know. If you have the 2 Powes instead of the 1 Malhotra, maybe you’re not even in a position to be in a 5 on 3 situation 1 goal up (or down) with less than 5 minutes. We’re talking in pure possibilities here, which is enjoyable and fruitful for thinking things through, but is not as useful when it comes to actually trying to critically, empirically, and objectively analyze data. The theoretics help us think up the questions — we have the questions at hand — and that’s very useful. Now, the data is answering the questions for us: All things considered, over the course of a season, having 1 guy who is exceptional at faceoffs at $2m is not as useful as having 2 guys almost as good but significantly worse in the FO department. Perhaps there are 1-2 games you will lose by virtue of having a 53% FO guy in the 5 on 3 disadvantage rather than the 58% FO guy (and, again, the probability of any single average NHL centerman winning an FO is ~50%, with some players beating luck by a pinch and others suffering at their talent’s expense — Malhotra is better than luck, Powe slightly worse, but Powe would never center a 5-3 disadvantage anyway, so that point is moot to begin with), but that’s almost beside the point. You give up those 1-2 games, because by having those 2 Powes instead of that 1 Malhotra will give you 2-3 games total, for a net gain. We’re being real inaccurate with these win estimations, yes, but the basic point is true. The farther above replacement-level player your net talent is, the more games you will win. Thus, 2 Powes has a more significant net effect than 1 Malhotra, at the same price. When our cap is constrained the way it is, that means the most efficient way to maximize wins (and not have to deal with that 4th-to-7th precipice) is via suffering a lost FO here and there for the sake of scoring an extra 10 goals and preventing an extra 5 on the whole. The stuff regarding 1 FO every 5 games is merely to provide some perspective — that 1 FO could be in game 7 of the Stanley Cup, up by 1 goal with 15 seconds left at a 2 man disadvantage… but it’s not too likely that it will be. Should we prepare for the scenario? Of course. If we can, but if we’re looking to maximize wins and we don’t have that much cap space, and it came down to signing 2 Powes or 1 Malhotra and having to go to war with that, give me the Powes. That gives us the best chance of having a 2 goal lead with 15 seconds in game 7, and that is what I would prefer. That’s all I’m saying, and hopefully we can all agree with that.

And the sample size issue takes care of that, essentially. Yes, there would be some differences still to be taken into account (as I said), but the bigger the sample size, the less likely it is that those 20 extra FO wins will be likely to come at crucial times. That’s the nature of sample size: the bigger it is, the less individual times are important.

That was a typo — I got the numbers from a site, and checking back, it seems that they’ve changed their numbers — added a 1 before each. Doh. The 1 per game argument, however, I would say, is still of moderate importance — more than I believed, yes, but I still believe 2 Powes > 1 Malhotra. To see if points & FOs are correlated, one should do a regression — doing so, the stats show that FOs & points don’t have a huge correlation. They can be important, as you suggest, in the blue zone, but overall FO wins is not really related to team wins & team points scored/given up.

Digging around more, I just found this site (knowing behindthenet, I should have looked there sooner — another mistake on my part), and it really does an admirable job summing up my own arguments (and coincides with what I expected):

behind the net’s faceoff vs. wins relationship

Overall, that means that one out of every 40 lost face-offs resulted in a goal in the next 20 seconds. In a league where teams score just 2.5 goals per game, that makes a significant difference.

He continues:

The most significant result of this analysis is that teams should use their best face-off men on face-offs deep in their own end to decrease the likelihood of being scored on. Similarly, they should also use their best face-offs takers in the offensive zone. If a team improves its face-off winning percentage in these situations from 50% to 60% (say, by signing Yanic Perreault, assuming all other things are equal), it can expect, on average, to improve its goal differential by 25 goals over the course of the season. In today’s NHL, this translates into an additional three or four wins, which is obviously an advantage any team would take.

This, I trust. I don’t think it invalidates my earlier arguments — 50% guy vs. 58% guy will still win 48% of the draws, for example, and that means just a few faceoffs (taking into account your corrected #s) over the course of a week. It also means that FO specialists can improve our GS amount by 25 goals per year — this is great, and makes Malhotra more valuable. I would still prefer 2 Powes (if we’re continuing to use Powe as a proxy for Malhotra minus his FO gifts), and I find it hard to believe you would disagree with that. Remember, as I’ve said all along, Powe is a proxy for Malhotra without his FO gifts, just as Malhotra is a proxy for a defensive guru who is also good at FO’s. Whether or not Powe is a good proxy for Malhotra in all but the FO dept is a completely different argument — obviously, they’re two very different players (Powe is faster, but not as talented defensively or offensively, although I don’t think he’s really AHL fodder as you suggest, but again, that’s a separate discussion).

I’ll leave my diary the way it is so that people could follow the progression & see what you’re referring to in your #s correction.

I agree with you — we can agree to disagree ;)

I completely agree, as well, that 2 wins is a gigantic swing — I think the only real difference in our discussion is that I think that 2 Powes (etc) means more Ws than 1 Malhotra. Perhaps I’m wrong — the #s I used turned out to be typos, after all! But thanks for the invigorating discussion and excellent eyes on the FO # catch (I’m not sure how I missed that myself — I guess I saw the number and assumed it was right, even though in retrospect it should have been obvious that it was wrong).

I bet they will get better with time, though.

I think you haven’t read all of my comments, and my rebuttal to your statement is in them.

you’ll find my arguments and a good discussiona bout them between myself and MarioD. That should give you a two-sided view on the importance of faceoffs.

Also, just to emphasize, I never say “Not-important,” I just say they’re a bit overrated. Still important, just not as important given certain circumstances I’m sure you’ll find if you read the entire article & discussion.