NHL hockey unequal point values

Rita · March 06, 2009, 09:37:58 AM

I was wondering if any of the math geeks had done any calculations on how the OTs/SO in the NHL makes some games worth 3 points and what effect that has had on teams making the playoffs, or not. I never had enough free time to go through the seasons to figure it out, but that might change in about 2 months!

I'm guessing not much difference, since they haven't changed it and you don't hear a lot of complaining by the teams. But then again, this is a league run by Gary Bettman.

It just seems so wrong that a team that loses a game still gets a point while the winning team gets 2. I'd be okay with dividing the points up 0.5/1.5 or making all games worth 3 points such that if you won in regulation you got all 3, and OT/SO games were worth 1 pt to the loser and 2 pts to the winner.

This is something that bothers me, especially as the playoffs draw closer and I'm hoping that the Carolinas and Floridas don't get in but the Buffalos of the NHL world do.

jtwcornell91 · March 06, 2009, 09:56:13 AM

Unfortunately international soccer has used a non-zero-sum point system for years (3 points for a win, 1 for a tie); it means two teams are both better off splitting a pair of games than tying both. I guess for hockey the best plan is for two teams to split a pair of games, with each team winning in OT/SO.

International hockey actually uses a zero-sum version of the NHL madness (you get 3 points for winning in regulation), whose most annoying feature is considering a decision in overtime equivalent to a shootout.

Trotsky · March 06, 2009, 10:01:42 AM

There are only two acceptable ways of doing this:

1. The right way: ties, no shootout and either no overtime or no points for an overtime loss

2. The barely tolerable wrong way: all wins the same value, all losses 0 points, break a tie by infinite overtime, shootout, fashion show...

DeltaOne81 · March 06, 2009, 11:17:40 AM

Well, 2 will never happen, and 1 won't either (again).

Given the reality of the NHL that there will be shootouts, I do agree with Rita that the best situation is probably:
3 points for a regulation W
2 points for a OT/SO W
1 point for an OT/SO L
0 points for a regulation L

(note: if you want to move OT W/L along with regulation instead of with SO, that's fine)

I haven't done the math (nor will I), but I just wanted to add that I heard a radio color person recently refer, derisively, to "loser points". A term I enjoy and wish to spread around.

I'd also like to add the flip side "fake winner points", for wins occurring in a SO, but that's not quite as catchy.

Okay, I lied, one quick bit of "math" (if it even qualifies as that): a quick peak at the NHL standings shows that the order of points is pretty similar to the order of wins. And there are no gross deviations (a team with several less wins being ahead due to a lot of OTLs). But that doesn't give any insight into the effect of SOWs, which aren't listed separately.

Trotsky · March 06, 2009, 11:26:22 AM

I love the metric "Loser Points" (LP), where:

LP = OTL + SOW

That simplifies the standings, which can then be

W, L, T, Pts, LP, PLP

with Post-Loser Points (PLP),

PLP = Pts + LP

The question then is whether there's any difference in the standings between Pts and PLP.

My advice on shoot-outs, after throwing up, is that the compromise between hockey fans who loathe it and casual hockey fans and infants who love it is to have it count as a tie-breaker. So, a game ending in overtime is a tie. The team winning a shootout gets 1 SOW, which counts for no points. If teams are tied in points at the end of the year, the first tie-breaker is SOW.

jtwcornell91 · March 06, 2009, 11:40:16 AM

[quote DeltaOne81]
Given the reality of the NHL that there will be shootouts, I do agree with Rita that the best situation is probably:
3 points for a regulation W
2 points for a OT/SO W
1 point for an OT/SO L
0 points for a regulation L
[/quote]

This is exactly the International/European system.

Note that Hockey East experimented with shootouts back in the 90s, and the point system they used was

5 points for a regulation/OT W
3 points for a SO W
2 point for an SO L
0 points for a regulation/OT L

which seems fine to me, and also makes shootouts worth less compared to games. (The results are not far from Greg's "tiebreaker" scenario.)

The other advantage to a plan like this is for a college hockey conference is that you could actually do weekend playoff series without infinite overtime, as first-to-eight-points series. (So you could win the series with two wins, a win and a SOW, a win and two SOL, or two SOW and a SOL. This is equivalent to an first-to-four-points series in the "standard" points scheme, where WLT and TTT tied series are settled by one or three shootouts, respectively...)

RichH · March 06, 2009, 12:29:09 PM

[quote jtwcornell91]
The other advantage to a plan like this is for a college hockey conference is that you could actually do weekend playoff series without infinite overtime[/quote]

Yeah, but infinite overtime can be pretty awesome.

Josh '99 · March 06, 2009, 12:46:41 PM

[quote Rita]I was wondering if any of the math geeks had done any calculations on how the OTs/SO in the NHL makes some games worth 3 points and what effect that has had on teams making the playoffs, or not. I never had enough free time to go through the seasons to figure it out, but that might change in about 2 months![/quote]My gut feeling is that it must work out basically the same in the long run, but that doesn't make the system not stupid.

KeithK · March 06, 2009, 01:30:27 PM

[quote jtwcornell91]
The other advantage to a plan like this is for a college hockey conference is that you could actually do weekend playoff series without infinite overtime,[/quote]
I'm sorry but this is not an advantage. Playoff games are not supposed to end in ties or artificial constructs.

The HE system is the best if you feel the need to award points for ties. Zero sum, minimal advantage for the SOW. Greg's suggestion is even better. Of course, just allowing ties (what's the frickin' big deal?) is by far the best option.

RichH · March 06, 2009, 02:03:39 PM

[quote KeithK][quote jtwcornell91]
The other advantage to a plan like this is for a college hockey conference is that you could actually do weekend playoff series without infinite overtime,[/quote]
I'm sorry but this is not an advantage. Playoff games are not supposed to end in ties or artificial constructs.

The HE system is the best if you feel the need to award points for ties. Zero sum, minimal advantage for the SOW. Greg's suggestion is even better. Of course, just allowing ties (what's the frickin' big deal?) is by far the best option.[/quote]

Hmm. Your ideas are intriguing to me and I wish to subscribe to your newsletter.

Josh '99 · March 06, 2009, 02:47:30 PM

[quote KeithK][quote jtwcornell91]
The other advantage to a plan like this is for a college hockey conference is that you could actually do weekend playoff series without infinite overtime,[/quote]
I'm sorry but this is not an advantage. Playoff games are not supposed to end in ties or artificial constructs.[/quote]*applause*

Jim Hyla · March 06, 2009, 04:13:21 PM

[quote KeithK][quote jtwcornell91]
The other advantage to a plan like this is for a college hockey conference is that you could actually do weekend playoff series without infinite overtime,[/quote]
I'm sorry but this is not an advantage. Playoff games are not supposed to end in ties or artificial constructs.[/quote]Why not, playoffs with first to 3 points can be very exciting, especially when the second game is tied.

KeithK · March 06, 2009, 05:18:57 PM

[quote Jim Hyla][quote KeithK][quote jtwcornell91]
The other advantage to a plan like this is for a college hockey conference is that you could actually do weekend playoff series without infinite overtime,[/quote]
I'm sorry but this is not an advantage. Playoff games are not supposed to end in ties or artificial constructs.[/quote]Why not, playoffs with first to 3 points can be very exciting, especially when the second game is tied.[/quote]
It's exciting when a team that won the first night can play for a tie? Maybe we should just go back to two game total goals series.

DeltaOne81 · March 06, 2009, 07:16:56 PM

[quote Trotsky]I love the metric "Loser Points" (LP), where:

LP = OTL + SOW[/quote]

Well, except, that's a misnomer. Because a SOW is not a loss, even by traditional hockey standards, by which it would have been a tie. You could call them 'non-winner points' but that's not nearly as catchy or degrading.

What about 'fake points' & 'phony wins'? Such that 'fake points' = 'loser points' (OTL) + 'phony wins' (SOW)

Rita · March 06, 2009, 07:21:44 PM

For regular season games, I don't see anything wrong with a tie. Why does everything these days have to have a "winner"?

As long as the NHL is going to resolve all ties, why don't they go right to the shoot out? There isn't any difference (at least with the current point system) between winning in OT v. winning in the SO.

Yeah, 4 v.4 provides more open ice and presumably more goal scoring opportunities, but with the way the call penalties, there are plenty those in regulation time.

I get the loser point criticism, but if the NHL is insistent in its desire to decide tie games by the stupid (IMHO) shootout method, I don't mind the losing team getting a share of the points as long the total amounts given out for a game do not exceed what a regulation win is worth.