2009-03-16

Another Day, Another Bettman Point


Ooohhh!! A point! Big whoop! I know exactly how you feel, Gary.

In fact there are too damn many of your points floating around these days. In the last week the Oilers have accrued 3 of them, in consecutive overtime losses to Montreal, Atlanta and Colorado. But that's just the tip of the iceberg; in those same 5 days Tuesday-Saturday, fully half -- 21 of 42 -- NHL games awarded the bogus, er bonus point that gets awarded jointly to any two teams which can't decide a game in regulation. It's the most cockamamie system imaginable.

Of course it’s in a team’s best interest to win their games in regulation, but a club prone to playing a lot of close games – which is the vast majority of them – is better served by playing conservatively and going for overtime. A late game-deciding goal has far greater negative impact on the team allowing it than positive for the team scoring it. Assuming that Bettman points are split 50/50, which of course they are on a league-wide basis, every regulation tie is worth an average of 1.5 points. The law of averages dictates that the late GA means the loss of those 1.5 points, the late GF gains just 0.5. Might as well wait for overtime and go for it then, with your 1 point assured if you fail but full value for a victory given if you succeed.

This system does nothing less than compromise the competitive integrity of the game. If that sounds like a serious charge, it is. When two teams and their coaches reach an in-game situation where the score on the board serves both their interests for the time being, the system is broken.

Gary Bettman and his cronies have failed in their fiduciary capacity as stewards of the game, and the shambles that their gimmicks have made of the standings, the record books, and the very games being played in front of their paying customers cannot be undone.

***

Interesting to look at the cumulative standings of all NHL teams over the past ten seasons.

Season * GP * W * L * T * OTL = Pts. Pct.
-----------------------------------------
1998-99 2214 945-945-324- 00 = 2214 .500

* Introduction of Bettman Point Ver 1.0 *

1999-00 2296 1002-888-292-114 = 2425 .528
2000-01 2460 1078-956-304-122 = 2582 .525
2001-02 2460 1081-960-298-121 = 2581 .525
2002-03 2460 1073-918-314-155 = 2615 .532
2003-04 2460 1060-915-340-145 = 2605 .529

* Introduction of Bettman Point Ver 2.0 *

2005-06 2460 1230-949- 0 -281 = 2741 .557
2006-07 2460 1230-949- 0 -281 = 2741 .557
2007-08 2460 1230-958- 0 -272 = 2732 .555
2008-09 2060 1030-792- 0 -238 = 2298 .558
(2008-09 through Saturday, March 14)

For perspective it begins with the last year of standings sanity, where all games were worth the same two points and teams by definition played .500 hockey against themselves. But in 1999 that changed with the introduction of the Bettman point, awarded to teams that lost games in overtime (or to teams that won them in OT, depending on your interpretation). The idea, or so it was explained at the time, was that teams were playing too conservatively in overtime, hanging on to their one point. A win for the winner, a tie for the loser will put an end to that.

Overtime indeed became more wide-open; the percentage of games actually decided in OT doubled from 24.7% in 1997-98 to 49.7% in 2002-03. How much of that was due to the Bettman Point and how much to the 4-on-4 format that was introduced at the same time cannot be disentangled. The heavy cost, however, was that instead of hanging on in overtime for their one point, coaches started to do so in regulation, often for entire third periods at a time or even longer. And given the “new math” of Gary Bettman’s NHL, frequently both teams would be doing so simultaneously, cuz it was in both their interests. Coaches ain't stupid, and most of them passed Grade 6 arithmetic.

After the lockout – another blight on the historical and statistical continuity of the sport, and the second such which happened on Gary Bettman’s watch – the powers-that-be went a step further and brought in the shootout, thus ensuring that the third point would now be awarded in every game that reached overtime, not just the half or so of them that resulted in an overtime goal. Thus the value of a regulation tie soared from 1 point pre-1983, to an average of 1.00 points during the 15 years that decided overtimes split the points 2-0 and undecided OTs 1-1, to almost 1.25 points during the first variation of the Bettman Point, all the way to 1.50 points in the shootout era.

The results can be seen in the blocks above, which show the league winning percentage jumping in two discrete steps from its natural .500 (1917-99) to about .528 (1999-2004) to about .557 (2005-2009). Key word: “about”, as the median, once a reliable constant, now fluctuates from season to season, indeed from day to day. In past seasons I have tracked this median figure and observed a general upward slope; as the playoffs approach and the points become more dear, OT games become more common. Last year for example, I divided the season into 5 * 246-game segments:

# 3-pt. games * Segment % * YTD % * Mean Pts%
---------------------------------------------
1. * * 39 * * * 15.9% * * * 15.9% * .5396
2. * * 49 * * * 19.9% * * * 17.9% * .5447
3. * * 59 * * * 23.9% * * * 19.9% * .5498
4. * * 65 * * * 26.4% * * * 21.5% * .5539
5. * * 60 * * * 24.4% * * * 22.1% * .5553

While Bettman & Co. delivered on one part of their promise, no more tie results, in reality there are more tie games than ever.

Season * % Ties * % Regulation Ties
-----------------------------------
1917-18 * * 0 % * unknown
1927-28 * 16.8% * unknown
1937-38 * 18.8% * unknown

1947-48 * 18.9% * 18.9%
1957-58 * 16.2% * 16.2%
1967-68 * 17.1% * 17.1%
1977-78 * 18.3% * 18.3%

1987-88 * 11.5% * 17.4%
1997-98 * 15.5% * 20.5%
1998-99 * 14.6% * 20.1%

1999-00 * 12.7% * 22.6%
2000-01 * 12.4% * 22.3%
2001-02 * 12.1% * 22.0%
2002-03 * 12.8% * 25.4%
2003-04 * 13.8% * 25.6%

2005-06 * ----- * 22.8%
2006-07 * ----- * 22.8%
2007-08 * ----- * 22.1%
2008-09 * ----- * 23.1%

The above list takes a sample once per decade, and annually during the Bettman Point era. Pre-1942 I don’t have a reliable source of info on OT results, so don't have a handle on what percentage of games were tied after regulation. Following the elimination of regular season OT due to wartime travel restrictions right through 1983 all games were 60 minutes and a tie was a tie, splitting the (two) points. Roughly 18% of all games ended in ties.

After 1983 the 5-minute overtime started to prune the number of tie results down, even as the number of games tied after 60 rose to about 20% during the dead puck era. But in 1999, the Bettman point was introduced and the number of regulation ties surged to new highs. It has remained high ever since, with the last 9 seasons yielding the 9 highest percentages of the entire sample, with 22-26% of games tied through 60 minutes every season.

***

That extended rant aside, the rules are the rules, and the teams must play within them. How do they apply to the Oilers?

Recent results suggest that the Bettman Point (a.k.a. the “loser point”) has saved the Oilers from what could/should have been a major pratfall in the standings. 1-1-4 = 6 points is much less painful than 1-5, 2 points would be. Getting to overtime has saved our bacon.

It’s what we have done when we get there that is disturbing. The play-for-OT strategy only really works if you can win at least your share of overtime sessions and/or shootouts. I would argue that it’s the “winner point” that is the true Bettman Point, the two points for regulation having been split in the traditional manner of a tie and now only the third, free lunch point that’s up for grabs and awarded to the winner of the mini-game. It’s here where the Oilers have been an epic fail, playing under 11 minutes of overtime the last two weeks and being outscored 4-0. Four goals against in half a period. Yuck.

The Oilers have been outshot in each one of those overtimes and cumaltively by a 9-2 count, so clearly something is amiss. One can start with goaltending: 4 GA on 9 shots is just brutal. Makes one wonder if tired goalie syndrome is likely to manifest itself at the end of a long, tense night. This past week Roli posted a Sv% of .902 in regulation, but only .500 in OT over the three games which included two decidedly weak goals through the 5-hole. Ugh-ly. That said, a team which took two penalties in overtime and couldn't kill either one, allowed one clear breakaway in the dying seconds, and which failed to backcheck effectively (if at all) in their most recent defeat, can't just lay it all on the goaltender.

So the Oilers crawl up the standings, one painful point at a time. Across the league, teams earn on average something over 1.11 points per GP in the Second Bettman Point Era, an average which surged to 1.25 during this recent outbreak of OT affairs. So 1 point a night is the equivalent of hanging on by our fingernails but gradually sliding away. We gotta win some games.

11 comments:

Scott Reynolds said...

Good work Bruce. Did you look at all of the overtime results in the early years or just those you listed in the sample?

As for the Oilers... yeah... they need to win some games.

Jonathan Willis said...

The Natioanl Society of Words Not Commonly Used (NSWNCU) thanks you for "fiduciary".

And I thank you for an excellent post!

David S said...

Something inside tells me that MacT is reverting to a strategy he's used before, that of actively playing for OT to get that 1.5 points average. Problem is, his 4X4 guys have completely let the team down. This is astounding to me considering the talent we seemingly have at our disposal - to the point where I've actually been salivating at the prospect of 4X4. Obviously that's changed.

However, I'd also have to say you're probably right when it comes to Roli being burnt out by the end of the game (thus being unable to hold it together for OT). You'd think that would be the cue to bring in JDD for a few.

Bruce said...

Did you look at all of the overtime results in the early years or just those you listed in the sample?

Scott: Originally I sampled once per decade from 1917-18, but for the first quarter century I know of no source which tabulates how many games were won and lost in overtime, just how many remained tied and therefore appeared in the standings. Incomplete information is still information, however, so this morning I have revised that middle section of the post if you are interested.

Something inside tells me that MacT is reverting to a strategy he's used before, that of actively playing for OT to get that 1.5 points average.

More like "passively playing for OT", David, but yeah, my game theory antennae reckon it's best to play a lot of overtimes. The actuarial tables are all in your favour if you do.

You still got to win the mini-games, though. At least some of them.

I include 4v4 as another type of special teams play, an area where the Oilers have just been bad all over this year. We've certainly reeked in OT be it 4v4 or 3v4 and have the record to prove it.

Scott Reynolds said...

Thanks for the heads-up Bruce. I was actually wondering if you had looked at every single season or if you just looked at those that you sampled. Basically, I'm curious if you think the years are all representative of the decade. The trend is obvious so I'm not trying to take away from your overall point, just curious if some decades had more ties than others before the Bettman point.

Bruce said...

Scott: That's on the "to do" list. From previous dabblings into this sort of thing years ago, I believe there may be one or two outlier seasons with a lot of ties which are not included in my sample, but otherwsie it's fairly representative I think.

Scott Reynolds said...

Thanks Bruce. I guess I'm mainly prodding to see if it's the lower scoring seasons that have more ties since that's what I would expect. Again, not to diminish the effect of the Bettman point which is clearly increasing the number of tie games in regulation. And I do agree that it is somewhat farcical that you have two ostensibly competing teams with complementary goals.

Bruce said...

Scott: Funny you should ask that. Back when I first did this research a couple years ago, I had the same question. In fact for "make it fit" reasons I suppressed a column from my spreadsheet which showed goals per team per game. But since you asked, it is instantly :) available:

Season: * % Reg.Ties * GPG
1917-18 * * ? ? ? * * 4.75
1927-28 * * ? ? ? * * 1.90
1937-38 * * ? ? ? * * 2.53

1947-48 * * 18.9% * * 2.93
1957-58 * * 16.2% * * 2.80
1967-68 * * 17.1% * * 2.79
1977-78 * * 18.3% * * 3.30

1987-88 * * 17.4% * * 3.71
1997-98 * * 20.5% * * 2.64
1998-99 * * 20.1% * * 2.63

1999-00 * * 22.6% * * 2.75
2000-01 * * 22.3% * * 2.76
2001-02 * * 22.0% * * 2.62
2002-03 * * 25.4% * * 2.65
2003-04 * * 25.6% * * 2.57

2005-06 * * 22.8% * * 3.08
2006-07 * * 22.8% * * 2.95
2007-08 * * 22.1% * * 2.78
2008-09 * * 23.1% * * 2.91

We are still resttricted to the sampled seasons (I may have to return to that soon), but a couple of trends are apparent. One is way more ties in the Bettman Point Era(s) than at any other time. The other is that the number of reg. ties appears to have peaked pre-lockout. This despite the fact that the disincentive to go for the win in regulation time is stronger than ever now, cuz now you have not one but two additional chances to earn your second point in "safe time", either overtime or the shootout. In fact it is guaranteed that one of the two teams will get it, so in its Schrodinger's Cat phase it is (0 + 1)/2, which is to say 0.5 points per team as we've already discussed.

In the old days the reward of going for the win equalled the risk of taking the loss: a late winner was your only chance to get the two points, and essentially you would take that point from the other team. Nowadays, simply by going to overtime the two teams create, out of the ether that passes for Gary Bettman's brain, a third point which one of them takes from the whole rest of the league, not just that opponent but 28 innocent bystanders. Did I mention it's idiotic?

Mathematically, under the old-fashioned tie rules, a late GWGF was +1 and GWGA -1. Under Bettman1, those numbers changed to about +0.77 and -1.23. Under Bettman2 of course, they are +0.5 and -1.5. To allow a late regulation winning goal is, in essence, triple liability. You have to be insane to take any chances in that environment.

Anyway, two things have changed post-lockout that "should" reduce the number of regulation ties, namely higher scoring games and a change in league policy w.r.t. late-game powerplays, which used to be verboten, esp. in already-tied games. A team with a powerplay will actually try to score in a tie game (ignoring for the moment Oilers' powerplay in Montreal in the dying minutes last week). Whereas teams at even strength continue to look after their own interests, which can be summarized thusly: defend, defend, defend, now let's think about attacking. The points system dictates it must work that way. The logical result is you get 21 tie games in 5 days, with "exciting finishes" derived at the considerable expense of deadly boring stretches of regulation play. The third period of the Atlanta game (where the score remained tied 3-3) and the second of the Colorado game (where it remained tied 1-1) were two of the worst periods of hockey I've seen all year; I'm sure glad I didn't pay $120 or whatever the heck it is to watch it, not to mention $7.75 a beer to make the pain go away.

The system is broken.

Ribs said...

Why not just get rid of the tie point altogether? Even before the shoot-out system was implemented it would have created better hockey.

A tie gets no points. Play 5 in OT for what it's meant to be; extra time to determine the winner. If there is worries about parity or wonky standings, add the shoot-out to end the no-point ties. OT's become more important. Shoot-outs become more important. Better hockey.
Tada.

Bruce said...

Why not just get rid of the tie point altogether?

Ribs: The tie point is traditional. We need to reflect before trashing another tradition.

There are two schools of thought as to how to repair the broken system with something that's "fair and balanced", as FoxTV would put it.

1) Go strictly to winning percentage. Play every game to conclusion (as now), but recognize only the winner and loser, not two different types of loser where one gets rewarded with the tie point. Thus the situation is always winner takes all, and a hard-fought tie game like last night's can be entirely decided in the little skills competition at the end. You lost in a shootout? Tough. You lost. In fact you can do away with the points system altogether, since points would always be an even number represneting Wins*2. Hockey standings could be expressed in the identical manner as baseball, basketball, and (effectively) football, just winners and losers, none of this namby-pamby "tie" shit.

2) Award three points for every game, not just those that go to OT. Split the points 3 and 0 in a game decided in regulation, or 2 and 1 if decided in OT or a shootout. Teams playing safe to protect their 1 point down the stretch would always be sacrificing a potential point at the same time. The two teams involved simply split the three points differently. Teams playing to a tie no longer split 2 points 1-and-1, but 3 points 2-and-1. But they are competing with each other for those points rather than creating them at the expense of non-participants. This way the tradition of the tie point is maintained and the novelty of the shootout is embraced. The team that wins the game outright is suitably rewarded. You want all three points? Go out and get 'em.

I favour #2, cuz much as I enjoy the shootout, I prefer the idea of it deciding one third of a game's spoils (as now) rather than the whole enchillada. The key to both alternatives is that all games are worth the same, which would guarantee that the interests of the two teams on the ice are always at loggerheads. No more of this nudge-nudge-wink-wink "let's wait for a break or overtime, whichever comes first" joint strategies that happen all too frequently in Gary Bettman's New NHL.

Bruce said...

One of the things that happens frequently, although certainly not always, in the Bettman Point era is games that are tied partway through staying tied. Oftentimes what starts out as a good positive game gradually slows down as both teams protect their assets. This can be seen not only in the scoreline of a game like last night's, which was tied early in the second:

STL 0-1-0-0=1
EDM 1-0-0-0=1


... but is also reflected in the shots on goal:

STL 12-10-08-2=32
EDM 15-09-09-1=34


... and, if one cares to look, in the shots AT goal, which broke down this way last night:

STL 25-18-15-6=64
EDM 32-17-20-3=72


57 attempted shots in the first, then just 35 in each of the last two periods where the game state was 1-1.

Something similar happened in the ATL game, which was a wide-open affair until Bogosian's PP marker knotted the count at 3-3 midway through the second, then suddenly dried up.

ATL 2-1-0-1=4
EDM 1-2-0-0=3


For that game I have just the shot count by period. It displayed a similar trend, indeed a little more extreme: 31 total shots in the first period, then just 29 in the second and third combined. The whole second half of the game was downright painful to watch.

ATL 17-06-07-3 - 33
EDM 14-09-07-2 - 32


There are games where one of the teams scores anyway, as Colorado did on Saturday to break a longstanding 1-1 deadlock, and that forces the game to open up for awhile as the trailing team is forced to push the play. But there are way too many games where that tie-breaking goal doesn't happen cuz neither team is prepared to risk a lot to go after it.