First, I'd like to welcome Ken Socrates' hockey blog, Hockey Gone Wild, to the sidebar. Do go read!
(update - oh, yeah... helps to include the link. Fixed now.)
Second, I'd like to show off a little. I've been harping on the pity points in the NHL... it reduces competitiveness at the end of games, doesn't eliminate the "boring" ties (games still are even after 65 minutes), artificially makes teams look better than they are, and makes it harder to catch the teams ahead of you in the standings.
Tiresome? Well, you'll notice I haven't run the Real Standings for a while. But check this out from The Hockey News' Edward Fraser. Essentially, it's a pithier and better-written version of all of my harping over the past few months.
I'd like to take credit, but I've just discovered Mr. Fraser's work, and way back in December 2007, it turns out that he invented the Real Standings, with a twist - he eliminated ties entirely and went to a 2-0 system. In recognition, the next Real Standings here will be named the Fraser Standings. (I'm not changing the tag, though.) And yes, I will actually try to run them sometime over the weekend. It takes a little while to catch up if I haven't updated it regularly.
Showing posts with label real standings. Show all posts
Showing posts with label real standings. Show all posts
Thursday, January 28, 2010
Monday, December 14, 2009
Sharp-saving
This started ten days ago... I read about Hawerchuk's quick item here at Tango's site. It's devoted mostly to baseball and sabermetrics - excellent for fans and students alike. The art of recording and analyzing hockey data is jumping forward, and Tango posts a couple of times a week on it.
In any case, I stuck a comment out there... and, like most of my comments in places like this, it made no visible impact. I don't mind that, actually, since what I don't know about advanced statistics could fill a typical introductory textbook on the subject, no matter what the real standings posts may say. But I do find it interesting to follow the topics as they develop and see other people who may have thought along similar lines, and see what they post about it.
If all this puck talk bores you, I do apologize. I haven't the energy to risk much more right now.
In any case, I stuck a comment out there... and, like most of my comments in places like this, it made no visible impact. I don't mind that, actually, since what I don't know about advanced statistics could fill a typical introductory textbook on the subject, no matter what the real standings posts may say. But I do find it interesting to follow the topics as they develop and see other people who may have thought along similar lines, and see what they post about it.
It is not surprising that winning teams tend to have higher save percentages, but this save percentage split is much larger than normal (this year goalies have averaged .939 in all wins and .873 in all losses). Even though these teams outshot their opponents by a 2-to-1 margin on average, they still lost 40% of the time. This indicates that often a goalie does not have a low save percentage because of a low shot total, but rather that he has a low shot total because of a low save percentage.Emphasis mine, just to highlight the similarity to what I'd written last week:
Those two trends seem to suggest that stopping a high percentage of shots is what pushes the totals higher, not the reverse. In his four worst save % performances of the year, [Cam] Ward was yanked every time, meaning that he didn’t play well enough to be allowed the chance to face a lot of shots - in much the same way as a pitcher getting smacked around doesn’t get a lot of IPs.Incidentally, the author of that Puck Prospectus piece, Phil Myrland, blogs at the provocatively-titled Brodeur is a Fraud. Many of his sidebar links go to other worthy, puck-oriented writers (haven't checked them all). At this rate I may need a separate hockey section of the blogroll.
If all this puck talk bores you, I do apologize. I haven't the energy to risk much more right now.
Friday, December 11, 2009
Real Standings 4½ - now with value-added commentary!
I'm running this earlier than usual so the links below wouldn't grow stale. The standings are updated through Thursday's action. -nf
As a twist, I've added an additional line to the end of the tables, which you will see at the end - the teams are still sorted the way the NHL wants them to be, but next to "actual points" is a new column labelled 3-1-0. That simply means, three points for a win, one for a tie, and nothing for a loss.
I love the idea behind the system, in that it would assess a premium on winning the game. I'd extend overtime to ten minutes, or even a full period (twenty minutes), to further increase the chances for a decision, and scrap the shootout, because it's inherently unfair to have teams cop extra points on a skills competition after the game is finished - especially one that is more luck than skill. Teams would definitely push for more wins, even at the cost of an occasional loss: 1-1 would be better than 0-0-2; heck, 3-5 would be better than 0-0-8. Notice how, below, the Sabres would be well ahead of the Bruins, on the strength of having four more wins? That's the idea - reward teams that win their games, knowing that ties are no longer an option in the playoffs.
The only downside is that, at the extremes, it could become almost as unfair as the system it replaces.
Assume two teams go to OT every game. One goes all out for every point and finishes 30-52 for 90 points. The other tries to stall for the sure point every time, only countering in the most obvious situations (blocks leading to breakaways, for example). They split their games with the all-out team, draw everything else, and finish 3-3-76 for 85 points.
A team that has 49 more losses than another would look very odd indeed above them in the standings, even given the 27 more wins. Not that either would make the playoffs anyway, but to take a less extreme example - say the same two teams with 20 OT games each, not an unheard-of total. They each go 40-22 in regulation, but the tie-up team takes its 20 draws, while the all-outers split their games 7-13. The 47-35 team will finish ahead of the 40-22-20 team by one point. That could be a problem.
One compromise is to muddle up the two systems - three for a win, nothing for a loss, but only for regulation and overtime. Once the game gets to a shootout, each team gets a point, with the successful team (it could hardly be called a "win") gets the third point. The standings would look odd: just as an example:
A. 35-30 (12- 5) 134
B. 38-34 ( 7- 3) 131
C. 34-28 ( 7-13) 129
D. 33-33 (10- 6) 125
E. 30-30 (11-11) 123
That's regulation record (shootout record) total points, where the total points equals 3*wins + 2*SO wins + SO losses. It looks odd, but people would get used to it if you broke it up in pairs like that; but I dislike it for a different reason. If you go through it and subtract shootout losses from the total points, you would make just one change: team C would finish behind team D, which would be more fair, because D's winning percentage (43/82, .524) is better than C's (41/82, .500). Whenever possible, we want the teams winning more often to be the playoff teams, because in the playoffs, it's win or lose, no pity points, and no pat on the back for hanging in there for 60 or 65 minutes.
The thing is, nobody - NOBODY - complained that the 2-1-0 system was unfair. They complained that there were too many ties, that the ends of games were boring. The league added the five-minute regular-season OT in response, but ties didn't go down "enough," so they cooked up solutions to a problem that really didn't exist, and made it worse. All the fixes debated above, and elsewhere in the hockey world, have varying merits, but the simplest fix would simply be to go back to the old, not-really-broken system... just make the OT a real hockey period, sudden death, and not the five-minute afterthought we currently have.
As a twist, I've added an additional line to the end of the tables, which you will see at the end - the teams are still sorted the way the NHL wants them to be, but next to "actual points" is a new column labelled 3-1-0. That simply means, three points for a win, one for a tie, and nothing for a loss.
I love the idea behind the system, in that it would assess a premium on winning the game. I'd extend overtime to ten minutes, or even a full period (twenty minutes), to further increase the chances for a decision, and scrap the shootout, because it's inherently unfair to have teams cop extra points on a skills competition after the game is finished - especially one that is more luck than skill. Teams would definitely push for more wins, even at the cost of an occasional loss: 1-1 would be better than 0-0-2; heck, 3-5 would be better than 0-0-8. Notice how, below, the Sabres would be well ahead of the Bruins, on the strength of having four more wins? That's the idea - reward teams that win their games, knowing that ties are no longer an option in the playoffs.
The only downside is that, at the extremes, it could become almost as unfair as the system it replaces.
Assume two teams go to OT every game. One goes all out for every point and finishes 30-52 for 90 points. The other tries to stall for the sure point every time, only countering in the most obvious situations (blocks leading to breakaways, for example). They split their games with the all-out team, draw everything else, and finish 3-3-76 for 85 points.
A team that has 49 more losses than another would look very odd indeed above them in the standings, even given the 27 more wins. Not that either would make the playoffs anyway, but to take a less extreme example - say the same two teams with 20 OT games each, not an unheard-of total. They each go 40-22 in regulation, but the tie-up team takes its 20 draws, while the all-outers split their games 7-13. The 47-35 team will finish ahead of the 40-22-20 team by one point. That could be a problem.
One compromise is to muddle up the two systems - three for a win, nothing for a loss, but only for regulation and overtime. Once the game gets to a shootout, each team gets a point, with the successful team (it could hardly be called a "win") gets the third point. The standings would look odd: just as an example:
A. 35-30 (12- 5) 134
B. 38-34 ( 7- 3) 131
C. 34-28 ( 7-13) 129
D. 33-33 (10- 6) 125
E. 30-30 (11-11) 123
That's regulation record (shootout record) total points, where the total points equals 3*wins + 2*SO wins + SO losses. It looks odd, but people would get used to it if you broke it up in pairs like that; but I dislike it for a different reason. If you go through it and subtract shootout losses from the total points, you would make just one change: team C would finish behind team D, which would be more fair, because D's winning percentage (43/82, .524) is better than C's (41/82, .500). Whenever possible, we want the teams winning more often to be the playoff teams, because in the playoffs, it's win or lose, no pity points, and no pat on the back for hanging in there for 60 or 65 minutes.
The thing is, nobody - NOBODY - complained that the 2-1-0 system was unfair. They complained that there were too many ties, that the ends of games were boring. The league added the five-minute regular-season OT in response, but ties didn't go down "enough," so they cooked up solutions to a problem that really didn't exist, and made it worse. All the fixes debated above, and elsewhere in the hockey world, have varying merits, but the simplest fix would simply be to go back to the old, not-really-broken system... just make the OT a real hockey period, sudden death, and not the five-minute afterthought we currently have.
Monday, December 07, 2009
Classic (Standings) IV
Your Eastern Conference:
And your Western Conference:
This week's charts highlights one of the big downsides to the pity-point system: the playoff races aren't as tight. Five of the six divisions would have closer races if the league still scored under its previous rules. Only the Central Division would be otherwise, as Chicago's lead over Nashville and Detroit would move from five and six points to six and eight points. The Southeast and Northeast are the strongest the other way - Washington's nine-point edge over Atlanta would be a less-comfortable five, and in the old Adams Division (the Northeast), Boston's lead over Buffalo would actually be a deficit. That's rather a huge deal. Instead of the third seed as the weakest division winner (and a home-ice advantage over sixth-seeded Buffalo), the Bruins would slide all the way to seventh, and have to open the playoffs on the road, against Ovechkin and the Capitals.
Further, it's harder to make up ground in the new system. Not only are certain teams further behind than otherwise, but to make up that extra distance, they have to hope the teams ahead of them don't scrape points after regulation. That just makes it harder. On the surface it's more exciting to look at the standings and see your team at or above .500, a thrill that 80% of the league currently enjoys. It's not so much of a thrill, however, when you realize that even a great run isn't likely to move you up in the standings so long as the teams you chase are getting extra points from a skills competition after the actual games end. Think of the Columbus Blue Jackets and their fans. 13-11-5 sounds better than 11-11-7... except now they're eleventh in the conference, well out of the playoffs. They have to pass Vancouver and Detroit and catch Dallas, currently eighth. But in the old system, they'd already be ahead of Detroit, and they'd only be behind Dallas by one win (31 points to 29). They could do that on Wednesday against Florida, who are fairly poor. Oh, but not now... and not with Dallas' Marty Turco leading all goaltenders in overtime games. If he wasn't 1-5 in shootouts, it would be even worse.
(OK, to be fair, Columbus has Steve Mason, who is the only goalie in the league currently WORSE than Turco in the shootout this season. But is it fair to either team to have it come to this?)
And your Western Conference:
This week's charts highlights one of the big downsides to the pity-point system: the playoff races aren't as tight. Five of the six divisions would have closer races if the league still scored under its previous rules. Only the Central Division would be otherwise, as Chicago's lead over Nashville and Detroit would move from five and six points to six and eight points. The Southeast and Northeast are the strongest the other way - Washington's nine-point edge over Atlanta would be a less-comfortable five, and in the old Adams Division (the Northeast), Boston's lead over Buffalo would actually be a deficit. That's rather a huge deal. Instead of the third seed as the weakest division winner (and a home-ice advantage over sixth-seeded Buffalo), the Bruins would slide all the way to seventh, and have to open the playoffs on the road, against Ovechkin and the Capitals.
Further, it's harder to make up ground in the new system. Not only are certain teams further behind than otherwise, but to make up that extra distance, they have to hope the teams ahead of them don't scrape points after regulation. That just makes it harder. On the surface it's more exciting to look at the standings and see your team at or above .500, a thrill that 80% of the league currently enjoys. It's not so much of a thrill, however, when you realize that even a great run isn't likely to move you up in the standings so long as the teams you chase are getting extra points from a skills competition after the actual games end. Think of the Columbus Blue Jackets and their fans. 13-11-5 sounds better than 11-11-7... except now they're eleventh in the conference, well out of the playoffs. They have to pass Vancouver and Detroit and catch Dallas, currently eighth. But in the old system, they'd already be ahead of Detroit, and they'd only be behind Dallas by one win (31 points to 29). They could do that on Wednesday against Florida, who are fairly poor. Oh, but not now... and not with Dallas' Marty Turco leading all goaltenders in overtime games. If he wasn't 1-5 in shootouts, it would be even worse.
(OK, to be fair, Columbus has Steve Mason, who is the only goalie in the league currently WORSE than Turco in the shootout this season. But is it fair to either team to have it come to this?)
Tuesday, December 01, 2009
The real standings, hopefully improved
These images should at least be larger and more legible - I uploaded each conference as a separate image. Blogger lets me make each individual picture a bit bigger, with better resolution, if I do that rather than upload the whole shebang as a unit.
A table would be the gold standard, however. Hm... in a series of posts mocking the league's pity points, my method could at best be described as a shootout loss - thus deserving of a pity point itself. At least I've successfully employed the concept of irony in my blogging.
Nonetheless I will still decry; heck, at least I'm trying to improve. The NHL seems more than happy about their system, and I think I see why. Say you're CommissionerGarrison Keillor Gary Bettman. You have to interest a non-market such as Phoenix to accept a transplanted team like the Winnipeg Jets, who finished the 1993-94 season a dreadful 24-51-9. Or, let's say you're trying to expand into new markets... but your expansion Tampa Bay Lightning were less than an exemplary example - that's 17-55-10 in '97-98 and 19-54-9 in '98-99. Yeah, who wants in?
So, the very next year, Bettman introduces an overtime loss column. Instead of Tampa duplicating its 19-54-9 record, it gets to say it was 19-47-9-7. Voila! And the brand-new Atlanta Thrashers avoid 60 losses by moving four of them to this OTL column. Huzzah! So what if these points wind up costing the Carolina Hurricanes (37-35-10) a playoff spot in favor of the inferior Buffalo Sabres (35-36-11, but four of the losses were in OT)? And by 2001-02, there were no 50-loss teams in the league; or, scratch that, there were three, but only if you actually count the OTL as an L. And the year after that, 27 of the 30 teams could say, hey, we got points in more than half our games! Somehow only three NHL teams managed to lose 40 or more games.
The extra column drove purists crazy, but it kept the owners happy. Nobody wants to say they were well-below .500, so the extra points were a godsend. Then, after they blew up the 2004-2005 season, they came back with the shootout - because if points for losing was good, then points for winning after the game was over was doubleplusgood!
As a result you have the table below, where 24 out of 30 teams are .500 or better. Everybody's above-average!
Well... not Carolina. Even their pity-point record is 5-17-5; it's almost impossible to be that bad under any circumstance. That translates into a full-year record of 15-51-15 with one game to play; even under the old rules, it's a record of futility that few teams have matched. But under the old rules, the 'Canes would be 2-19-6. There's just no way to sugar-coat that. It's epically bad, a SIX WIN pace. Even if they could scrape the seventh win in the last game, that's a full-year of 7-57-18, and only the ties would keep them from being possibly the worst team in the history of the league. Their seven wins would break the old 1975 Washington Capitals record for the fewest in the expansion era.
A table would be the gold standard, however. Hm... in a series of posts mocking the league's pity points, my method could at best be described as a shootout loss - thus deserving of a pity point itself. At least I've successfully employed the concept of irony in my blogging.
Nonetheless I will still decry; heck, at least I'm trying to improve. The NHL seems more than happy about their system, and I think I see why. Say you're Commissioner
So, the very next year, Bettman introduces an overtime loss column. Instead of Tampa duplicating its 19-54-9 record, it gets to say it was 19-47-9-7. Voila! And the brand-new Atlanta Thrashers avoid 60 losses by moving four of them to this OTL column. Huzzah! So what if these points wind up costing the Carolina Hurricanes (37-35-10) a playoff spot in favor of the inferior Buffalo Sabres (35-36-11, but four of the losses were in OT)? And by 2001-02, there were no 50-loss teams in the league; or, scratch that, there were three, but only if you actually count the OTL as an L. And the year after that, 27 of the 30 teams could say, hey, we got points in more than half our games! Somehow only three NHL teams managed to lose 40 or more games.
The extra column drove purists crazy, but it kept the owners happy. Nobody wants to say they were well-below .500, so the extra points were a godsend. Then, after they blew up the 2004-2005 season, they came back with the shootout - because if points for losing was good, then points for winning after the game was over was doubleplusgood!
As a result you have the table below, where 24 out of 30 teams are .500 or better. Everybody's above-average!
Well... not Carolina. Even their pity-point record is 5-17-5; it's almost impossible to be that bad under any circumstance. That translates into a full-year record of 15-51-15 with one game to play; even under the old rules, it's a record of futility that few teams have matched. But under the old rules, the 'Canes would be 2-19-6. There's just no way to sugar-coat that. It's epically bad, a SIX WIN pace. Even if they could scrape the seventh win in the last game, that's a full-year of 7-57-18, and only the ties would keep them from being possibly the worst team in the history of the league. Their seven wins would break the old 1975 Washington Capitals record for the fewest in the expansion era.
Monday, November 23, 2009
The real standings
Weird weekend. The Chiefs, Lions, and Raiders all won yesterday. I feel like I woke up in 1993.
The NHL's "real standings" this week -
Not quite as illegible as last time. Trying to make the picture bigger just blurs everything. My HTML-fu is weak, and I can't get a table to look like anything but an affront to statistics everywhere. The dumb thing is, the table looks amazing in Excel, but I can't just dump the spreadsheet page to the web... or, to be more precise, *I* can't just dump the thing on here. A skilled person could. Hope I can find one before next week, but there's that Thanksgiving thing... you know, the one everyone seemed to skip this year to get right from Halloween to the high holy shopping season.
The NHL's "real standings" this week -
Not quite as illegible as last time. Trying to make the picture bigger just blurs everything. My HTML-fu is weak, and I can't get a table to look like anything but an affront to statistics everywhere. The dumb thing is, the table looks amazing in Excel, but I can't just dump the spreadsheet page to the web... or, to be more precise, *I* can't just dump the thing on here. A skilled person could. Hope I can find one before next week, but there's that Thanksgiving thing... you know, the one everyone seemed to skip this year to get right from Halloween to the high holy shopping season.
Monday, November 16, 2009
The real standings
I'm on record as not liking the concept behind giving free points to losing teams in hockey games.
In our own low little league it's not so bad, we get sorted into divisions and everyone, no matter how downtrodden, gets to play a playoff game against a similar team. In the NHL, where making or missing the playoffs could lead to extra millions for a team, none of whom are currently in the best financial shape? Yeah, could be a problem.
Psychologically, I also don't like the concept of a game being decided by a glorified skills competition, especially one where only one skill is on display.
Finally, the whole rationale behind it was to eliminate the allegedly-pernicious tie, to prevent stuff like this. That's a LOT of ties, right? Looks a lot better than, say, today's standings... right?
Yeah... not so much. So I put myself to a little research project over the weekend: if there were no "pity point" and games ended after overtime, what would the standings be like? That table is below.
It's simple, if a little time-consuming. Shootout wins are taken out of the win column and moved to the tie column, and overtime losses are taken out of the tie column and moved to the loss column. No other adjustments are needed. The table has the current standings on the left, the adjustments in the middle, and the Old School Adjusted Standings on the right.
It really should be a table, not a picture - you have to click to embiggen - but the major hunk of the work is done. I hope to be able to run this every Monday, hopefully in a more legible format.
In our own low little league it's not so bad, we get sorted into divisions and everyone, no matter how downtrodden, gets to play a playoff game against a similar team. In the NHL, where making or missing the playoffs could lead to extra millions for a team, none of whom are currently in the best financial shape? Yeah, could be a problem.
Psychologically, I also don't like the concept of a game being decided by a glorified skills competition, especially one where only one skill is on display.
Finally, the whole rationale behind it was to eliminate the allegedly-pernicious tie, to prevent stuff like this. That's a LOT of ties, right? Looks a lot better than, say, today's standings... right?
Yeah... not so much. So I put myself to a little research project over the weekend: if there were no "pity point" and games ended after overtime, what would the standings be like? That table is below.
It's simple, if a little time-consuming. Shootout wins are taken out of the win column and moved to the tie column, and overtime losses are taken out of the tie column and moved to the loss column. No other adjustments are needed. The table has the current standings on the left, the adjustments in the middle, and the Old School Adjusted Standings on the right.
It really should be a table, not a picture - you have to click to embiggen - but the major hunk of the work is done. I hope to be able to run this every Monday, hopefully in a more legible format.
Subscribe to:
Posts (Atom)