[Cracked Sidewalks] Revisiting Luck

[lawdog77]

Absolutely. Here's the kenpom definition:

Luck – A measure of the deviation between a team's actual winning percentage and what one would expect from its game-by-game efficiencies. It's a Dean Oliver invention. Essentially, a team involved in a lot of close games should not win (or lose) all of them. Those that do will be viewed as lucky (or unlucky).

Providence is probably around the 30th-40th best team in the country. Torvik has them at 37, Pomeroy at 46, Haslametrics at 60. Are they as bad as losing to Virginia by 18 or Marquette by 32? Probably not. But they also aren't close to as good as their 18-2 record.

Consider this. In 11 games against top-100 opponents on Pomeroy, Providence has a -2 combined scoring margin. You would expect that team to be 5-6. If they're lucky, 6-5, or really lucky, 7-4. Providence is 9-2 in those games. That's absurd luck and history shows those teams tend to pay the luck piper the first weekend in March.

Reminds me of the 1960 World Series between Pittsburgh and New York. Yankees outscored the Pirates something like 55-27 but  lost the Series.

[IrwinFletcher]

Absolutely. Here's the kenpom definition:

Luck – A measure of the deviation between a team's actual winning percentage and what one would expect from its game-by-game efficiencies. It's a Dean Oliver invention. Essentially, a team involved in a lot of close games should not win (or lose) all of them. Those that do will be viewed as lucky (or unlucky).

Providence is probably around the 30th-40th best team in the country. Torvik has them at 37, Pomeroy at 46, Haslametrics at 60. Are they as bad as losing to Virginia by 18 or Marquette by 32? Probably not. But they also aren't close to as good as their 18-2 record.

Consider this. In 11 games against top-100 opponents on Pomeroy, Providence has a -2 combined scoring margin. You would expect that team to be 5-6. If they're lucky, 6-5, or really lucky, 7-4. Providence is 9-2 in those games. That's absurd luck and history shows those teams tend to pay the luck piper the first weekend in March.

Madison is almost equally as lucky as Providence.

[brewcity77]

Madison is almost equally as lucky as Providence.

They're up there, which is why they were one of the topics of the article, though Providence is currently the luckiest team in kenpom history and ranked #1 while Madison is 7th. In two possession games, which essentially should be a coin flip (one three bounces out at one end and in on the other and the result flips) Providence is 7-0 while Madison is 9-1, with the only loss coming to Providence. And that's why those teams are going to be overseeded come March.

[Dr. Blackheart]

Absolutely. Here's the kenpom definition:

Luck – A measure of the deviation between a team's actual winning percentage and what one would expect from its game-by-game efficiencies. It's a Dean Oliver invention. Essentially, a team involved in a lot of close games should not win (or lose) all of them. Those that do will be viewed as lucky (or unlucky).

Providence is probably around the 30th-40th best team in the country. Torvik has them at 37, Pomeroy at 46, Haslametrics at 60. Are they as bad as losing to Virginia by 18 or Marquette by 32? Probably not. But they also aren't close to as good as their 18-2 record.

Consider this. In 11 games against top-100 opponents on Pomeroy, Providence has a -2 combined scoring margin. You would expect that team to be 5-6. If they're lucky, 6-5, or really lucky, 7-4. Providence is 9-2 in those games. That's absurd luck and history shows those teams tend to pay the luck piper the first weekend in March.

The luck factor is also an indication that his models aren't very good for predicting Providence, remembering he pools all the 358 schools when modeling with a bunch of indexing. So "luck" is the residual random error, that he then throws into his prediction models.

As to making it out of the first weekend, only 23.5% of the 68 teams in reality make it out of the first weekend? How do the unlucky teams compare to the lucky teams as a control?

[MarquetteDano]

Reminds me of the 1960 World Series between Pittsburgh and New York. Yankees outscored the Pirates something like 55-27 but  lost the Series.

With pitchers that is more common in baseball.  Team has 4 starters in series and three are aces and one sucks.  Then the relief sucks too.  This kind of thing can happen.

[rocky_warrior]

What does that tell you? Obviously check back after the official NCAA Bracket comes out, but if you see Wisconsin, Providence, or West Virginia in your bracket, don't expect to see them in the second weekend.

To me, this sounded like sour grapes when posted.  Now confirmed.  And we have yet to see what WI does tomorrow.

[brewcity77]

To me, this sounded like sour grapes when posted.  Now confirmed.  And we have yet to see what WI does tomorrow.

Or it was pretty much correct, as West Virginia's luck ran out before the tourney, Wisconsin bowed out in the first weekend, and Providence has succeeded in playing to seed but still needs another win to truly be an outlier on par with 2013 Marquette (mentioned in the original linked 2019 article).

[panda]

Or it was pretty much correct, as West Virginia's luck ran out before the tourney, Wisconsin bowed out in the first weekend, and Providence has succeeded in playing to seed but still needs another win to truly be an outlier on par with 2013 Marquette (mentioned in the original linked 2019 article).

Pretty unlucky Wisconsin lost their starting Pg in the first half