Selection Sunday Discussion

[herodotus]

The problem with using margin of victory is that it often lies. Let's say two games end with a 10 point margin. Same thing right? Wrong! Game one had the loser with a one point lead with 1:30 left in the game. The winning team hits a three to take a 2 point lead with 1:20 left. After a miss by the loser, the winner drains another three with 50 seconds left. The loser sends the winner to the foul line 3 times in a desperate attempt to catch up. They make all 6 shots, while the loser misses long threes. Final score 80-70. In the second game, the winning team has a double digit lead from the middle of the first half on. They lead by 25 with 5:30 left, when the coach empties his bench. The scrubs are outscored by 15 points the rest of the game. In KenPom, these results are identical. These are not rare results, endings like this happen all the time. Thus, KenPom rewards teams that run up the score, and punishes teams that keep trying when a game gets away from them in the last 2 minutes.

[anXUfan]

Should teams be seeded mostly based on what they've earned (win/loss focused RPI) or how they are projected to perform in the future (MOV focused KenPom)? I lean towards what they've earned.

[XUFan09]

The problem with using margin of victory is that it often lies. Let's say two games end with a 10 point margin. Same thing right? Wrong! Game one had the loser with a one point lead with 1:30 left in the game. The winning team hits a three to take a 2 point lead with 1:20 left. After a miss by the loser, the winner drains another three with 50 seconds left. The loser sends the winner to the foul line 3 times in a desperate attempt to catch up. They make all 6 shots, while the loser misses long threes. Final score 80-70. In the second game, the winning team has a double digit lead from the middle of the first half on. They lead by 25 with 5:30 left, when the coach empties his bench. The scrubs are outscored by 15 points the rest of the game. In KenPom, these results are identical. These are not rare results, endings like this happen all the time. Thus, KenPom rewards teams that run up the score, and punishes teams that keep trying when a game gets away from them in the last 2 minutes.

Yes, there's still variance in the model. If you are seeking a perfect model without variance, it's fool's gold. There's still a lot less variance than for the RPI, and over a 30+ game season this variance generally gets balanced out. Just concerning your example of a team continuing to fight, in a 30-game schedule every team has probably been on both sides of this situation, even Xavier, who is notable this season for not really having close games. So, those situations tend to balance themselves out.

You also talk about a team getting rewarded for running up the score and to some extent that is true, with the example that comes to mind first being Louisville. However, it is really hard to run up the score. You make it sound like it's something easy to do, but the losing team generally keeps fighting. And, becasue they're losing, they usually are giving a greater effort than the team in the lead. Thus, if a team can push the margin further, with the goal of putting the game further out of reach, that says something positive about that team's ability. If a team can just maintain a margin that they have established, that also says something positive about them. It's my one major complaint of this Xavier team. When they have the chance to finish putting away a team, they too often let them hang around, which is risky.

Just consider the end-of-season Xavier vs. Creighton game. Xavier had extended the lead to 18 points with 5 minutes to go. Now, when a team has opened up a game to that margin with some time left in the game, what's the best way to ensure they win? Simple. Push the lead further, out to 23 to 25 in the next couple minutes. If Xavier did that, Creighton wouldn't have almost completed a comeback in the last five minutes of the game. Even if they hadn't extended the leader further but had been focused on both sides of the ball, they would have at worst played even with Creighton and kept the score around 18.

Basketball involves a lot of luck when dealing in small sample spaces. Over a larger sample space, such as the majority of a game, luck is less impactful. In these larger sample spaces, significantly better teams should be able to open up margins such that the luck in the smaller sample space at the end of the game doesn't matter. A good example of this is with Xavier (who isn't always bad about a lead). Against Villanova at home, they opened up the game to a 15-point lead with 2 minutes to play. A comeback was close to impossible at this point. Then, Villanova started getting hot, hitting three-pointer after three-pointer, including three of them in the last 50 seconds. It didn't matter, though, as Xavier had run up the score a bit just before that meaningless onslaught.

In essence, it's a team's self-interest to run up the score. To not do so is to risk losing the game against a team you are beating. But, some teams are better at running up the score than others and their risk of losing is thus lower.

[SCS]

Creighton and Tulsa seem to be scheduling pretty damn well... I mean, I know Creighton had some cupcakes this year, but even still got Arizona St playing there. Last year they got Oklahoma and St Mary's at home. 2 years ago, they got Tulsa, Nebraska, Cal there.

Tulsa got Wichita at their place 2 years ago. last year, they got Creighton and Oklahoma at their place. This year Wichita again.

It is much much easier to get to Omaha and Tulsa than it is to Olean.

[XUFan09]

Should teams be seeded mostly based on what they've earned (win/loss focused RPI) or how they are projected to perform in the future (MOV focused KenPom)? I lean towards what they've earned.

This is a bad question, since a team's RPI is not directly a factor in the selection and seeding process, whereas their Kenpom rating is. I think teams should be seeded based on what they earn, but the quality of their wins and losses should be judged by Kenpom or a similar system rather than by RPI. For example, Xavier went on the road and beat the Kenpom #68 Georgetown team, not the RPI #106 Georgetown team. I used Georgetown as an example, because no matter how disappointing they've been this year, there are not 105 teams better than them in Division I. They've just lost a lot of close games, and their Kenpom rating is a better reflection of how they've performed this year. As a result, their Kenpom rating is also a better reflection of how tough they were to beat.

Ironically, RPI is a better measure of what a team has earned in the season, but we use it to evaluate how challenging a team's opponents were. Kenpom is a better measure of how challenging a team's opponents were, but we use it to evaluate how well they have played (similar to what they have earned).

[HoosierPal]

Seton Hall as a 6 is a travesty.

Are you talking about the same Seton Hall that was gifted a W over St Jojns two weeks ago? You can't just look at the last few games.

[stever20]

travesty is way too strong of a word. in the polls that got released today, they were 20 in in the AP and 21 in coaches. Or in other words, right on the 5/6 cut line. If anyone has a right to be upset, it's Arizona. They were 17 in AP/16 in coaches. And they're a 6.

[JPSchmack]

And about St Bonnie's how can a team like Monmouth get those games, and you can't? Part of it is you didn't do an exempt tourney where you can get those big games OOC some times.

The A10 doesn't allow it's teams to take more than one buy game a year. Also, when looking for easy home games, the P5 powers prefer to avoid games against teams that might be dangerous. No one expects a MAAC team to be a threat. The better A 10 teams are viewed as dangerous. The exempt tourneys don't want teams like Bona, because they don't want to risk their marquee being upset, and not reaching the final. Monmouth was mistakenly viewed as garbage, which is why they got those chances.

It’s not like we’re turning down Maui and Atlantis. You have to be INVITED to these tournaments, and they want big fan bases to travel to Maui & Atlantis and buy resort tickets.

Monmouth “got in” to the Orlando tournament BECAUSE THE MAAC IS THE HOST and there’s always a MAAC team in the field.

Bonaventure used Rochester to try and get a P5 team to come play…

2001 - Bona beat C-USA Charlotte (back when they were good)

2004 - #16 Syracuse agreed to play us because we just got hammered by the NCAA and lost all our recruits and best players. They beat out 7-21 team by NINE.

2006 - West Virginia agreed to play us because Mike Gansey was one of those players who transferred after the scandal and he’s from Rochester. They beat us by 5.
2010 - St. John’s beat us by 1
2011 - Va Tech beat us in OT
2012 - NC State won on a buzzer-beater, and later went to the Sweet 16 that year (We got a 14-seed vs Fla St, and only lost by 3).

That’s 1-5 by an average margin of 2.7 points. No one’s been back since.


We played Saint Joe’s in Rochester this year in a conference game (And won). We only played there because we couldn’t find ANYONE DECENT to play us OOC

[JPSchmack]

The problem with RPI isn’t margin of victory, which is stupid. The real problem with RPI (And the selection process in general) is that it was DESIGNED IN 1981.

That was back when 18 conferences had teams we’d call “power teams” like:

Big East, ACC, SEC, Pac-10, Big 8… plus…
Texas & Baylor in the Southwest, Memphis & Florida State in the Metro, Xavier & Butler in the MWCC, Creighton in the MVC, Temple, Pitt & WVU in the Atlantic 10, Notre Dame & DePaul as independents, Utah & BYU in the WAC
UNLV in the Big West, Western Kentucky was good and in the Sun Belt, etc, etc etc.

With everyone spread out, everyone had 2-6 Top 50 games IN CONFERENCE max because very few leagues played more than 14 conference games and half the schedule or more was OOC games.

With only 12 one-bid leagues, BUBBLE TEAMS were all 45-55 in the RPI and “Top 50 RPI wins” meant “Everyone in the discussion for an NCAA bid.”

Well now, there’s 8 good conferences, a couple “mid-majors” like the WCC, and 22 one-bid leagues.

#72 RPI is getting an at-large, so Top 50 wins don’t tell you jack anymore. Bona ONLY had 3 Top 50 wins. In FOUR GAMES, while Stanford & K-State went 3-14. (Ditto bad losses. Monmouth got HAMMERED for 21-3 against bad teams while Wisconsin was 9-6!).


Because everyone plays 18 conference games, plus conference tournaments, TWO THIRDS of your SOS is NOTHING but the records of your conference opponents. And since every conference goes .500 against itself…

If the Big XII goes 100-20 OOC against TERRIBLE TEAMS, every B12 game is gonna add .6333 to their SOS.
If the Big XII goes 60-60 OOC against GREAT teams, every B12 game is gonna add .5000 to their SOS.
(I’m just using them cause they’re a balanced conference schedule)

So they CAN get great SOS without actually playing anyone. As long as the bottom teams of their league clown stomp terrible teams. Then their heavy hitters can get marquee wins.


And that’s the A-10’s problem: Pac-12 can game the RPI, split with each other, and get seven bids.
Oregon State (lost to Valpo), USC (lost to Monmouth), Arizona (beat no one with an OOC SOS worse than ours!)

If the A-10 does the exact same thing… it’s “Who’d you beat OOC? You’re just playing RPI games!”

If we do what the P5 does, we get screwed for it.
If we go schedule tough OOC and lose on the road (like EVERYONE does), our RPIs crumble and it’s like Linda Bruno’s “Anyone, Anywhere” policy years: ONE BID LEAGUE.

The committee screws us no matter what we do:
Bona beats Top 50 A-10 teams, wins the league, RPI 30 = Sorry, no good OOC wins!
GW beats Virginia & Hall OOC = Sorry, 68 RPI RPI too low!
Davidson did exactly what Vandy did. Davidson wasn’t in the discussion.


For 15 years, we had Greg Shaheen STEERING the committee, telling them how to do the job. His replacement lets the committee decide HOW to pick the teams now. The new evolution of the NCAA process is so jacked up they made John Calipari seem smart: Just tell us what to do, and we’ll try and do it. Don’t shift the criteria season to season and team to team.

EVERYONE complains about everyone else “gaming the RPI,” but if you fix the RPI so that any attempt to game it FORCES you to have a good season or fail trying, then we could just use the RPI to pick the field. Why don’t we just do THAT?

[stever20]

so it sucks that St Bonnie's can't get teams to come play.. But then don't cry when your OOC SOS isn't good.

Monmouth got hammered because they lost 3 games to sub 200 teams. Wisconsin lost 1 time. Wisconsin also had 5 top 50 wins to only 1 for Monmouth.

Big 12 went 103-23 OOC (.8175) against the #3 OOC SOS.
Pac 12 went 111-33 OOC (7708) against the #1 OOC SOS.
A10 went 109-57 OOC (.6566) against the #6 OOC SOS

So tell me again how the Pac 12 gamed the RPI exactly? They played the toughest OOC SOS of any conference in the country, yet had a good record. So let's see what the A10's record would be against the #1 SOS, and then we can talk. Unfortuately the conference you could talk the most about is the Big East- they go 95-30(.7600) but only against the #17 OOC SOS this year.

Oregon St- #58 OOC SOS
USC- #86 OOC SOS
Arizona did have the #234 OOC SOS- but they did beat Gonzaga- and even Fresno was better than anyone you beat. They also went 12-6 in conference play. 13 top 100 wins.

GW yeah they beat UVA and Seton Hall. But their OOC SOS overall was 210. lost to DePaul. Also 1-5 road/neutral vs top 50 teams
Davidson. 3-9 vs top 100 teams. Vandy was 7-10 vs top 100 teams. So no, Davidson didn't do exactly what Vandy did at all....