PDA

View Full Version : Visualizing the RPI of GU's Opponents over Time



AK457
02-24-2014, 08:27 PM
There's been a decent amount of discussion on GU's chances of getting an at-large big to the NCAA Tournament this year and how RPI among other factors is used to invite and seed teams to the tournament. This year, GU has a current RPI of 28 and SOS of 88, but these factors alone aren't enough information to make an informed guess on their selection into the tournament -- we should also take into account the distribution of GU's opponents' RPI. This is important because of an effect on their record that is not captured by RPI or SOS metrics. The probability that GU wins a game is likely according to an s-curve relationship (imagine this picture http://2.bp.blogspot.com/-zNdt4hrywzw/T610iC3SBRI/AAAAAAAAALk/y0s2YlNQZMU/s1600/S-Curve.PNG without the axes labels) where the x axis is opponent's RPI ranking and the y axis is the probability of winning. The important trend being that GU is essentially as likely to win against a team with an RPI of, say, 250 as it is against a team with an RPI of, say, 200, but beating the team with the RPI of 200 will cause GU's RPI to be higher than having beat a team with an RPI of 250. In other words, if teams were able to accurately predict the RPIs of their opponents, the optimal strategy would be to load up on teams where victory is almost certain but where their opponents' RPI is as high as possible, which brings us to this years distribution of GU's opponents' RPI relative to previous years:

http://i.imgur.com/uIvaqVG.jpg

You'll notice that our opponents are less frequent at both the high end (1-50) and low end (250-350) of the RPI range relative to previous years (e.g., we have zero opponents in the 300-250 range, which has never happened in the last 7 years), which means that our RPI is more of a result of us having not playing against poor teams than having beaten good teams. This is problematic during the selection process given that the committee awards teams for having played good opponents, and they are sure to have figures like the one I provide here to look at that are able to disentangle RPI, SOS, and the team's record. So go Zags and win the conference tournament!

Oregonzagnut
02-25-2014, 12:39 AM
Interesting. VERY interesting.

While we have zero RPI opponents lower than 300, we ALSO have the fewest games of any year between 0-50. that hurt us the most IMO.

BTW, that top plot is just like the "Law of diminishing returns". Is Few working much harder to simply maintain the status quo of 27 wins and a 2nd round tournament?

caduceus
02-25-2014, 03:30 AM
Good stuff, AK. Two things to consider. First, the major screw up in Maui. Win that first game and we had potentially two high-RPI games that follow. No doubt that Dayton loss cost us more than any game this season.
Second: Battle in Seattle. Big name team pulls out of the game, leaving us with So. Alabamistan.

That's potentially 3 more top-50 games, one fewer cupcake, and one fewer non-DI team game that doesn't count in the RPI. Unlucky.

GoZags
02-25-2014, 05:05 AM
Good stuff, AK. Two things to consider. First, the major screw up in Maui. Win that first game and we had potentially two high-RPI games that follow. No doubt that Dayton loss cost us more than any game this season.
Second: Battle in Seattle. Big name team pulls out of the game, leaving us with So. Alabamistan.

That's potentially 3 more top-50 games, one fewer cupcake, and one fewer non-DI team game that doesn't count in the RPI. Unlucky.

In addition to these comments .... SMC being in the 60's could potentially cost GU 3 games against a 'Top 50 school'. But '13/'14 isn't over. GU could play Top 50 BYU in the WCC tourney and one or two more Top 50's in the dance. But caduceus is correct, what happened in Maui and Seattle WAS "unlucky" vis a vis Top 50 games.

cbbfanatic
02-25-2014, 06:48 AM
You'll notice that our opponents are less frequent at both the high end (1-50) and low end (250-350) of the RPI range relative to previous years (e.g., we have zero opponents in the 300-250 range, which has never happened in the last 7 years), which means that our RPI is more of a result of us having not playing against poor teams than having beaten good teams. This is problematic during the selection process given that the committee awards teams for having played good opponents, and they are sure to have figures like the one I provide here to look at that are able to disentangle RPI, SOS, and the team's record. So go Zags and win the conference tournament!

this is the general approach i was referencing in my view that the wcc isnt materially better than past years (from watching the talent on the floor), and that they have just more effectively managed their rpi numbers (not a bad thing). I am guessing a similar trend could be seen with the other conference teams... fewer games against the very bottom, fewer against the very top, living in the 100-150 neighborhood more than before.

this type of rpi management is what has made the #s less meaningful over the last 7-10 years, to me. and i would guess the committee sees it as well.

Oregonzagnut
02-25-2014, 02:24 PM
I think the WCC has improved from top to bottom. But the addition of BYU cannot be ignored. If BYU stays and Gonzaga can weather this storm, teh next few years in the WCC will be inching closer to the power 6-7 conferences as far as RPI goes.

But one thing to remember, if you take the wC power numbers from KenPom and Sagarin, the WCC is stronger overall. You cannot schedule to pre arrange those ratings. So IMO the WCC has improved from teh 13th best conference 4 years ago to the 8/9th best conference now and 33% of that is due to BYU, 33% of that is due to better RPI scheduling strategy, and 33% really is because whatever was attractive about the WCC to BYU, is likely attractive to the better recruits.

The WCC is perfect for us and I see no greener grass anywhere else. At this point.

seacatfan
02-25-2014, 02:33 PM
this is the general approach i was referencing in my view that the wcc isnt materially better than past years (from watching the talent on the floor), and that they have just more effectively managed their rpi numbers (not a bad thing). I am guessing a similar trend could be seen with the other conference teams... fewer games against the very bottom, fewer against the very top, living in the 100-150 neighborhood more than before.

this type of rpi management is what has made the #s less meaningful over the last 7-10 years, to me. and i would guess the committee sees it as well.

Effectively managing RPI numbers is not a bad thing at all. The Mountain West and Missouri Valley conferences figured it out years ago. You look at the high RPI numbers of some of their teams over the years and you actually look at the team and their schedule and you're left scratching your head why the RPI is so high. Certainly has benefited both conferences at Tourney selection time.