This is the flaw in RPI. Win games and it almost counts against you Where as if you scheduled more teams along the lines of Mt St Marys at 169 they would have a much better RPI yet still a bad team. Not much difference really.
Bubble Watch
- Thread starter vcl3
- Start date
This is the flaw in RPI. Win games and it almost counts against you Where as if you scheduled more teams along the lines of Mt St Marys at 169 they would have a much better RPI yet still a bad team. Not much difference really.
I thought they modified it a few years back. A road win is 1.4 weighted, a home loss is 0.6 weighted. Only for your winning percentage, not the OPP or OPP2 percentages. Something like that.
I stand corrected. Didn't know that adjustment happened. But that weighting only applies to the 25% of your rank that comes from your own winning percentage
Is this site different rankings then on the ESPN site? I just looked up on ESPN today and it said we were 55 in RPI still, 26 BPI still, and 8 SOS. We were 55 in RPI on Feb. 6. So we won over OU and on the road at KState (whose RPI is 48) and we gained no ground according to their RPI rankings. What?????
Yes, in all three cases it's winning percentage. Individual games don't directly affect anything outside the weighting given to home/road/neutral wins in calculating your own winning percentage
Yes, ESPN only updates periodically. RealtimeRPI is as it says, it updates after every game.
We moved up another spot last night from #47 to #46 because of some abstract game result somewhere.
http://realtimerpi.com/college_Men_basketball_rpi.html
We're 46 in RPI, 26 BPI, 8 SOS according to ESPN?
http://www.espn.com/mens-college-basketball/bpi/_/page/2/view/overview
https://fivethirtyeight.com/features/the-ncaa-is-modernizing-the-way-it-picks-march-madness-teams/
Thought this was as good of a place as any to post. Talks about the lengths the NCAA is going to marginalize RPI a little bit and lean more towards using a lot of rankings to create a composite ranking.
Thought this was as good of a place as any to post. Talks about the lengths the NCAA is going to marginalize RPI a little bit and lean more towards using a lot of rankings to create a composite ranking.
I think our RPI of 46 is more indicative of our entire season than the 26 we are in BPI/Kenpom.
I think we're playing like 26 or even better the past few games, but not our entire season as a whole. KenPom and BPI's flaws when used to evaluate a season are they reward a close loss too much when the human mind really doesn't give a crap about that outside of gambling. RPI's flaw is as you pointed out that scheduling the right or wrong "bad" teams can skew the result more than it should.
Almost any expert would probably rank us somewhere between the two or closer to our RPI, AP certainly doesn't think we're 26 or we'd be ranked/nearly ranked. Neither human voter poll has us in the top 35.
Well, actually they have been routinely getting votes in AP and coaches poll for a lot of the season usually putting them in top 35 and top 30 in the country. I would guess if they beat TCU they again will be close to being ranked. So that doesn't really fit with what the RPI has generally had ISU at.
I'm just saying it's useful to look at both when the trend is to worship KenPom/BPI as some perfect alternative to RPI. There have been many years lately where RPI represented us better and higher when KenPom/BPI had us lower than we should have been for punishing close wins and ranking us below teams with lots of close losses.
I don't think either tells the story this year. We haven't been below 30 in KenPom even at times we were really struggling and clearly weren't a top 30 team. Likewise I don't think we've been been a sub 50 team but our RPI has dipped there several times.
Again, this goes back to what the metrics are used for. RPI is an attempt to be an explanatory metric, using only wins, losses, opponent's (and opponent's opponent's) wins and losses, and location. One could make an argument that because wins and losses, regardless of the score, are what really matters, RPI (or something like it) is an effective way to evaluate a team's resume. However, it could also be argued that the RPI specifically places too much weight on who a team played, rather than how a team actually performed, and opens itself up to being gamed by savvy schedulers (like Larry's Colorado State team a couple of years ago).
In contrast, KenPom, Sagarin, BPI, etc. are predictive metrics that use score margins/efficiency ratings to predict how good a team is. Scores are more effective than wins/losses in predicting the future performance of a team. Bill James showed that this was true many years ago in baseball, and that result has been applied successfully to many other sports in the years since. It has consistently been shown that whether a team wins or loses a close game isn't particularly predictive of how it will perform going forward - a lot of the result is based on luck.
In short, I think people tend to conflate RPI with KenPom/BPI, when in reality they are measuring different things and have different uses. Which one is "better" depends on your priorities, I suppose.
Could have sworn it said 55 when I checked at 9:00 this morning. But, yes I rechecked...glad its 46.
The committee has in the past weighted improved recent play so if we have the Solomon and Burton from the KSU game in these upcoming gsmes we may receive a seed bump past our overall record. Imo it's turning more into the subjective "how good" does the committee think you are like in the CFB playoffs. With a good Burton and recent Young we're arguably a top-25 team
The committee intentionally removed any "last 5 games" or "last 10" metric from their nitty gritty sheets a few years ago. With those metrics two teams who played exactly the same schedule and performed exactly the same way could be evaluated very differently simply based on where their easy and hard games fell on the schedule. A smart move to remove them in my opinion. A front loaded schedule with an easy coast to the finish should not be an advantage to gaining a berth or seeding in the tournament.
I agree it was smart to remove the Last-X metric for reasons you stated.
Another measuring device I question, or wonder how they evaluate it, is road/neutral record. Although I agree it's good to be able to win away from home, do they analyze where non-home games occur, i.e., the opponent strength in those games? A team could have an impressive record in road/neutral, but maybe a lot have come vs. marginal teams.
Iowa State has a mixed bag in that regard, snapshotting it today. ISU is 4-5 away, 3-1 neutral ... 7-6 combined, that looks good to me. But how much so? RPI-based, ISU's three top-50 wins all are away from home (at KU, at OSU, Miami-neutral), three of the 6 top-50 losses are away from home, but two neutral wins are sub-200 (Indiana State, Drake).
Just curious.
Of course. Like I said, imo it's more subjective now, which is great for us this year.
Specific to Iowa State: Obviously, winning as much as possible is No. 1 priority (always the case). But getting indirect help wouldn't hurt, either, to at least balance any additional missteps that might happen down the stretch.
A few examples to want, RPI-wise:
* Miami staying in top-50
* KSU getting back into top 50 (and TCU staying in top 50, as long as it isn't at head-to-head expense).
* Vandy climbing into top-50
* Iowa getting back into top-100 and staying there (yeah, *ugh* to wanting that).
Impact of all that is probably minimal at this point, I mention it only as fringe-assistance, until we directly enter Lock status.
A few examples to want, RPI-wise:
* Miami staying in top-50
* KSU getting back into top 50 (and TCU staying in top 50, as long as it isn't at head-to-head expense).
* Vandy climbing into top-50
* Iowa getting back into top-100 and staying there (yeah, *ugh* to wanting that).
Impact of all that is probably minimal at this point, I mention it only as fringe-assistance, until we directly enter Lock status.
