Buy/Sell: Too much weight is given to RPI

[Hanmudog]

Ole Miss has the highest RPI in the West yet they are easily only the third best team in the West. I keep seeing where many of the talking heads have Bama not making the NCAA tourney but have UT, UK, and Vandy as locks yet Bama beat UK and UT head to head and had a much better conference record.

It just seems like every year the committee is looking less and less at how teams finish or things that occur during a season like injuries and suspensionsand focusing more on just the RPI. It is almost as if a bad start dooms your entire season.</p>

 

[57stratdawg]

It's a flawed system.

If Michigan State gets in after losing all year, I'm going to be pissed.

 

[skeptic]

I know the SOS is important. I know the RPI is important. But those who point to the RPI and SOS as two different arguements for or against a teamare just not getting it.... are they?

Ifthe SOS is low thenthe RPI is already penalized. If the SOS is high, the RPI is consequently already given a bump. Those who argue that two equally RPI'd teamsare different because one's SOS is stronger has to then ignore the better record by that team with the weaker SOS required to strengthen the RPI. Either use the RPI alone or the Record + SOS.

http://mrsec.com/tags/Mississippi-State

The blind comparisons article abovebears this out. The author argues thatIlinois should be picked because it has similar (but weaker) RPI but a better SOS (but worse record.) The team who happened to have a weaker schedule, Butler, had a better record, but is stuck in a weaker conference and could not build a SOS to match.

I understand why they talk about each. But use the tool as it was designed to be used. Using both blows the entire RPI concept.</p>

 

[patdog]

75% of your score comes from who you played and only 25% comes from how you actually did against those teams. If they changed the formula to1/2 your record,1/3 your opponents record, and1/6 your opponents opponents record, then it would be a pretty valid ranking.

 

[wpnetdawg]

that SOS carries greater importance in the RPU calculation because of its greater weighting. The greater weighting is used because the greater variance that exists in winning percentage than exists in strength of schedule.

If you look at all the teams in the NCAA, the top winning percentage is .936 and the lowest winning percentage is .036. There is a range of .900 between the top and the bottom. The standard deviation is .189.

However, the strength of schedule for all teams tend to deviate toward the middle. The top SOS in the country is .622 and the bottom SOS is .367 giving us a range of .255. The standard deviation is .049.

SOS has about 3.9 times less deviation, but only counts 3 times as much in the RPI calculation. Consequently, winning percentage still is the most important factor in the calculation.

 

[8dog]

I think any formula that looks only at records its pretty stupid.

 

[RebelBruiser]

because I think it's both.

I know as MSU fans you don't like the RPI because it actually holds you accountable for games you play before January, and Stansbury's teams lately have crapped the bed big time in those games, which costs you late in the season when you're playing better.

You can't completely discount those games. They count. Should they count just as much? No, but they still count. Alabama is on the bubble because of what they did before January. You can't just write that off like it didn't happen. If you were going to do that, then they should just call that an exhibition season and only count conference games. I think the RPI could benefit from a weighting of games based on when they were played, but I wouldn't make it too dramatic of a weight, maybe split each schedule into 3rds and count the middle third with the same weight you do today 1.4 and 0.6 being the numbers used for wins and losses depending on whether they were home or road. Count the first third games as 1.3 and 0.5 weights and the last third as 1.5 and 0.7 weights. That might help a little, but I wouldn't make it too dramatic.

That said, RPI is a weak formula for any sport. It's too simplistic. It has more value the more games are played, so for baseball is has more value, but for basketball it's too simple to count it as your only measure.

I think the selection committee, however, doesn't use it as much as you think. They won't take one team over another just because they are ranked 5 spots ahead in the RPI. They look at the entire resume. They consider the what have you done lately measure. They consider injuries. They consider quality wins versus bad losses.

On that note, that's where I think RPI's best measure comes in, in helping determine good wins and bad losses. It's still not perfect, but if you lose to a team outside the Top 100, it's not a good loss. If you beat a team in the Top 50 or Top 25, that's a good win. Take Marquette as an example. It sounds like they may get in as the 11th Big East team, with an RPI of only 68, and only an 18-13 record.

However, when you look at their losses, they came to teams ranked 5, 7, 9, 11, 13, 20, 22, 23, 26, 31, 32, 62, and 91 in the RPI. Meanwhile, they have wins over teams ranked 9, 15, 18, 23, 82, 91, and 92. They're only 4-11 against the Top 50, but they've got a decent number of good wins to go with almost nothing you could consider a bad loss.

 

[RebelBruiser]

I think any formula that looks only at records its pretty stupid.

Just curious, but where do you get your other numbers if you don't use only winning percentage or some variation of it?

 

[8dog]

Get an intelligent selection committee without regional bias that can cover the various parts of the country and tell me who the best 68 teams are (I realize its not the best 68 but you know what I mean). Watch a ton of games. Tell me the best teams. There are enough people committed to college basketball nowadays and enough tv coverage that this seems possible.

I know its a pretty big undertaking, but its better than just relying on some math. Beating UNC at home and Beating Old Dominion at home amounts to almost the same thing under the RPI with the exception of the last 25%. While I appreciate SDSU is a good team, I don't think beating them should amount to the same thing as beating Ohio State.

And if a committee member really believes in the RPI, he can use it as a guide if he needs it.

Maybe its a better system than I give it credit for, but I've just never liked it.

If nothing else do this:

As someone pointed out above, if anything, people shouldn't analyze both SOS and RPI. You punish/reward teams twice for it when you dod that.

Quit drawing the line at top 50 wins. If I have 3 wins over teams 55, 57, and 59 is that worse than 1 win over number 45? A team's resume that is generally studied would say yes.

 

[patdog]

I know as MSU fans you don't like the RPI because it actually holds you accountable for games you play before January, and Stansbury's teams lately have crapped the bed big time in those games, which costs you late in the season when you're playing better.

that Stans has lost all those Dec. games than we are that they count in the RPI. I do think that this year, the Dec. losses shouldn't hurt us as much because we didn't have our full lineup available until January. But generally speaking, hell yes the Dec. games should count. If Stans is going to continue to insist on scheduling such a weak non-conference schedule, he'd damn well better go something like 12-2 or better.

 

[drunkernhelldawg]

I am very skeptical that the Big East teasm are much if any better than teams from other conferences. The committee's reliance on RPI and preseason rankings has caused the dance to lose some credibility with me. The bottom line is that those teams play each other and their RPI doesn't suffer win or lose. I know that we could beat them about as often as we'd lose to them, but we'll never get the chance to show it because the dance is rigged by a small group of broadcasters, writers, and coaches. I think it's a bunch of crap that you can lose your chance for the dance by losing any of your first several games. Sports is not just about winning every game; it's about competing, getting better, and the getting even better than that when the big game is on the line. RPI is great for making money but it does not enhance competition and it certainly doesn't have anything to do with fairness. So I actually hate it. We've won four of five but Big East teams with losing streaks are going to the dance while we and the rest of our league stay home. We're bad as a league, but we're not that bad. Put the SEC into the Big East and I bet our teams would be at least fifty fifty in the top ten there. But screw it, I'm done.

 

[RebelBruiser]

and any committee is going to be looking for measurables to help rate teams.

After the selection committee basically opened their doors to guys like Lunardi and others to let them in on how the process works, I don't think you can get a whole lot better than what they do.

Maybe you could tweak the formulas some like I said, but I don't think you can just throw out numbers and solely use the eye test. You have to have measurables of some sort to help you sort teams. You end up with around the Top 50 teams in the tourney, and you probably have 70 or so teams on the list for those spots.

The fact that Alabama would even be under consideration shows that they don't use RPI as heavily as the average fan perceives. It's just one measuring tool among many, and I believe the eye test plays a role too. It's not just pure numbers. If it was, you could just make out your bracket based off the RPI. That's not the way it works. Since they only use it as a tool, not a final decision, I don't think it's that bad. Tweak it some to weight more heavily what you do late, sure, but you can't just eliminate all measures and go purely on the eyeball test when there are over 300 teams out there playing varied schedules in varied conferences.

Ultimately, it really doesn't matter all that much. The tourney is big enough that what happens on the bubble has zero effect on what the tourney is trying to accomplish, which is trying to make sure they include every team that might have a claim at being the best in the nation. They do that, and the bubble teams rarely make it past the first weekend and pretty much never make it past the second, so the argument on the bubble ultimately becomes about a banner, not whether or not your team may have had an argument for the national title.

It's not like the BCS that leaves out an undefeated TCU team or an undefeated Auburn team in 2004 or undefeated Utah teams a few times. There are teams in the BCS that have legit claims every year to being the best in the nation, but not in the NCAA tournament. If you have any remote claim, you will be given your chance to prove it.

 

[RebelBruiser]

If you were to look at an RPI before the season, every team would be ranked No. 1 with a 0.000 RPI.

The formula populates itself as you play based on how you do and how the teams you played against do.

And I disagree with your premise. I've watched enough Big East basketball to see that those teams are way better coached than the crap we see in the SEC. The AAU culture still sneaks its way in, but they play a much better brand of basketball than we do in the SEC. A few select games SEC teams can and do win against the Big East, but that league is much deeper and just better overall.

 

[8dog]

don't watch nearly the basketball Im talking about. Plus, as ADs their bias is pretty strong.

I want 20-30 guys that have nothing to do but watch hoops all year. Pay them generously (I think they can afford it).

Keep the RPI if you must but I do think it gets way too much emphasis.

 

[Hanmudog]

For example, State is a better team than Ole Miss yet theRPI has Ole Miss about 20-30 spots ahead of us. It does not factor in that we were without Bost and Sidney for the first part of the season or that we beat Ole Miss twice head to head. The problem is that an objective observer that did not follow our season closely might would lazily look at the RPI and assume that Ole Miss is the better team. Playing a tougher schedule does not mean a team is necessarily better.

I just used State and Ole Miss as an example but the same could be said of Alabama when comparing them to UT and Kentucky. I am just worried that too much weight is being given to the RPI and not enough committee members look atthings that happen during the season.