Current RPI

[ecasadoSBU]

You really think teams are holding back in any situation except the final couple garbage minutes against a truly terrible opponent? Teams already beat those teams by huge totals.

But I really don't see how you can say there's no difference between two teams with significantly different efficiencies.

I simply think that judging team by point differential opens a can of worms.

A lot of times, teams empty their bench if they are winning by double digits but you wouldn't see this happening if PD mattered. For sportsmanship sake I don't think is right to base how good team is based on point differential.

I think the point that I'm really trying to get across is that I understand the RPI is far from perfect, but any system based on a formula (including BPI, KenPom) will be fan from pefect. So why not keep things as simple as possible.

Win your games,
Play teams that win a lot of games,
play teams that win a lot of games against teams that win a lot of games

It a simple formula that values victories over everything else. Isn't that what basketball is about? Winning above all

Well, this sort of proves my point, since I think one problem is that Hawaii is probably under-rated in the RPI because of its methodology. Hawaii is 55 places higher in Kenpom and 62 places higher in BPI.

For a Top 50-100 team, Binghamton (KenPom 336/BPI 342) isn't comparable to (underrated by RPI) Hawaii isn't the right, but 6-7 Byrant (KenPom 201/BPI 218) or 7-8 BU (KenPom 202/BPI 195) or (sigh) 7-10 South Florida (KenPom 213/BPI 206).

If Bryant played Binghamton, Bryant wins...but to a top team, there's hardly a difference, yet the RPI, in particular, makes distorts reality as if there were...I mean, merely playing a low level team hurts you deeply.

And, look, you've presented your opinion fairly here, but let's not pretend there isn't something here biasing you. The Big East looks absurdly good by the RPI's measure. Also, according to the RPI, a 13-4 Providence (which lost to Brown [RPI 226/KenPom 244/BPI 264] is the 13th best team in the country (KenPom has them 50, BPI 42).

According to RPI, your team is better than

Louisville (RPI 23/KenPom 9/BPI 11)
Maryland (RPI 17/KenPom 19/BPI 17)

And a host of other schools (UNC, West Virginia, etc.) that I don't want to include the numbers for... including UConn (RPI 63/KenPom 30/BPI 36).

The RPI is a crock.

And as for accounting for who plays in the game...of course they should factor that in...and the NCAA always has.

tzznandrew, I think you have a valid point that Hawaii may be underrated by the RPI currently but don't you think that by March its RPI rank would be a more accurate reflection?

I think the difference between BPI/KenPom and the RPI is that the first two provide more accuracy on smaller sample size but when March comes around I think the committee does a pretty good job of choosing which teams belong/do not belong and the RPI is fairly accurate.

For example... UConn may be RPI63 right now, on the outside looking in. In my opinion, I think the team belongs in the NCAA and its likely that by March, its RPI rank will reflect that even if it doesn't reflect it right now.

Lastly, I just wanted to clarify that I have no allegiance to Providence. I do support the Big East as a whole but the only two individual teams that I actively root for are Stony Brook and Connecticut. I'm probably not free from biases, but my viewpoint has always been the same regarding this... even before conference realignment kicked Connecticut out of the B.E

All of I wanted to do was have an honest discussion about the RPI because I really do think that it is a fair and simple way to judge who's in/who's out when March comes up. I really value wins over everything else but I'm sure a lot of people think differently about this

 

[nyhusky89]

Tulsa's RPI would be a lot worse if their loss to Division II SE Oklahoma State counted. Let's all be thankful it doesn't.

Dont know if this has been said, but thats actually what gonzaga does. instead of playing 200+ out of conference, they play D-II teams which dont count against the RPI.

Thanks for the correction. For some reason I totally forgot about this. I remember reading that it indeed weights away wins higher than home wins.

Regarding your second point. Is it fair to say that 1-15 Binghamton (RPI 345) is the same as playing against 11-5 Hawaii (RPI 206)? ... I doubt it. I totally disagree.

Also I don't think whether a specific player is playing or not should be taken into account. It makes the formula unnecessarily complex and how can you really tell the impact an specific player can have on a game against that specific team. Our best bet is to keep that out of the formula. You either win or you lose, and your chances of getting to the NCAA should be based solely on wins/loses not who plays that game...

I also dont think its necessary to take into account the amount of points a team wins by. This can be very misleading and this doesn't say anything about how good a team is compared to another. If you add that to the formula You will have teams blowing out other teams just for the sake of boosting the ranking

The RPI is the fairest method that the NCAA could qualify teams without falling into pit falls. I would leave it the way it is

First of all the RPI is awful. That isn't a debate. Ken Pomroy would not have a job if the RPI was a worthwhile predictor of future results (i.e. who is better). They use the RPI specifically because it doesn't use margin of victory because of the scandal at BC, in 1980. Math has come a long way since 1981 and that should be reflected in the rankings. Unfortunately, the RPI is a far too simplistic way to look at things; it vastly overstates strength of schedule to the point that just playing a tough schedule will often make a team look good regardless of wins and losses. It fails to account for who plays, the pace of the game or the margin of victory, all critical factors in determining who the best team is.

Now specifically to your claims. You don't think margin of victory is important? Or who plays? Do you not look at pace adjusted stats either? Come on man. The rankings should be the most accurate possible reflection of each team the day the season ends, which means it absolutely matters how much they win by and how fast they play and who played and where they played. You are right about one thing, margin of victory in a single game means nothing, however over the course of a season it means a great deal. It also matters because schedules differ in strength. If a team plays a weak schedule but dominates, RPI will have them appreciably lower than a team that plays a good schedule and eeks out games. RPI sees two teams winning and judges them based on who they played. Kenpom/Massey judge them on who they played and how the beat them. If you don't think margin of victory is helpful i don't know what to tell you.

As far as players go, does it makes sense to you that Rutgers beating Wisky without FK should be looked at the same as if they beat them with him in the lineup?

 

[Jluc]

I prefer the eye test.

 

[Leebo]

I think the point that I'm really trying to get across is that I understand the RPI is far from perfect, but any system based on a formula (including BPI, KenPom) will be fan from pefect. So why not keep things as simple as possible.

You can't just call both of these systems "far from perfect" and expect us to believe it.

RPI is "far from perfect" the way the idea of a flat Earth was far from perfect. It makes sense only if you don't look at the details.

BPI and kenpom types of systems are "far from perfect" the way a globe is. It's not an exact representation, but it's not nearly as dumb as thinking we live on a flat Earth.

 

[7774]

I know you're a big fan of the Big East, but you're really homering it up in here with things that make no sense.

A lot of times, teams empty their bench if they are winning by double digits but you wouldn't see this happening if PD mattered. For sportsmanship sake I don't think is right to base how good team is based on point differential.

Point differential matters now, in that it affects people's perception of a team(i.e. people started thinking Kentucky was an unbeatable juggernaut when they beat Kansas by a bajillion points. Now people have more doubts about Kentucky because they are barely winning in OT), so I highly doubt much would change. Most teams empty their bench in the last minute only, if that.

I think the point that I'm really trying to get across is that I understand the RPI is far from perfect, but any system based on a formula (including BPI, KenPom) will be fan from pefect. So why not keep things as simple as possible.

So because all the metrics have flaws, we shouldn't try to find the most accurate one and use it? Because of simplicity? Lets just rank everyone by winning percentage then, it's even simpler than the RPI.

It a simple formula that values victories over everything else. Isn't that what basketball is about? Winning above all

Winning is the thing that is most valued in all these calculations.

I think the difference between BPI/KenPom and the RPI is that the first two provide more accuracy on smaller sample size but when March comes around I think the committee does a pretty good job of choosing which teams belong/do not belong and the RPI is fairly accurate.

If you look at the documentation the tournament committee provides which details how they select the teams for the tournament, they specifically say that RPI is only 1 of many statistical tools they use in their analysis, so they are most likely looking at BPI and KemPom, as well as other stuff. It's pretty silly to focus on RPI just because it's most closely tied to the selection process in people's minds.

 

[AtlHusky]

It a simple formula that values victories over everything else. Isn't that what basketball is about? Winning above all

But RPI doesn't value a team's victories above all else. A team's winning percentage is only 25% of the formula. I understand you have an opinion here, and it's perfectly acceptable for you to express it, but that doesn't mean you're not dead wrong.

 

[captainron2]

Well, this sort of proves my point, since I think one problem is that Hawaii is probably under-rated in the RPI because of its methodology. Hawaii is 55 places higher in Kenpom and 62 places higher in BPI.

For a Top 50-100 team, Binghamton (KenPom 336/BPI 342) isn't comparable to (underrated by RPI) Hawaii isn't the right, but 6-7 Byrant (KenPom 201/BPI 218) or 7-8 BU (KenPom 202/BPI 195) or (sigh) 7-10 South Florida (KenPom 213/BPI 206).

If Bryant played Binghamton, Bryant wins...but to a top team, there's hardly a difference, yet the RPI, in particular, makes distorts reality as if there were...I mean, merely playing a low level team hurts you deeply.

And, look, you've presented your opinion fairly here, but let's not pretend there isn't something here biasing you. The Big East looks absurdly good by the RPI's measure. Also, according to the RPI, a 13-4 Providence (which lost to Brown [RPI 226/KenPom 244/BPI 264] is the 13th best team in the country (KenPom has them 50, BPI 42).

According to RPI, your team is better than

Louisville (RPI 23/KenPom 9/BPI 11)
Maryland (RPI 17/KenPom 19/BPI 17)

And a host of other schools (UNC, West Virginia, etc.) that I don't want to include the numbers for... including UConn (RPI 63/KenPom 30/BPI 36).

The RPI is a crock.

And as for accounting for who plays in the game...of course they should factor that in...and the NCAA always has.

Of course the eye test says Providence isn't better than Louisville or Maryland, BUT, they have the 10th toughest SOS in the country and have played 10 top 100 teams and won 8. That's why despite the loss to Brown their RPI is so strong.

 

[auror]

If you look at the documentation the tournament committee provides which details how they select the teams for the tournament, they specifically say that RPI is only 1 of many statistical tools they use in their analysis, so they are most likely looking at BPI and KemPom, as well as other stuff. It's pretty silly to focus on RPI just because it's most closely tied to the selection process in people's minds.

I agree with all your other points, but just because they SAY that RPI is only one of many tools, doesn't mean it's not still too invasive in the selection process. The teams are grouped by RPI and "good wins" and "bad losses" are derived from RPI, and the nitty gritty sheets given to the committee show things like top 100 RPI wins, etc. It's still a part of everything.

 

[7774]

I agree with all your other points, but just because they SAY that RPI is only one of many tools, doesn't mean it's not still too invasive in the selection process. The teams are grouped by RPI and "good wins" and "bad losses" are derived from RPI, and the nitty gritty sheets given to the committee show things like top 100 RPI wins, etc. It's still a part of everything.

Agreed. At least they only look at it towards the end of the year when it's most accurate.

 

[ecasadoSBU]

Fair enough. I like all your points and I don't fear being wrong. Good to learn from the folks that have probably been following this for much longer.

This discussion has increase my interests in the way these rankings work. I'm doing more research on the accuracy BPI and KenPom systems. I'll probably be convinced by the time I'm done... lol