PDA

View Full Version : RPI Analysis



curmudgeon
04-23-2014, 09:22 AM
Gentlemen,

When all results were in last night, we sit 32nd in the RPI. I believe that we still have a chance to host. Last year was a fluke, as it is the only year I ever remember that the actual Top 16 teams in the RPI hosted. Most years, there is a team in the high 20s and even low 30s that hosts.

Let's geek out a little bit.

While most consider "our RPI" as 32, our RPI is actually 0.5626. The #25 team in the RPI is Old Dominion, at 0.5699

Taking 2 of 3 from everyone from here on out is the minimum that we need to do to get into the hosting discussion, which we will look at below. Here, I wanted to show what taking 2 of 3 and sweeping A&M would do for our RPI.

First, lets look at taking 2 out of 3.

This would improve our RPI to 0.5710. If everyone in front of us held steady or dropped, our ceiling would be 25th. Of course, teams ranked 20-24 could fall and open up a slot or two, but taking 2 of 3 would be a high of 25 and a low of 32.

Now the sweep would improve our RPI to 0.5772. A sweep gives us a ceiling of 20th.

Again, this is not saying if we take 2/3 we will be 25th and if we swept we will be 20th. It says that is what is possible.

Now, a look at the teams ranked 20-31 that we need to jump.

Teams ranked 25-31
31. Tennessee (at LSU)
30. Mercer (at Florida Gulf Coast)
29. West Virginia (Marshall, Kansas State)
28. Georgia Tech (at NC State)
27. Stanford (Cal)
26. Miami (at FAU, at Clemson)
25. Old Dominion (at MTSU)

Mercer and Old Dominion probably need to sweep to hold serve. A loss to one of those sub-100 teams would hurt them badly. They are both on the road, so they are likely to drop at least one.

Tennessee might pick up 1 against LSU, but that would not offset a sweep of A&M by MSU. If we were to take 2 and UT takes 1 from LSU, they would probably stay ahead of us.

West Virginia, Georgia Tech, Stanford and Miami all have similar tasks to us this week, playing a team ranked between 60-80. Miami and Tech are on the road, so we do not need them to sweep a road series.

Now, a look at teams ranked 20-24. These will be hard to catch this week, but worth watching because we will have to catch them.

24. Louisville (Morehead State, Connecticut)
23. Oregon State (Oregon)
22. Oregon (at Oregon State)
21. Maryland (at Boston College)
20. Indiana State (at Butler, Dallas Baptist)

Louisville has five games against really bad teams this week. Going 5-0 will probably not help them that much. If somehow they went 3-2, they are passable this week with a sweep. Oregon/Oregon State winner will move into host position. If one of them were to sweep, the loser could drop but not too far. Maryland losing 2/3 to BC would be huge. Indiana State will probably hold serve, as Dallas Baptist is a top 35 RPI team.

So what about hosting? Last year, there were about 18 teams at most competing for 16 host spots. This year, its a little more up for grabs.

First, the eight locks:
Florida, Florida State, Virginia, South Carolina, Indiana, Texas, Rice, LSU

And four probable locks:
Cal Poly, Ole Miss, Louisiana-Lafayette, Washington

National seeds are bolded

Four more spots and here are the teams in the hunt - my order.
Alabama (but dropping rapidly)
TCU (almost a lock as co-leaders in the Big 12)
Louisville (lower RPI, 3 GB in weaker conference)
Oregon/Oregon State winner
------------
Kentucky (7th place in conference)
Vanderbilt (8th place in conference)
Houston (needs to start winning, dropping rapidly despite top 10 RPI)
Texas Tech (low conference place)
Mississippi State (low RPI)

There are two things working here. Not only do we have to pass some teams nationally and have some teams lose. We have to finish in the top three of the SEC. I have five SEC teams hosting, and two more on the waiting list. If TCU and Houston were to drop, two non-SEC teams would take their place. So, instead of watching teams like TCU and Houston, Lousville and Texas Tech, we have to have teams like Ole Miss, Kentucky and Alabama start losing at the same time we are winning.

Ole Miss goes to Kentucky this weekend. It probably does us more good for Ole Miss to take 2 of 3, even though it strengthens their bid, we are closer to Kentucky's bid. Another argument is that if we end up head to head with Ole Miss for a host spot, we went 2-2 and we draw great crowds. Ole Miss is going to have a higher RPI and they draw great crowds as well - trust me we will lose that battle. We have a better chance if we finish 2-3 games ahead of Kentucky.

Alabama is at South Carolina. South Carolina needs to win that series without sweeping. Bama is going to keep dropping in future weeks. We don't need USC to start running away with second place.

BulldogBear
04-23-2014, 10:10 AM
I don't really know how all that RPI mathematics formula works, etc.

Just trying to understand. So, we host Tennessee in a couple of weeks. Would we not want them to continue to have success and therefore be a higher RPI team when we play them and perhaps take 2 or find a way to take 3? Or does their hypothetical success against LSU offset this? I guess my question is, how much role does head to head play factor in RPI?

esplanade91
04-23-2014, 10:21 AM
At this point if we host we're headed to Charlottesville for a rematch with those pompous asswipes who are currently ranked #1.

curmudgeon
04-23-2014, 10:34 AM
Tennessee having success against LSU helps us on two fronts:

1. We are a half game behind the Corndog Bastards and they own the tiebreaker.
2. It keeps Tennessee as a Top 30/possibly Top 25 RPI team so when we have success against them, it helps, but that help might be offsetted by the damage it does to LSU since they swept us. If Tennessee were to take 2/3, and we were to sweep, then we would see a positive gain.

Like I said, you almost have to ignore RPI at this point and trust that it will take care of itself (it will). If we end up winning the West, then we will finish first or second overall in the SEC. With LSU, South Carolina and Ole Miss pretty much being locks to host, they would not pass the second place team to take three teams. If we were to finish third, they might pass us to get two, and if we were to finish fourth, they would definitely pass us to get one.

I would consider us a 100% lock at hosting if we finish first, and an 80% lock if we finish second in the SEC. That's what we have to shoot for, and we have the schedule to do it. Third place is a crapshoot and fourth or lower is a trip to Charlottesville.

We are the 2014 version of 2013 Arkansas. They finished third in the SEC with a RPI at #30 and both us (#5 in SEC) and South Carolina (#4 in SEC) hosted over them. If they would have finished second, they would have hosted.

BulldogBear
04-23-2014, 10:35 AM
Or another question would be:

If SOS factors in at 2/3, how much does W-L against said SOS factor. Suppose you got the toughest schedule but go 20-36 against it. SOS or not, you got no business in the postseason. I see this reasoning in football sometimes too. It's better to lose 45-0 to Alabama than beat Troy 45-0. Sure it looks great IF you play well against the tough opponents but if you get murdered why are you getting credit for that?

engie
04-23-2014, 10:45 AM
I don't really know how all that RPI mathematics formula works, etc.

Just trying to understand. So, we host Tennessee in a couple of weeks. Would we not want them to continue to have success and therefore be a higher RPI team when we play them and perhaps take 2 or find a way to take 3? Or does their hypothetical success against LSU offset this? I guess my question is, how much role does head to head play factor in RPI?

Head to head doesn't count for much.
RPI = 1/3(home/away weighted WP) + 1/3(opponents' home/away weighted WP) + 1/3(opponents' opponents home/away weighted WP)
33% win percentage
33% opponents win percentage
33% opponents' opponents win percentage
Thus, 2/3 of RPI is SOS regardless of whether you win or lose. You only "control" the 33% in the win percentage column.

Home games count as 0.7 win and 1.3 loss -- attempting to account for "home field advantage"(Boyd and others have shown where this weighting is ridiculous for baseball -- and is the primary reason for him formulating ISR and Nolan formulating NPR). Road games are the inverse -- 1.3 win or 0.7 loss. Neutral site games count normally.

Basically, when it comes to SEC games, we need to pull for all the teams we play -- when they are playing the teams that we don't. Otherwise, in series like Tennessee vs LSU, it's ultimately a wash for us at the end of the season...It hurts the same amount it helps regardless of outcome... Note, I'm talking strictly in terms of RPI here...

engie
04-23-2014, 10:48 AM
Or another question would be:

If SOS factors in at 2/3, how much does W-L against said SOS factor. Suppose you got the toughest schedule but go 20-36 against it. SOS or not, you got no business in the postseason. I see this reasoning in football sometimes too. It's better to lose 45-0 to Alabama than beat Troy 45-0. Sure it looks great IF you play well against the tough opponents but if you get murdered why are you getting credit for that?

The manipulation of RPI like you are describing is why teams have to be .500 mandatorily to make the postseason...

There have been a number of years where Florida was right around .500 with a top 10 RPI -- because they play a ridiculously tough SOS every year with FSU, Miami, and all the other in-state schools... Fall below .500 -- and you are out. Manipulate it and manage to stay above .500 -- and you are in. That said, RPI can say you deserve to host -- but in that instance, you wouldn't get a bid...

curmudgeon
04-23-2014, 10:49 AM
Stanford is good to look at. They have a losing record but have played the 2nd toughest schedule, so they are in the top 30 of the RPI. They will not make the postseason if they have a losing record as there is an NCAA rule that you can't make the postseason with a losing record, unless you win your conference's auto bid.

Stanford will be rewarded with a regional bid if they can get above .500. Remember, RPI is just one of the components (although it is the biggest). Conference strength, conference finish, performance in last 10 games all play a role in the selection.

Look at a team like Tennessee. RPI is 31st. That's the second weakest #2 seed on a straight S-curve. Then you factor in that their record is 54th best against the 33rd ranked SOS. That drops them to a high 3 seed in the committee's mind. Then they are 5-5 in their last 10, that drops them a little lower. Then they are 12th in their conference and they become a questionable bubble team.

vs. a team like Mississippi State. RPI is 32nd. That's the weakest #2 seed on a straight S-curve. Record is 51st against the 31st hardest schedule. That might move them a little but not much. 7-3 in the last 10 - moves them up a little bit. 4th place in the toughest conference. Solid 2 seed.

See how it works?

The reason RPI doesn't work well in football is the smaller amount of games. Remember early in the baseball season you see teams like Bryant ranked in the Top 5. With only 12 games, that's why there are different formulas.

The Croom Diaries
04-23-2014, 11:04 AM
You have six SEC teams in Regionals, not five. Just sayin.

Why are you so down on Kentucky? I would think they'd be more of a lock than LSU right now 9 spots ahead in RPI. They both have a 200+ loss. UK does have two more 100+ losses but their SOS is 3 and LSU is 50.

BulldogBear
04-23-2014, 11:04 AM
Thanks engie, carmudgeon. That clears it up some, at least for a layman.

curmudgeon
04-23-2014, 11:17 AM
Kentucky's problem is simple. 7th place in the conference. They move to the top 5 and they can host.

Sorry about the six teams. That changes things. I'm editing it. SEC will not have more than five hosts under any circumstances.

curmudgeon
04-23-2014, 11:26 AM
Essentially Alabama, Ole Miss, Kentucky, Vandy and MSU are competing for at least one, probably two hosting spots. It would be hard to argue that we are any better than fourth in that group right now.

The Croom Diaries
04-23-2014, 01:14 PM
This is an interesting article from Aaron Fitt today on RPI: http://www.baseballamerica.com/college/rpi-analysis-requires-a-closer-look/

He says in there that it's 25% win percentage, 50% opp. win percentage, 25% opp. opp. win percentage.

engie
04-23-2014, 01:17 PM
This is an interesting article from Aaron Fitt today on RPI: http://www.baseballamerica.com/college/rpi-analysis-requires-a-closer-look/

He says in there that it's 25% win percentage, 50% opp. win percentage, 25% opp. opp. win percentage.

Then they have modified the formula slightly -- which wouldn't be surprising. It's changed 3 or 4 times over the last decade in baseball...