On the Perversity of the RPI (Part 1) | Syracusefan.com

On the Perversity of the RPI (Part 1)

Quazzum69

Stable Genius
Joined
Oct 27, 2012
Messages
3,925
Like
6,131
I'm sure some most of you are familiar with the Ratings Percentage Index, or the RPI. Over the next few weeks leading up to the NCAA Tournament, I will periodically write about some very peculiar features of the RPI and how it affects Syracuse. I am no Joe Lunardi (he's an expert, right?), but just a curious investigator looking for some discussion.

I've calculated the weighted win percentage portion of the RPI (accounts for 1/4 of the RPI) for all possible outcomes of the remaining nine conference games. There are 20 (or 4 x 5) possible outcomes for each combination of home wins and losses. Keep in mind that which teams we beat has no effect on the RPI - it's just how many wins or losses.

If Syracuse were to go 3-1 in home games but 0-5 in away games, their weighted win percentage (which is 1/4 of the RPI, remember) would be 0.825. Now, if we were to go a very impressive 5-0 in away games but a pathetic 0-4 in home games, the RPI would be 0.823! Similarly, going 5-0 away and 2-2 at home would produce 0.907 but going 3-2 away and 3-1 at home would produce 0.906.

You can already start to see how teams/conferences can quite easily maximize the RPI via scheduling. If we go 2-3 on the road (likely wins @ FSU and MD; toss-up games at Pitt, Duke and Virginia) and go 4-0 at home (all gimmes, IMO), it will be equivalent to going 3-1 at home and 4-1 on the road.

It is astonishing that only losing to Clemson at home and Duke away would equivalent to if we win all home games but lose to Florida State, Maryland and Pitt away.

Clearly, it is much more important to win at home than to win on the road (or lose on the road than lose at home, if you prefer) so it is better to have easy home games and hard road games. Having the weaker teams at home and harder teams away would produce a higher expected RPI, I believe. This seems odd, doesn't it?

Anyways, people should stop worrying about the road games and its effect on our seeding. As long as we take care of business at home, we'll be fine.
 

jncuse

Living Legend
Joined
Feb 19, 2012
Messages
16,510
Like
27,089
http://www.rpiforecast.com/wizard/index.html

Just use the RPIwizard. Basically, you enter all our results over our remaining games (W or L), and the rest of the country's games are projected as well. The app will spit out what our projected ending RPI and SOS is.

I am not too worried about our own individual RPI. It should not make a difference in our seed.

I would worry about the conference RPI and how it affects me, if I was a team on the bubble like Clemson and Florida St. The ACC is probably a little bit better than the RPI says it is. (merely comparing KP to RPI). It will result in extra bad losses, less top 100 wins. Things that hurt bubble teams.
 

jncuse

Living Legend
Joined
Feb 19, 2012
Messages
16,510
Like
27,089
Anyways, people should stop worrying about the road games and its effect on our seeding. As long as we take care of business at home, we'll be fine.

While your post did a good job showing the perversity of RPI, I believe your concluding point that I have quoted is inaccurate.

Your post is basically, "All we need to worry about is maximizing RPI" and the seed will sort itself out.

But seeding, especially on the top 2 lines, is not sorted by RPI. It is sorted by assessing top 25 wins, top 50 wins, quality road/neutral wins, top 50 record, any bad losses. Whether we are 5th or 7th in RPI is totally irrelevant in the discussion. Its the quality win profile / top 50 record that will matter most.

No matter the flaws in the formula, all top teams that look like #1 seed contenders will have good RPI's. Where they stand amongst each other in the RPI metric is mostly irrelevant to the committee.

All this being said, we have developed a nice cushion so we can likely get away with 3 road losses at the end, or 3 losses anywhere. But that would be the case no matter how the RPI impacts us.
 
Last edited:

Forum statistics

Threads
160,514
Messages
4,343,131
Members
5,639
Latest member
GoldyGopher

Online statistics

Members online
36
Guests online
424
Total visitors
460




Top Bottom