The RPI is one strange animal...

[maroonmania]

Just checking the recent updates on realtimerpi and trying to figure out how Ole Miss can take a 32 point thumping on their home floor to a team with a lower RPI, have only a 10-7 record and yet still maintain a 67 RPI with the double arrows heading upward. Here's another interesting one, South Carolina (89) still sits 3 spots ahead of us (92) with a 12-4 record and a 201 strength of schedule while we are at 12-5 with a 104 strength of schedule. So while you might say SOS is the sole reason for Ole Miss's RPI it seems to mean little when comparing MSU with South Carolina.

 

[muddawgs33]

There's just more that factors into your RPI other than SOS. Like losing at home hurts your RPI more and we have lost 2 home games to teams with a RPI of 144 and 204 and a combined record of 16-18. Msu has won 3 road games but 2 of them had RPI's of 188 and 226. USC has lost 1 home game and it was to a 16-1 Clemson team with an RPI of 8. They have won 2 road games and 1 over a 33 RPI team in Balyor. The fact is that losing to 5-11 Charlotte with an RPI of 204 at home is what is killing our RPI right now.

 

[Hanmudog]

Seem like the formula needs a little tweaking. There is really little difference in a team ranked 300 and one ranked 200 yet playing them kills your RPI. There should be some kind of sliding scale once a team gets past a certain ranking. Other than the win at home against Arkansas, Ole Miss has no wins of any note. State on the other hand has beaten WKU, Houston, Alabama, Vandy, and Arkansas. The formula is letting those early games against poor teams affect our ranking a bit too much.

 

[UpTheMiddlex3Punt]

The RPI doesn't seem to be a good indicator of performance until much later in the season. In mathematical terms, we are nowhere near convergence.

 

[RebelBruiser]

RPI is a simple formula:

25% of your rating is based on your winning percentage.
50% is your opponents' winning percentage (not including games against you).
25% is your opponents' opponents' winning percentage (not including games involving you).

Also, when you see SOS listed, they are just adding the second two components I listed above related to opponents' record and opponents' opponents' record.

And depending on the RPI calculation, they factor in home and road differently. I believe on most they count road wins as 1.4 wins and road losses as 0.6 losses. Home wins count 0.6 wins and home losses count as 1.4 losses.

So, let's say you play 5 games, 3 at home and 2 on the road. You go 2-1 at home and 1-1 on the road. Your record is 3-2. However, your RPI record is 2.6-2.0.
2 home wins x 0.6 = 1.2 wins
1 road win x 1.4 = 1.4 wins
Total of 2.6 wins

1 home loss x 1.4 = 1.4 losses
1 road loss x 0.6 = 0.6 losses
Total of 2.0 losses

Neutral site games count as 1 win or 1 loss. And I believe some calculations have a Semi-Home and Semi-Road calcuation where the values change to 1.2 and 0.8 instead of 1.4 and 0.6. For instance, when we played Southern Miss in Biloxi, it was a Semi-Home game for USM since it wasn't on their home floor, but it was essentially a home venue for them. So, in beating us, they got 0.8 wins, and in losing, we got 0.8 losses.

I also believe that some formulas have some bonuses and such that you get for quality wins, etc., but I could be wrong about that. The most important thing to remember is that it obviously gets more accurate with more data. Right now, I guess our schedule or something is keeping our RPI relatively high, or the location of our losses. Who knows? It won't stay there long. We'll be outside the Top 100 eventually.