Imagine a conference where...

[Jamar Howard 4 President]

You are completely mashing two different concepts. My premise is that RPI itself is flawed because too much weight is placed on what a team has no control over, other than scheduling. I then proved that on average 69% of the final weighting is based on SOS -- which a team has no control over, other than scheduling.

The variance of Win Rank and SOS Rank is a completely different animal (that I showed only for interest). It has nothing to do with the weight.

So now the next step is to see how variance furthers your argument that the elite teams are properly selected and seated, which I will do when I get more spreadsheet time.

So far this statement still holds for the top 68 RPI spots:

On average, a WHOPPING 69% of RPI is based SOLELY on the strength of schedule.

More to come on your premise that the variance of SOS rankings makes the heavy weighting of SOS irrelevant -- but not until I get more down time.

I didn’t realize that you were trying to analyze these 2 categories separately. I assumed that you mentioned both so you could talk about both at the same time. I guess I misunderstood. I’m curious to hear your thoughts on variance.

Here’s the reason I jumped the gun and started talking about variance. In reality, I don’t think it is fair to separate weight and variance. Each one can be a complete distortion without also checking to see what the other tells us. Let me explain.

We aren’t really interested in a team’s RPI. It is some number between 0 & 1. What we really want is to rank each team by RPI and then look at RPI rank. This sounds like a meaningless point, but it isn’t. No one really cares what the RPI formula comes up with, as long as the order is correct. The actual value assigned to a team isn’t important. Right now it is between 0 and 1. If it was between 0 & 50, no one would care. The important thing is the relative rank of each team with respect to every other team.

What if every division 1 team was equal and no team was better than any other? Then, you could make SOS 99% of the RPI and it wouldn’t matter because everyone’s SOS would be the same. It would all be based on the 1% of the RPI that looked at wins & losses, and you would ultimately still be able to correctly rank teams relative to each other. The differences between RPI values would be very small, but there would still be differences, and the RPI rank would actually be 100% based on that small 1% value of wins & losses. In this analogy, 99% of the weight is based on SOS, but 100% of the final RPI rank is effectively based on wins/losses by the team in question! Obviously this analogy breaks down because if every team is equal, then every record would also probably be equal, but you get the point. Weighting only matters if there is variance within the area that is being weighed.

Now back to your example. You came up with 69% for the top 68 teams. I believe you would have come up with exactly 75% if you had used all D1 teams. Since you were simply re-calculating weight based on a subset of teams, you came up with a slightly different number. You choose the top teams, whose good winning %’s help make the SOS less of a factor. Had you choosen the bottom 68 teams, you probably would have come up with 81% instead of 69%. However you want to look at it, it is indeed true that roughly 75% of the actual RPI value for each team comes from SOS. However, as I showed in the paragraph above, this doesn’t mean anything until we know how much variance there is within this 75% and how much variance there is in the other 25%.

As I showed previously using your analysis of the top 68 teams:

Win Rank varies by 0.095
SOS Rank varies by 0.091
0.095 / (0.095 + 0.091) = 51%

This means that, on average, 51% of the difference you see in the RPI (value, not rank) from team to team is based on an individual team's winning %. If you want to debate whether 51% is too much/not enough control, fine. However, compared with your initial calcs, my 51% number gives a much more meaningful answer to the question "just how much control does a team have over their own RPI once the schedule is set?"

[Jamar Howard 4 President]

Bottom line is this:

http://espn.go.com/blog/collegebasketballnation/post/_/id/12511/scheduling-still-difficult-for-mid-majors

I may be a good hitter but only three at bats is not a worthy representation to show my ability.

Give me 12-15 at bats, then we will talk.

There is some truth to what you are saying. However, I hear many people complain about their favorite team not getting chances to prove themselves despite the fact that they lost 4 or 5 games to bad teams. A team that is going to be capable of proving themselves against good competition should also be capable of winning the majority of games against weak competition. Of the numerous teams that wanted a tougher schedule but couldn't get it, there were only a handful each year that actually took care of business in the games that they did play.

[Jamar Howard 4 President]

- BCS teams get an RPI boost by not having to play on the road in non-conf (False)

How in the heck do you assert this?

It is 180 degrees out of phase with RPI. RPI recognizes a substantial advantage to playing a home game.

The advantage is particularly good for BCS teams for many reasons:

Bigger stadiums therefore bigger crowds.
More media attention.
Not being stiff and tired from traveling.
Travel time which interferes with studies (putting external pressure on players).
Travel time interferes with practice and game study time.

I think it will be quite easy to show that teams with few away games fair much better than teams that travel a significant portion of their schedule.

And .. BCS teams have the money and scheduling privilege to demand home games.

As I said before, the RPI boost of playing on the road is greater than the actual added difficulty that comes with such a game. Since the RPI over-compensates for home court advantage, playing road games is actually better than playing home games (if we are looking at it from an RPI standpoint only) Here's an explanation:

Team's A and B both play a group of 20 teams. Every team involved is evenly matched, so it would be assumed that they should win 50% of their games at neutral sites. Home court advantage is to be considered 65/35.

Team A plays 20 games, all of them at home. As expected, they go 13-7.

Team B plays 20 games, all of them on the road. As expected, they go 7-13.

These teams should be considered equals, but the RPI doesn't see it that way. According to the RPI:

Team A's adjusted record is (13 * 0.6) wins & (7 * 1.4) losses = 7.8 wins & 9.8 losses = 44% winning %

Team B's adjusted record is (7 * 1.4) wins & (13 * 0.6) losses = 9.8 wins & 7.8 losses = 56% winning %

Both teams were equals, both teams played as expected based on home court advantage, and yet the team that played all the road games came out ahead. This is why I say that the BCS teams are not gaining an advantage (in the RPI) by playing all those home games. If anything, they are hurting themselves.

[lostshocker]

Both teams were equals, both teams played as expected based on home court advantage, and yet the team that played all the road games came out ahead. This is why I say that the BCS teams are not gaining an advantage (in the RPI) by playing all those home games. If anything, they are hurting themselves.

that's only part of the equation. they are offsetting the negative with positives to the other aspects of the equation. overall the net result is an advantage by playing all home games.

[Jamar Howard 4 President]

Could you please be a little bit more vague? :whistle:

[lostshocker]

Could you please be a little bit more vague? :whistle:

i know you're not as dense as you are trying to be. but if you wish to feign ignorance more power to you.

sad thing is the adjustment was added to try shift the results to what some people wanted to see. maybe we can get charlie epps to come up with an algorithm and it will only put haves into the tourney.

[Jamar Howard 4 President]

lostshocker, I'm not trying to be anything. I'm trying to have a logical conversation, and I really have no idea what you are talking about. Why are you so mysterious with your opinions?

How does playing a bunch of home games help a team's RPI? What parts of my post about the RPI formula overcompensating for home court advantage do you disagree with?