I use a formula I call "Wins Only" to rank the tournament. Every team you mentioned falls in exactly the same spot just outside the tournament. The cut line to make the tournament as of Monday was 58 pts with UNC-Wilmington getting the last seed on a tiebreaker with Louisiana (say thank you midweek win over Wake Forest).
The teams in your article Louisana 58pts, Kansas 56pts, College of Charleston 54pts, Xavier 52pts, Ole Miss 51pts and James Madison 51 fall into a cluster between 42 and 55.
The biggest snub I find in both RPI and DSR is South Alabama scores a 61 in wins only earning them the 26th best score and better than Oregon St.,Coastal Carolina and far better than Florida. But South Alabama has an RPI of 65 and a DSR of 68.
Why do we place value in losses?
I think a "wins only" type system works fine in a small, controlled population of teams (like MLB, or even a single conference), but when trying to compare 300+ teams and select 64 for a tournament it's not super useful. A team that goes on the road and goes 1-2, losing two 1 run games to a Top 5 team has arguably had a better weekend than a team that won 3 easy games against a 50 RPI. Ultimately wins matter but when trying to compare teams that some play head to head the way teams lose also matter.
@sean-miller The value of a Top5(quad 1) win is 4 pts. The value of a quad 4 win is 1 point. So the team that beat the top 5 opponent would get 4 pts and a better weekend than the team that swept, which would have 3 points.
To your specific example Top 5 vs. Top 50. I assure it's easier to beat #5 Clemson once at Clemson than it is to sweep #50 Texas anywhere. Or for specific in league comparison. It's easier to beat Kentucky once in Lexington than to sweep Auburn.
In the "wins only" system, who you beat matters, who beat you does not. With the schools I mentioned South Alabama is 15-11 vs. Quad 1&2 and has winning record against both groups. Florida is 12-19 and has a losing record against both quads. Yet Florida is at 18 RPI and South Alabama is at 58. It seems Florida is getting a huge benefit by getting drubbed nightly in the SEC while USA is getting buried for occasionally losing a conference game.
Good article.
I certainly think there's a lot of value in using muliple metrics and comparing them.
@joseph-davis To be clear, I'm NOT arguing that RPI is the answer. It's not. But I do think how you lose, and to whom, is important.
@sean-miller A simple question about your belief that who and how you lose matters...
Why? Why should Arkansas get more credit for beating Alabama once and losing to them twice than South Alabama gets for playing Alabama once and beating them? They both beat Alabama one time, they should both get 4 points.
My system does currently use the RPI quad system because that's the dataset available to me You get 4 pts for Q1 wins, 3 for Q2, 2 for Q3 and 1 for Q4, 0 for a loss.
@joseph-davis First, using Q1, Q2, etc is a good thing. That addresses the "who." If we're both 5-5 but you're 5-5 in the SEC/ACC and I'm 5-5 in the Summit those are two very different things
Second, yes, I think how you lose matters. If you play Arkansas, go 1-2 and your wins and losses are all 1 run results I would say you're very similar in quality to Arkansas. ...on the other hand I play them and win by 1 in a rain shortened game and then I lose by 15 runs and 20 runs I think we'd all agree that I got lucky once and got pummeled twice. We would both have 1-2 records vs the same team but everyone would agree we're in a different spot quality wise. Call it the "eye test" if you want, but I think it's very important.
I tried to read through the DSR equation over lunch. Haven't from had a chance to digest it as it's pretty complex but it seems like an interesting concept. Overall I'd tend to agree it's likely an improvement over RPI. But I do question two things...
(1) While RPI tremendously overvalues road games I think DSR may undervalue game location. Only applying a 4% value to location is too small in my opinion unless it's being used artificially low to offset RPI.
(2) I'm not sure I'm convinced that the very early season calculation of team strength is a good metric. Looking at average stats from the previous 3 seasons in a college sport with tremendous turnover strikes me as a real hope and a prayer. Let's say I played a 4 game set at Stanford to start the season this year. They're coming off 3 straight Omaha seasons with very good offensive production. But they're dog poo this year. I would get a huge boost in DSR that would be completely unwarranted.
I'll have to look at this more when I have the time but I am intrigued. But as Rooney said, no metric is perfect.
@sean-miller I don't think road records should count at all. While I understand it may be harder to win on the road than at home. Team A should be advantaged overr Team B because the conference scheduled Team C at Team B and Team A at Team C. Especially when you're talking about a 3 game series, it could be significant difference in value for the exact same outcome, a 2-1 record against Team C.
All individual samples are unique and hold no real value within themselves. If the ONLY data you have is the 3 games against Arkansas, it appears one team may be better. But in each of those series there were unique events that impacted the final outcome. The other 53 data points in the season will provide a more clear picture.
I use losses as the #4 tiebreaker between 2 teams. The first tiebreaker is Q1 wins, then Q2 wins, Q3 wins then Q4 losses. I use Q4 losses instead of Q1 losses because Q4s are games you should have won.
I agree 100% there is no value in a preseason calculation of team strength and it should not be used. A system will completely integrate itself in 3 weekends.
Is there a reason the DSR rank is no longer available on this site? Seems strange to just have it removed.
@campo118 It's still there. On the main D1 page go to the Rankings tab in the top nav bar, hover over it and the drop down will show the DSR.
Not quite sure what you mean by "road records shouldn't count at all." Are you saying we just wipe the slate clean on the road? or that they shouldn't be weighted at all? I think you're saying the latter.@sean-miller I don't think road records should count at all. While I understand it may be harder to win on the road than at home. Team A should be advantaged overr Team B because the conference scheduled Team C at Team B and Team A at Team C. Especially when you're talking about a 3 game series, it could be significant difference in value for the exact same outcome, a 2-1 record against Team C.
Either way I completely disagree. If you look at teams across the entire nation it's pretty clear it's much more difficult to win on the road and the outcomes back that up. The question to me is how, and how much, you adjust for that. Again, I think RPI overvalues being on the road (and undervalues wins at home), but on it's face the 4% boost in DSR seems too little to me.
And for your tiebreakers do you count Q1, Q2, etc. as a true Top 25/50/100/100+? Because the current Q system also gives HUGE boosts for road games that it didn't prior to this year. So when calculating your rankings did you use the new adjusted Q system? or a true Q like we used last year?
