6/4/2013 7:01 AM
I just looked at the Tark Projection Report (D3).  With 1 conference game to go, I'm sitting in 9th place, which seems reasonable.  What gets me is the team immediately above me.  My team is ranked 6th, RPI 7.  Wooster is ranked 21, RPI 15.  I can understand how this has happened.  Wooster has a lot of "solid" wins against opponents in the RPI 25-50 range, which I have to assume is bolstering their PR ranking.  But I thought the new system was supposed to reward quality wins.  As I said, Wooster's wins are solid.  But they aren't great.  Their best win was against RPI #21 at home; their best road win was against RPI #36, and they have no other top 25 RPI victories (they do have 5 losses against the top 25, and do have another shot to pick up a 2nd win in the final game this PM).  Their worst loss was at home against RPI #45, but to be fair this was their only loss outside the top 25 (19-6 overall).

Methodist, on the other hand, is 23-2.  Our worst loss was against RPI #33 at home by one point; our other loss was on the road at RPI #1 by 11.  We have wins against RPIs #6, 6, 11 (road), and 14 (road).  Again, I'm sure the reason they're ahead of us is because of the RPI 25-50 wins, of which I have only 1.  And I certainly don't mean to disparage warshot, who has had a solid season and did a great job of scheduling.  To be honest, if I'd known we'd have 5 sims in the conference this season I would have scheduled at least a slightly tougher non-con schedule.  We're still the #2 conference in Tark D3, but I was counting on an even stronger season of conference games when I scheduled.  All the same, this seems like a failure of the system to me.  I have 4 wins that are far better than any of his and he has more and worse losses than I do.  His net SOS is better, but my RPI is still higher.  Again, my fault for not scheduling better.  Ideally I would have liked a high 2-seed this year since my team isn't really good enough to beat any 1 seeds, or many other 2- and 3-seeds, and I could use a little scheduling help in the postseason to make a deepish run.  All the same, does this feel weird to anyone else?
6/4/2013 10:39 AM
These two D3 schools' profiles are so similar, in my opinion it seems unreasonable to call one being slightly ahead of the other in the PR a "failure of the system."
6/4/2013 1:09 PM (edited)
This has long been a point of concern of mine.  "Good" losses count for far too much credit.  I think where he really gets you is the top 100 wins/losses.  14-6 vs. 8-2.  Is 14-6 better than 8-2?  Certainly not from a win % standpoint.  

Full disclosure - USA South sucks.

P.S.  This is udm_mike, so you know why I think USA South sucks, dahs.

6/4/2013 11:51 AM
I've thought since literally my 2nd season of WIS that scheduling (and beating) 8-15, 200+ RPI (or worse, 300+) has a disproportionate negative effect. Of course, this is true in real RPI too. Any formula that thinks beating a terrible team is easier than not playing at all is a bad formula in my book. 

Although this might not be directly what's hurting you here, as volume of top 100 wins has been noted. 
6/4/2013 1:08 PM
I'm gonna plug these numbers into my formula and see what turns up....not for any of you really, I'm just intrigued
6/4/2013 1:21 PM
Rank W L Team T10
1
19 6 Wooster 308.0316
2
23 2 Methodist 296.57152


Their schedule is considerably better than yours...you've played 15 opponents with an RPI of 115 or higher...conversely, Wooster has played 4 opponents +115
6/4/2013 1:33 PM
I think part of the problem is the difficulty in translating one set formula across the power spectrum.  For two teams in the 20s or 30s I guess I would probably want to put a significant emphasis on games against 25-50 or 50-100.  Certainly games against quality teams are important anywhere.  But I'd still imagine that human voters would rank us differently.  This isn't PR 27 vs. 28, it's the difference between a 2- and 3- seed, inside the top 10.  In general I would expect people to look more heavily there at record against top teams.  Against the top 25 I'm 4-1; they're 1-5.  I'd have to imagine human rankings would value those wins on the road at 6, 11, and 14 and use them as an indication that my team can compete with the top-10 teams they're ranking us amongst.  There is no such evidence for Wooster.  Again, their top win is against #21, and they did that at home.

All things considered, I think the new system is much better than what we had before the PR.  Frankly, I'm glad to see other people think this makes sense.  But it does feel wrong to me.  Given 2 teams with somewhat similar overall resume strengths, I would give the higher seed to the one with the 4 best wins amongst the 2.
6/4/2013 1:49 PM
In the end, you're trying to marginalize the season into the games that you want to consider while basically throwing out the rest of the stuff, while trying to create context with your 2/3 seed over 6/7 seed argument.

You want your big games weighed/weighted more, and they shouldn't be, the schedule is 25 games to this point, and each game should have equal worth.  I know with real life/game SOS, they have Wooster's SOS 6th and yours 33rd, but that's based on raw win percentages and it's really a rather ****** system.  There's a pretty sizable difference in your schedules, even just giving it the eye test when I was entering the data...I entered Wooster's first, and then when I saw your schedule and the RPIs of all the teams you played...I thought to myself...no chance, and I guess I'm kinda surprised that you're as close as you are in my rankings.

You can't weigh particular games/instances twice and other games/instances only once, just because they hold more importance to you and your cause...it just isn't fair...that's why reviewing wins v. top 25...25-50...50-100, etc is pointless, because in a good system it's already been analyzed and appropriated solidly and correctly.  I'm sorry, but you don't deserve to be ahead of Wooster at this stage of the game.
6/4/2013 2:08 PM
You will never get a system that gets everything right 100% of the time. There will always be "special circumsatnces"
I would be amused if, just once, someone posted a complaint that they thought they were rated too highly ...
6/4/2013 2:28 PM
I don't believe I'm ranked too low, FWIW, just that Wooster is a little high....  I forgot to mention in my last post that our one common opponent - something that certainly gets looked at by human seeding committees inside the top 10 like this - was a road win for me and a home loss for them.

WRT your last post, colonels, it's hard to imagine a sentiment that sounds so logical being more wrong.  In the real world it is ABSOLUTELY the case that "big games" are weighted more heavily in seeding.  That's unambiguously true.  If you watch any bracketology show it's about 50% of what they want to talk about.  Nobody mentions the game in which Duke beat Crap State on the road.  They mention the 2-point victory at UNC.  Frankly, any games against teams above 100 or 150 really don't matter at all unless they're losses.  So I have a few less games that matter, and in this case that certainly hurts me.  But to pretend that games against the toughest opponents don't and shouldn't carry more weight in terms of seeding is just ridiculous.  For a low- or mid-major with 3 or fewer wins two or three non-conference games against top-25 opponents are generally the difference between a 2-seed and a 5-seed.  Honestly, if you look at their resumes, that's most of the difference last year between Memphis getting a 6-seed and Gonzaga getting a 1.
6/4/2013 2:32 PM
I'm actually surprised that Wooster is in front of you.
6/4/2013 2:40 PM (edited)
Posted by dahsdebater on 6/4/2013 2:28:00 PM (view original):
I don't believe I'm ranked too low, FWIW, just that Wooster is a little high....  I forgot to mention in my last post that our one common opponent - something that certainly gets looked at by human seeding committees inside the top 10 like this - was a road win for me and a home loss for them.

WRT your last post, colonels, it's hard to imagine a sentiment that sounds so logical being more wrong.  In the real world it is ABSOLUTELY the case that "big games" are weighted more heavily in seeding.  That's unambiguously true.  If you watch any bracketology show it's about 50% of what they want to talk about.  Nobody mentions the game in which Duke beat Crap State on the road.  They mention the 2-point victory at UNC.  Frankly, any games against teams above 100 or 150 really don't matter at all unless they're losses.  So I have a few less games that matter, and in this case that certainly hurts me.  But to pretend that games against the toughest opponents don't and shouldn't carry more weight in terms of seeding is just ridiculous.  For a low- or mid-major with 3 or fewer wins two or three non-conference games against top-25 opponents are generally the difference between a 2-seed and a 5-seed.  Honestly, if you look at their resumes, that's most of the difference last year between Memphis getting a 6-seed and Gonzaga getting a 1.
And I'm sure you'll be shocked to know that I think the RL selection process as is, is garbage/pointless and that all of the analyzing of wins v. top 50, top 100, top 150 etc, is needless because a proper ranking system would have already correctly taken that stuff into account, so there's no need to re-go over it and add weight to things that have already been given appropriate weighting.  In my eyes, the problem with the process is human subjectivity, but since that's what most of the masses can only get their heads around, that's how it's done.  Taking into account only the 1, 2, 4, or 8 games that you think/want to matter out of the entire 30-35 game schedule is a crime.

I would like to see my ranking system or another suitable ranking system choose a field of 68 every year in real life, but that would never fly...too many simpletons/idiots would cry foul...people like to ***** about that which they don't and/or can't understand.

6/4/2013 7:51 PM (edited)
Honestly, I used to think about it more the way that you do.  But I've come around.  In the real world, the purpose of seeding isn't to reward teams for their regular season performance.  It's to create a tournament that will A) be fun to watch and B) determine the best team in the country AT THE TIME OF THE TOURNAMENT.  That's really the best it can do.  The third priority would probably be fairness to the teams and players in the tournament, and maybe a distant fourth rewarding regular season performance.  I used to hate it when teams got bumped or docked a few seeds when a key player got injured or came back from injury late in the season.  Didn't seem fair.  But at the end of the day it's the most fair for the rest of the tournament.  Are you screwing over a team's players, who worked hard all season, by dropping them from a 3-seed to a 6-seed after their star gets hurt?  Sure you are.  But it protects the integrity of the bracket.  Imagine if Jordan had been injured for most of a season and came back to help Carolina win an ACC tournament.  Maybe your ranking system would call that team a 7-seed.  Is it fair to the 2-seed that's going to hit them in the 2nd round to actually make them a 7?  Not really.  In the same way you don't want to reward a team TOO much for having a schedule with a bunch of teams that aren't exactly cupcakes, but don't pose a real challenge or prove very much about the strength of a top-10 program.

At the top of the bracket in paritcular you seed based on the apparent relative strengths of the teams.  Obviously that's really the goal everywhere.  Games against teams in the 30s and 40s don't really speak to relative strengths of top 10 teams.  They just don't.  It's a pretty good bet that most of the time both of our teams would beat those teams.  Games against top teams demonstrate the actual talent level of the teams involved.

Here's another way to look at it.  Looking at our schedules, just try to guess how each team would have done against the other's schedule.  My team was 4-1 against the top 25.  I would tend to guess, based on my results alone, that I probably would have won about 2 more games than they did against their schedule.  OTOH, they were 1-6 against the top 25, so it's unlikely that with 3 top 10 opponent games in conference and a road game against a team they lost to at home they would have won any more than 22 games against my schedule.  Obviously that's conjecture, particularly since to the best of my ability I'm taking team talent out of the equation aside from it's obvious influence on results.  But unless you think that my guesses are wrong - and I'm guessing you don't, since they seem fairly reasonable to me and I think I'm looking at this fairly objectively - it seems hard to say that based on the balance of play they seem better based on the schedules, even though if we switched I would have done ~2 games better against their schedule than they did and they would have done ~2 games worse against mine.
6/4/2013 3:43 PM
Posted by m4284850 on 6/4/2013 2:32:00 PM (view original):
I'm actually surprised that Wooster is in front of you.
They're not anymore.
6/4/2013 4:23 PM
I agree that they shouldn't be double-counted, but I do think the best system rates quality wins very highly (and doesn't have arbitrary cutoff points. A win over #25 shouldn't be worth significantly more than a win over #26). Not that I know what the best system is, just that this should be a feature. Last year, I'd rather have wins over Duke (RPI #1), Austin Peay (323), and Presbyterian (332) than Stony Brook (77), LSU (92), and Loyola-Maryland (98). 
of 4

Terms of Use Customer Support Privacy Statement

Popular on WhatIfSports site: Baseball Simulation | College Basketball Game | College Football Game | Online Baseball Game | Hockey Simulation | NFL Picks | College Football Picks | Sports Games

© 1999-2014 WhatIfSports.com, Inc. All rights reserved. WhatIfSports is a trademark of WhatIfSports.com, Inc. SimLeague, SimMatchup and iSimNow are trademarks or registered trademarks of Electronic Arts, Inc. Used under license. The names of actual companies and products mentioned herein may be the trademarks of their respective owners.