Interesting The Diamond Sports Ranking: A Modern Alternative for College Baseball and Softball – 6–4–3 Charts
Initial thought is including margin of victory(in any sport) is dumb. Im all for anything that removes all human input, including selection committee. Surely theres a program that will spit out the 'perfect' 64 team field.
All ratings systems should have explicit criteria. The difference between a computer rating and a human rating would then be that the algorithms would be more quantitative and more consistent in applying these criteria. But algorithms can also be more constraining, sometimes leading outcomes that seem abjectly wrong to our human sensibilities. Including margin of victory is necessary if the goal of your rating system is find the best (ie most powerful) teams. As the article points out, winning 6-5 over a team isn’t the same as beating that team 11-0. The main downside to including MOV is encouraging teams to run up the score. The DSR seems to deal with this by minimizing the contribution of MOV to the overall rating. Alternatively, some people prefer ranking teams by resume over power. If that’s the case, you won’t want to include MOV. The downside here is of course that you are more likely to get lower seeded teams in the NCAA tourney that are favored over their higher seeded hosts. I think algorithms have a place in such projects as tournament seeding, but human judgement also seems to have value. No system can be perfect, but I kind of like the football playoff committee, as they use both algorithms and humans. The humans on the committee must use metrics to justify their arguments.
I dont disagree, I just dont trust any human to not have own agendas. Possibly even paid to have 'own' agenda. Computers also dont know/care the name on jersey. Biggest flaw of human input. I would maybe take an average of RPI, ELO(pretty sure includes media poll) DSR.
Hopefully one thats never seen a baseball game in their life. Which is my guess on who came up w/ current computer rankings. Probably also applys media polls
Damn jhf—that sounds like the same idiots that have ruined MLB with analytics, spin rates, launch angles, and all the other stuff people who have never played the game push to be center stage. Simply put whether you’re talking about pitching, hitting, or playing defense—you teach the fundamentals of consistently duplicating the mechanics of the position. YOU CANNOT DO THAT WITH ANY COMPUTER PROGRAM OR PRINT OUT. And that’s all I got to say about that…..
100% agree on the analtyics, etc. On field stuff. But when ranking 305 teams. the only human Ill trust over computers will never be asked. Especially since I dont answer phone. Or we can just go liberal route and let all 305 in,,then who cares about RPI lol
Cohen(AU AD) is selection chair and I still dont like it. The humans making the selections. And if gong to use a committee, really shouldnt be any debate.. Everyone rank your field 1-64(auto bids included) take the average of those rankings. And theres the field, no debate. Or at least T32 to allow for geography placements after that.
Here's a bunch of rankings: Massey Ratings - College Baseball Rankings I believe all the computer polls except "Massey BCS" are score-based. (EDIT: ELO and RPI aren't either.) Here's how tournament selection goes in D1 Men's Hockey, which would seem to be a starting point for any "objective" system. Info - NCAA Tournament Pairwise Primer That said, "Pairwise" ends up pretty much being a tweaked RPI. This year, Pairwise only differed from RPI in the 5 vs. 6 seeds, which were then promptly flipped back to shorten travel for both teams. (10/11/12 were also shifted to avoid intra-conference matchups, but 10/11/12 Pairwise = 10/11/12 RPI)
margin of victory still has some human elements that can sway these algorithms- for example, one coach/team may decide to sit his starters late in a sure loss or even a sure win. If the other coach does not…the margin of victory can go up or down greatly based on who is on the field/court. one great example of this is Auburn hoops- they were ranked super high by computer models like kenpom and Bart torvik and it was mainly due to margin of victory. They often continued to pour it on with starters and rotation players even after the opposing team dumped the bench onto the court. They seemed to purposely push the margin higher when the game was out of reach in the final minutes. Most teams don’t. (Maybe they know how the models work) Winning by 5, 10, and 20 all weight differently in their models. And yet, Auburn had like 1 quad 1 victory in the net rankings. The least of any top 25 team. Barely winning games vs winning big should matter though I’m not sure the best way to model it.
This all sounds like it comes down to “style points” which I detest because it makes the game outcome influenced by the media which I think is absurd. Coach and play your style of game without the influence of what’s going to be written afterwards.
Not to get off subject, and youre probably right on Margin of victory influencing their high computer rankings.. But rest of that totaly false. Before Lior Berman injury, AU played 11 guys every game 15-20 minutes each.. That only leaves 3 or 4 on roster..lol Who all came in any time up 15-20 last 5 minutes. Is it their fault their so called 2nd 5,really just a first 11, were better than most teams first 5?? Who else were they suppose to put out there? lol Ran into a team that didnt miss much, but who they should beat by 30 9/10 times. But in 1 & done thats all it takes is 1 game... Why i hate that format. So chokers maybe, Running up score on purpose?? hardly.. Even if so, thats also not their problem, other team is allowed to play D. Rumor is all the ones that can return,, Broome included, will all return, so you might get to see them 'run up scores' again next year..
Yep if including margin of victory, then your are basically rewarding/penalizing styles of play. Baseball wise, if they wanted to give a bonus for a run rule win, I guess Id be ok with, not really..lol
Good points. If we want to predict future outcomes, it would certainly help to know the intentions of coaches, as some like to put in young players to get experience and others prefer to keep their foot on the pedal. I know a couple of the football algorithms have started to discount data during garbage time, which is predefined by the algorithm as a certain minimum lead with a certain maximum time remaining. Those of course aren’t perfect either, but I believe they help. Baseball might be a bit different in this regard, as FSU found out this weekend, since there is no game clock. As a result, the number of runs that can be scored in a single inning is theoretically infinite. I agree that there is simply no perfect solution, so I think our best bet when setting a tourney field is to include multiple kinds of analyses and invite many teams. The former should minimize the amount of bias and the latter minimize the effects of the biases remaining.
