Eh. Yes, cold has an impact on muscle injuries but heat and humidity is a different beast especially if the playing surface is turf. August is a brutal month in the south to begin with, July being just as bad. Starting preseason in July, in the South, on turf fields is a quick recipe for heat poisoning with too many coaches and athletic training staff telling players just to push through. In saying that though, my kid left the heat and humidity claiming she was never coming back to it. She's now experienced playing on turf that has iced over, playing in temps with windchills in the single digits and has learned to play with multiple layers on. If you asked her, she'd take being soaking wet in sweat, sunburns and drinking the daily gallon of water over iced turf fields any day.
Coaches should be disciplined and trainers fired, if that’s the case. It shouldn’t be in a coaches hands to begin with, so fire the ADs if they permit it.
Totally agree with you but the same can be said with the cold and ice. Is there a reason in the offseason to be on a turf field that you had to shovel before practice and stop out the ice to break it up, in 22 degree weather with a 15-20 mile an hour constant wind? I can't fathom, especially with ham, quad and knee injuries, how that is viewed as safer. Side note: Most southern teams respect the heat, its when non-Southerners show up with teams not used to it that some damage can be done.
I just have sent to the Division 1 Women’s Soccer Committee a proposal for RPI formula changes that would bring teams’ RPI ranks and their ranks as strength of schedule contributors into relative equality and that would vastly improve the RPI’s ability to rank teams from conferences properly in relation to teams from other conferences. I sent them a Summary and a much longer Memo with details and data supporting what I am proposing. Getting the NCAA to make RPI changes will take, at a minimum, a very strong Committee effort. If you are a D1 coach, you may want to use the links and download what I have sent the Committee. If you support the changes, it will help if you -- and your conference -- weigh in with your Committee member, letting him or her know of your support. (Or, perish the thought, if you do not like the changes, you can let him or her know that.) Here are the Committee members and their affiliations: Clifton Douglass, Chair Southwestern Athletic Conference (administrator) Debra Boughton Northern Illinois (administrator) Duane Bailey Louisiana Lafayette (administrator) Jen Klein Michigan (coach) Keri Mendoza Cal Poly (administrator) Lisa Varytimidis Austin Peay (administrator) Matt Mott Ole Miss (coach) Ron Rainey Dartmouth (coach) Scott Leykam Portland (administrator) These changes would greatly improve the RPI and make non-conference scheduling much more straightforward. I hope they will get a lot of coach support.
CP- Thank you for your work. I’m a long time lurker and d1 Women’s coach. I agree there needs to be changes to the RPI system. Do you have the RPI rankings for this past season with these recommendations? I would love to see. If I missed it somewhere, my apologies. What impact do you see the potential or probable elimination of overtime playing in the RPI (more ties)? What are your thoughts on implementing a Performance Indicator like Division 2 or something similar to the NET system used by D-1 basketball? thanks for all your work and insight as always for our game.
WhiteBeard, here is a table that compares the actual 2022 final RPI ranks (third column) to what the ranks would have been using the system I have proposed (fourth column). The table covers the Top 57 using the final RPI ranks, since no team ranked poorer than #57 (under the current RPI formula) has gotten an NCAA Tournament at large selection over the last 15 years. I have marked in yellow the teams that were in the RPI Top 57, but that drop out under the proposed system. They all either were Automatic Qualifiers or did not get at large Tournament positions. Then, I also have included teams not in the RPI Top 57, but that would have been in the Top 57 under the proposed system. For practical purposes, one could consider these teams that probably did not get consideration for at large positions but that probably would have gotten consideration if the NCAA used the proposed system. Except for the Houston-UCF switch, the teams that drop out of the Top 57 are from second-tier conferences in terms of conference strength and the teams that move into the Top 57 are from the top tier conferences (which is as expected). Since the teams that drop out of the Top 57 are either Automatic Qualifiers or did not get at large selections, the changed rankings for those teams would not have made any difference in the 2022 NCAA Tournament bracket. That leaves the question whether any of the teams moving into the candidate pool would have been likely to bump out other at large teams. I have not taken a careful look at that question, but my educated guess is that the at large teams would not have changed. If that is right, then the only difference the change to the proposed RPI system would have made is that coaches, in setting their pre-season schedules, would not have had to spend time trying to game the system. They simply would have been able to schedule teams based on their likely strength, knowing that however strong the selected opponents turned out to be, their RPI strength of schedule contributions would come very close to matching that strength. In my opinion, this would have been a big plus for D1 and the coaches. (Note: I have some additional changes to the RPI system to address its problem rating teams from different geographic regions in relation to teams from other regions and that do even better at rating teams from different conferences in relation to teams from other conferences. These are changes, however, that I am sure the NCAA never would consider. I will discuss these changes in a future post.)
WhiteBeard, in response to your other questions: I do not know what the effect would be of eliminating overtime games, beyond our ending up with more ties. Years ago, I did an experiment with an Elo rating system where I treated all games that had gone into overtime as ties. To my surprise, it ended up that there was little difference from treating the golden goal games as wins and losses. Based on that, I have the feeling getting rid of overtime games would not make much difference from a rating perspective, but I am not sure about that as it seems counter-intuitive. I am not familiar with the Division II performance indicator -- although I am a little familiar with the system the NCAA uses for FCS football, which is somewhat similar to part of what I have recommended. I do not know much about how the NET system works, but one of the reasons the NCAA went to it is that the basketball coaches (and some of the conferences) had figured out how to game the RPI through non-conference scheduling and they wanted a system where coach ability to do that was minimized. My assumption is that the NCAA intends to at some point move sports using the RPI to the NET system, although I have no idea when. If my assumption is right, I expect that the NCAA will resist changing the RPI now, saying they would rather wait for the NET system. If they try to do that, I think the Committee should either demand a timeline for when that will happen or push for the changes I have recommended now.
Love this! CP Thomas, it seems from the data above this would reward more of the Mid Major type teams… Can you explain the additional or different values this new metric rewards and penalizes?
The theory behind the RPI is that when evaluating teams you should look at two things: the team winning record and the team strength of schedule. This is a good theory. In addition, the theory behind it is that winning record should account for about 50% of a team rating and its strength of schedule should account for the other 50%. This likewise is a good theory. When computing what a team contributes to the strength of schedule of an opponent, the RPI goes away from the 50%-50% model. Instead, when it evaluates you as a strength of schedule contributor, it has your winning record accounting for 80% of what you contribute and your strength of schedule accounting for only 20%. This is why we get such big differences between a team RPI rank and its rank as a strength of schedule contributor. So, suppose you are in a very strong conference and are a middle-of-the-conference team. Because your conference is so strong, you are likely to end up with a roughly 0.500 conference record. Alternatively, suppose you are in a mid-tier conference and are a near-top-of-the-conference team. You are likely to end up with a very good conference record, say around 0.800. And, suppose that these two teams are roughly equal in actual strength. When the NCAA is computing what the strong-conference team contributes to the other teams in its conference, in terms of strength of schedule, the 0.500 is 80% of what it contributes from its conference record. On the other hand, 80% of what the mid-tier conference team contributes to the other teams in its conference is 0.800. And this difference is there even though the two teams are roughly equal in actual strength. The bottom line of this is that the current 80-20 strength of schedule contribution format biases the RPI in favor of teams from mid-tier conferences and against teams from top-tier conferences, even if the two teams are actually equal. So, to remedy this, the system I have proposed rebalances the strength of schedule components so that the winning percentage of your opponent counts for only 50% of its contribution to you and who it achieved its winning percentage against accounts for the other 50%. (Interestingly, looking at how ice hockey does it, it appears that this is what they do with their version of the RPI formula.) So, in fact, as my earlier post using the 2021 season as a basis for comparison shows, the current RPI formula is biased in favor of the mid-tier conferences and against the top-tier conferences. Part of the reason for the proposed changes is to remove this bias. Thus five of the six teams that the current version of the RPI had in the Top 57 in 2021, but that would drop out of that group using the recommended system, are from mid-tier conferences. And all of the teams that the recommended system would move into the Top 57 are from top-tier conferences. This is not by design. It simply is a result of moving to an ungamable and far less discriminatory rating system. Also as I pointed out above, however, this may not really hurt teams from the mid-tier conferences. Those that lose ranking positions but are Automatic Qualifiers will get into the NCAA Tournament anyway. And those that drop out of the Top 57 but are not Automatic Qualifiers may not be getting at large selections even if in the Top 57. Where the real difference may be is that teams from top-tier conferences that in the past have not gotten consideration for at large spots now may get consideration.
On looking over systems I have tested, I see that I did run a test of how the current version of the RPI would work if overtime games were eliminated except in conference tournaments. Overall, the correlation between RPI ratings, adjusted for home field advantage, and game results would be slightly poorer. On the other hand, in terms of the ability to rank teams from different regions and conferences properly, in relation to teams from other regions and conferences, would be about the same -- poorer in some respects and better in others. Overall and without spending a lot of time on it and not updating to include recent years, it looks to me like the impact on the RPI as a rating system would not be large. That suggests how such a change would affect the RPI probably should not be a major consideration in the debate over whether to eliminate overtime games during the regular season.