Posted by jimmychino on 7/20/2020 11:57:00 PM (view original):
Interesting. I did not realize the SIM considered things like “quality wins”.

Yeah, for some reason the reference to Top 50/100/200 wins was removed from the FAQ a while ago, but this was left in:

Tournament seeding is not directly based on RPI or SOS. Often a team can have a high RPI without actually having many impressive wins.

Silly Benis!

there was a major change to seeding a number of years ago to include quality wins - perhaps when the projection report initially came out? people are referencing the wrong metrics though, like vs top 60 proj report, top 200 wins... those aren't things.

what IS a thing is record vs top 25/50/75/100 rpi. each of those 4 components is significant, and is not weighted by home/away. there is NO consideration for quality wins beyond that. a win vs a #1 rpi team on the road is equal to a win vs a #25 rpi team at home, as far as this metric goes. regular rpi and sos are of course still impacted by the record of your opponent (and the record of their opponents) and the home/away bit.

the scheduling meta has changed as a result. in the olden days, it was pure play 10 fairly easy to win games in non conf on the road against solid w/l teams, to pad your future losses.

now, a combination approach is in order. winnable games against fairly easy top 100 rpi teams are very valuable now. you also want to stretch for some top 50 rpi wins as your team gets a little better than it is now (and top 25 when you are a top team). it is perfectly acceptable to play these games at home - in fact, i would even recommend it. you know, a mix is good. if you can slaughter the 50-100 rpi teams, great - play them on the road. but there is no longer a clear downside to playing teams at home in non conf - provided they provide a 'quality win', whatever that means for your team (for a low end NT seeker, that means barely top 100 rpi - for title teams, that means barely top 25 rpi or better).

Posted by kcsundevil on 7/21/2020 12:35:00 AM (view original):

Posted by jimmychino on 7/20/2020 11:57:00 PM (view original):
Interesting. I did not realize the SIM considered things like “quality wins”.

Yeah, for some reason the reference to Top 50/100/200 wins was removed from the FAQ a while ago, but this was left in:

Tournament seeding is not directly based on RPI or SOS. Often a team can have a high RPI without actually having many impressive wins.

this seems to suggest rpi is not a factor anymore. i'm highly skeptical. record vs top X was added as a factor, but the old factors remained, as far as i know. it might be semantics, maybe seble is using something highly similar to rpi in its place, i don't know. but even that would be news to me. perhaps the use of the word 'directly' is providing the skew there.

the FAQ says

The logic for ranking teams for the postseason essentially calculates a score for each game on the team's schedule. This game score is determined by the following components:
( https://help.whatifsports.com/hc/en-us/articles/360014115752-How-are-the-at-large-bids-for-the-National-Tournament-determined-What-about-the-seeding )

i guess there's a game-by-game analysis on seble's end rather than full aggregates, but i'm not sure this particularly matters. in theory it could mean the home/away weighting of rpi is discarded, but i see no reason to believe that is in the case. the way seble has described the changes along the way is in pretty different terms than what the FAQ says now, but as best i can tell, its all a result of the same change. i think its probably a matter of technical fact versus practical reality - if you go game by game and use roughly everything RPI uses, a reasonable person might say, the seeding is largely based on RPI, even if technically, this is not correct. i suspect its the same story for this record vs top X stuff. what i will say is, the record vs top X aggregates if you will (6-3 vs top 100 rpi etc) sure seem to be a great predictor of seeding, it seems to be a very significant factor. i personally wouldn't worry about the explanation in the FAQ - even if its a simplification. assuming the vast majority of seeding runs off of (record vs top 25/50/75/100 rpi irrespective of home/away and RPI) sure seems to lead to conclusions very much in line with the projection report, and unless someone finds that isn't the case, i wouldn't get bogged down in the details of a game by game versus seasonal analysis.

now that a half hour or so has elapsed, i'm kinda feeling like this is all familiar. probably not so much news to me as i discarded this information as useless.

I've always thought the projection report always did a really good job of getting the deserving teams into the NT. I don't really see any outrageous snubs.
Outside of this crazy rule that apparently teams that win their CT get automatic bids...

agreed, the projection report changes are one of the only unambiguously great changes HD has seen in my time. definitely some of seble's finest work. i also think basically saved the game after potential came out (2.0), and i think everyone pretty much agreed he made things better, but there were definitely criticisms he didn't do enough and also when the release you are fixing is such an unmitigated disaster, it sort of makes clearing the bar of 'did this clearly make things better' a relatively easy feat. meanwhile the projection report took a decent system and made it way better, so i think that kind of stands out in my mind.

Posted by franklynne on 7/20/2020 4:06:00 PM (view original):
in d3 naismith, my 20-8 hiram 66rpi-100sos made the nt over 18-11 wabash 68rpi-39sos..wabash beat us in the ct..wabash was in the next 4 out group..??

I was just as shocked as you were franklynne...