# Coach's polls vs. RPI...accuracy? **URL:** \ **Category:** The Game Room\ **Created:** [February 20, 2014, 2:02pm UTC](https://boards.straightdope.com/t/coachs-polls-vs-rpi-accuracy/681787 "2014-02-20T14:02:34Z")\ **Posts on this page:** 5\ **Page:** 1
**Author:** ![crucible](https://avatars.discourse-cdn.com/v4/letter/c/5f8ce5/32.png) [@crucible](https://boards.straightdope.com/u/crucible)\ **Post date:** [February 20, 2014, 2:02pm UTC](https://boards.straightdope.com/t/coachs-polls-vs-rpi-accuracy/681787/1 "2014-02-20T14:02:34Z")
There are generally some significant differences in NCAAB rankings between polls completed by humans and polls calculated by computers. I’m trying to find a study of some sort that tries to show that one way or the other is more accurate when the final tournament results are in Has anyone tried to set a point value for each team’s final level? For example, if you are in the NCAA tourney, and are eliminated in the first round, you get 1 point, etc. up to the championship game, 7 points. But, if the ranking you previously had, pre-tourney, said you should win that game, and you lost…only half the points for that level. Holes in that, I know. does someone have a better way to evaluate the accuracy of the RPI vs. coach’s polls? ---
**Author:** ![crucible](https://avatars.discourse-cdn.com/v4/letter/c/5f8ce5/32.png) [@crucible](https://boards.straightdope.com/u/crucible)\ **Post date:** [February 20, 2014, 2:04pm UTC](https://boards.straightdope.com/t/coachs-polls-vs-rpi-accuracy/681787/2 "2014-02-20T14:04:51Z")
I know one hole is that when the coach’s and writers do their ranking, they probably look at the RPI for some amount of guidance. So, further, the study should have a way of comparing how accurate either method is, versus the actual final result. ---
**Author:** ![crucible](https://avatars.discourse-cdn.com/v4/letter/c/5f8ce5/32.png) [@crucible](https://boards.straightdope.com/u/crucible)\ **Post date:** [February 21, 2014, 3:29pm UTC](https://boards.straightdope.com/t/coachs-polls-vs-rpi-accuracy/681787/3 "2014-02-21T15:29:25Z")
okay, nobody cares. How about we make it easier… Using the three major polls, for NCAAB, how many times did a higher rated team defeat a lower rated one in the 2013 postseason NCAA tournament? do I have to do all the work here? Where are the interns? ---
**Author:** ![astorian](https://avatars.discourse-cdn.com/v4/letter/a/ad7895/32.png) [@astorian](https://boards.straightdope.com/u/astorian)\ **Post date:** [February 21, 2014, 5:21pm UTC](https://boards.straightdope.com/t/coachs-polls-vs-rpi-accuracy/681787/4 "2014-02-21T17:21:25Z")
Ask some coaches, off the record, and most will probably admit candidly that they never fill out their own ballots. They routinely hand off their ballots to an assistant coach, a secretary, or a PR man and say, “Here, do this, I don’t have time.” And it doesn’t really matter anyway. Coaches, secretaries, and PR men all follow roughly the same procedure: Look at who was in the top 20 last week, drop the guys who lost a few notches, and move everyone else up a few notches. Not exactly scientific. Not that I take the computers much more seriously. But at least the stat geeks are TRYING. ---
**Author:** ![crucible](https://avatars.discourse-cdn.com/v4/letter/c/5f8ce5/32.png) [@crucible](https://boards.straightdope.com/u/crucible)\ **Post date:** [March 2, 2014, 1:36pm UTC](https://boards.straightdope.com/t/coachs-polls-vs-rpi-accuracy/681787/5 "2014-03-02T13:36:57Z")
Yes, I don’t have a clue why coaches would have an informed opinion about teams they haven’t been exposed to. You’re right. Sportswriters? That seems reasonable, but overall, I suspect the computer predicts the post season better than any human poll.