(08-22-2018 11:00 AM)stever20 Wrote:  will be really interesting to see how this impacts things. I do like the 10 point cap- think it could have been a bit higher but still good that there is a limit.

A capped margin of victory should be used for CFB as well. Along with using the point spread.

(08-22-2018 04:04 PM)Scoochpooch1 Wrote:  

(08-22-2018 11:00 AM)stever20 Wrote:  will be really interesting to see how this impacts things. I do like the 10 point cap- think it could have been a bit higher but still good that there is a limit.

A capped margin of victory should be used for CFB as well. Along with using the point spread.

Capped = bad. Sliding scale = good.

Example:

Winning by
1 = okay
5 = good
10 = better
15 = much better
20 = not much better than 15
25 = only a little bit better than 20
30 = almost no difference than 25
and so on

Some of the better computer models already do this(Sagarin, maybe Kenpom). A flat cutoff leads to bad results. Especially set as low as 10.

I should add to my post. New Mexico State lets say play West Texas A&M and Chicago State. If I am a school like New Mexico State, I would like to have West Texas A&M count in SOS to help their NET than playing Chicago State. Schools not in a P5 conference do not have any choices on who they play.

(08-22-2018 03:10 PM)DavidSt Wrote:  I do wonder. The playoff committee said you get more bump in your SoS if you play a school like NDSU than a Kansas in football.

I'm 90% certain that no one but you ever said that.

(08-22-2018 06:28 PM)johnbragg Wrote:  

(08-22-2018 03:10 PM)DavidSt Wrote:  I do wonder. The playoff committee said you get more bump in your SoS if you play a school like NDSU than a Kansas in football.

I'm 90% certain that no one but you ever said that.

Same here.

But I'm pretty sure that a win over a team that was 2-0 vs FBS over the past 4 years is better than a win over a team that was 3-40.

RPI will get its revenge and return when the NCAA using NET for men but not women causes a Title IX lawsuit.

(08-22-2018 04:51 PM)Hammersmith Wrote:  

(08-22-2018 04:04 PM)Scoochpooch1 Wrote:  

(08-22-2018 11:00 AM)stever20 Wrote:  will be really interesting to see how this impacts things. I do like the 10 point cap- think it could have been a bit higher but still good that there is a limit.

A capped margin of victory should be used for CFB as well. Along with using the point spread.

Capped = bad. Sliding scale = good.

Example:

Winning by
1 = okay
5 = good
10 = better
15 = much better
20 = not much better than 15
25 = only a little bit better than 20
30 = almost no difference than 25
and so on

Some of the better computer models already do this(Sagarin, maybe Kenpom). A flat cutoff leads to bad results. Especially set as low as 10.

If margin of victory is going to be a factor why not use a point system awarding points for ranges of victory margin but also for the score at the half. If Blue Blood U leads by 20 at the half and wins by 15 they had a dominant day but if MidMajor U trailed at half and was in a tight game until the final three minutes but managed to win by 15 after their opponent jacked up bad threes and fouled a lot, that is a different critter.

Likewise if you win by 5 in OT that's different from winning by 5 in regulation.

Did they factor in road games in this?


Sent from my iPad using Tapatalk

(08-22-2018 11:54 PM)arkstfan Wrote:  Likewise if you win by 5 in OT that's different from winning by 5 in regulation.

Heck, overtime in general should count for something.

(08-22-2018 11:45 AM)leofrog Wrote:  

(08-22-2018 11:26 AM)quo vadis Wrote:  

(08-22-2018 10:50 AM)jdgaucho Wrote:  Of key importance, game date and order were omitted to give equal importance to both early and late-season games.

I'm not sure I like that. The tournament isn't played at a random time, it is played at the end of the season, so it stands to reason that performance at the end of the season is more indicative of likely success in the tournament.

Also, in most sports with playoffs, there is the concept of "getting better". Yes, teams practice in October to get ready, but nobody can necessarily be fully ready for game play until games actually start being played. I like the idea of rewarding teams that make adjustments, learn from their early mistakes, and get better.

Sure, sometimes teams have better results the last part of the season because of luck - maybe their SOS was front loaded, maybe they luck out and tough teams have injuries when they play them late. But still.

How can you make an objective, unbiased computer ranking factoring in game dates? And, are the dates same for everyone?

Well, by having a line of code that adds weight to games played after a particular date, or set of dates?

(08-23-2018 07:44 AM)The Cutter of Bish Wrote:  

(08-22-2018 11:54 PM)arkstfan Wrote:  Likewise if you win by 5 in OT that's different from winning by 5 in regulation.

Heck, overtime in general should count for something.

Like the hockey overtime Loss gets you a point

(08-23-2018 11:37 AM)dbackjon Wrote:  

(08-23-2018 07:44 AM)The Cutter of Bish Wrote:  

(08-22-2018 11:54 PM)arkstfan Wrote:  Likewise if you win by 5 in OT that's different from winning by 5 in regulation.

Heck, overtime in general should count for something.

Like the hockey overtime Loss gets you a point

Exactly!

I want to see ranges because a 2 point win is not twice as good as a one point win or worth 50% less than a three point win.

Any game you could have lost with one shot should be valued equally and any game you could have lost with two shots should be valued equally rather than a 5 point win being worth more than a 4 point win.

I think one thing we're going to see is close wins vs bad teams really hurting teams going forward. A team like Providence last year who had 4 wins OOC vs mediocre to bad teams by less than 6 points- in RPI they don't get punished but in the NET rating- they will totally get killed by that.

(08-22-2018 11:45 AM)leofrog Wrote:  How can you make an objective, unbiased computer ranking factoring in game dates?

You can't.

There is no such thing as an objective, unbiased computer ranking. Doesn't matter whether it's RPI or NET or KenPom or any of the 100 different rankings that try to rank college football teams.

The formula used by a computer to rank teams reflects the biases of whomever created the formula. It's no more "unbiased" than a poll voter or committee member, and it's not inherently better than a poll or a committee vote. The one thing a computer ranking does do "better" is that the formula used by the computer applies its biases consistently to every situation.

I don't like factoring in margin of victory. It will change the way the game is played.

The point of the game is to "win." Not to "win by a lot." Not to "have a fast-paced offense." Not to "score a lot." It's winning or losing.

I can understand adjustments for the circumstances under which the game was played (game location, injuries to key players, and even who picked the referees). But the official metric used by the NCAA should not do anything that might change the goal of the game or the way the game is played.

(08-23-2018 11:37 AM)dbackjon Wrote:  

(08-23-2018 07:44 AM)The Cutter of Bish Wrote:  

(08-22-2018 11:54 PM)arkstfan Wrote:  Likewise if you win by 5 in OT that's different from winning by 5 in regulation.

Heck, overtime in general should count for something.

Like the hockey overtime Loss gets you a point

Right...you couldn't determine the game within standard regulation, so while the winner still gets something, the loser does, too.

I also don’t care for this sort of moral highground the system thinks it can impose upon the game to discourage blowouts. Some games are bound to get ugly, and that’s the beauty of the game...because when that didn’t matter with the RPI, it still happened. This encourages “decisiveness” but discourages a byproduct of a rewarded act. And I have a tough time understanding how winning a game in one or two overtimes by over ten points because the other team ran out of gas should count the same as a win against a team that never bothered to show up at the first whistle. And, again, how it puts mid/non majors in a bad spot because it will still come down to SOS and who you played rather than HOW you played.

It's funny- i don't think folks here appreciate just how big this is going to change college basketball.

(08-30-2018 09:23 AM)stever20 Wrote:  It's funny- i don't think folks here appreciate just how big this is going to change college basketball.

No one knows until the NCAA either releases the formula or the math nerds can reverse engineer it.

(08-30-2018 09:50 AM)arkstfan Wrote:  

(08-30-2018 09:23 AM)stever20 Wrote:  It's funny- i don't think folks here appreciate just how big this is going to change college basketball.

No one knows until the NCAA either releases the formula or the math nerds can reverse engineer it.

The fact that it's going to incorporate more than just who you play, where you play, and binary win/loss- is a huge sea change.

(08-22-2018 09:55 PM)IWokeUpLikeThis Wrote:  RPI will get its revenge and return when the NCAA using NET for men but not women causes a Title IX lawsuit.

There might be a case, except that RPI will still be used by all other men's and women's sports.