Pairwise and Bracketology 2013 Edition

[Numbers]

Re: Pairwise and Bracketology 2013 Edition

I agree with Almington about this, in several ways:
1) Variety of conferences is better
2) AH teams are not losing 7-1 in the tourney, and actually have had some success
3) More out-of-conference games next year due to the arrival of the Big 10 will make a difference in the ratings. A welcome difference. It will make the ratings better because of more intermingling of schedules, and therefore less influence by a small number of games.

[Patman]

Re: Pairwise and Bracketology 2013 Edition

If the AHA gets an at large, I'm sure the NCAA will take a hard look at preventing that from happening again. They might even resort to KRACH.

Black helicopters fly by night...

I actually expect an RPI adjustment for the open schedule next season... unsure how they justify these things, but i expect it to happen.

[Numbers]

Re: Pairwise and Bracketology 2013 Edition

Black helicopters fly by night...

I actually expect an RPI adjustment for the open schedule next season... unsure how they justify these things, but i expect it to happen.

What do you mean by 'adjustment'? Back to a good wins category - with varying degrees of effect for home-neutral-away?

Or, a change in formula?

Also, all those 'adjustments' just show that the RPI rating is an arbitrary, 'best guess' tool.

[Numbers]

Re: Pairwise and Bracketology 2013 Edition

Priceless,

Do you have one more update planned on "in the last 10 years, x % of teams above #16 in the PWR as of March 1 made the field??"

That has been interesting.

[FlagDUDE08]

Re: Pairwise and Bracketology 2013 Edition

Black helicopters fly by night...

I actually expect an RPI adjustment for the open schedule next season... unsure how they justify these things, but i expect it to happen.

We're not going independent.

[Patman]

What do you mean by 'adjustment'? Back to a good wins category - with varying degrees of effect for home-neutral-away?

Or, a change in formula?

Also, all those 'adjustments' just show that the RPI rating is an arbitrary, 'best guess' tool.

The percentage formula. Fact is, it varies across sports... Part of me wonders if they have somebody calibrate it for them.

I don't imagine some new component any time soon... HOWEVER... 2-3 years into our new system I would expect a more general change as the powers that be see the effect of the open schedules has on the results and the scheming that takes place.

[Priceless]

Priceless,

Do you have one more update planned on "in the last 10 years, x % of teams above #16 in the PWR as of March 1 made the field??"

That has been interesting.

I have one for the end of February and one for "the weekend before the playoffs" which is next weekend. I'll be posting the February recap on Monday (maybe Sun PM) and the playoff data the Monday after.

[Slap Shot]

Re: Pairwise and Bracketology 2013 Edition

Scooby's head must be sploding.

[LordofBrewtown]

Re: Pairwise and Bracketology 2013 Edition

The L16 or Last X games was a terrible criteria - you could be completely screwed depending on if your better opponents were front-loaded or back-loaded on the schedule. I'd like to see the TUC go away too, since the cliff presents such a problem. I think the best thing that they could do to change the pairwise is to adjust all wins/losses in the RPI based on home away (road win counts as 1.3, home win 0.7, or whatever the comined rate for all teams in current year is). If it has to be more simple than that, just use total (gross) road wins as a comparison (not road win %) and have that replace the TUC component. Might hurt the 'bigger' schools like my Badgers since they'll have less road games & less opportunities to win those, but having 6-10 more home games than some teams is a huge advantage for RPI & needs to be accounted for somehow

[Steve_MN]

Re: Pairwise and Bracketology 2013 Edition

Which, by the way, brings up this for discussion:

Disclaimer first: The current PWR follows the current RPI rankings pretty closely. As for the tourney field, only BC (+3), and Dartmouth (+3) are more than 2 spots off of their RPI rank. (I am ignoring that Providence wins the 4-way tie in the HE). Only BC and Lowell would be in the wrong bands (compared to if they simply used RPI), and only RPI (the school) would be out of the field, replaced by Omaha.

Now, that being said, what would everyone think of this:
To calculate each comparison:
RPI = 2 pts + tiebreak
TUC = 1 pt
ComOpp = 1pt
H2H = 1 pt unless the teams played more than 2 games and there was a sweep (then = 2pts)

My reasoning: First, if the RPI is doing what you want it to do, you should only override it in extreme circumstances. Hence, 2 points.
Second, in a long season, when you play someone is often more important than who is better, hence the change in the calculation for H2H.

Comments?

Doing this on the present calculation (from the calculator here http://www.elynah.com/tbrw/2013/cgi-bin/rankings.cgi) without being able to implement the h2h sweep, leaves all the rankings going by RPI.

Is their something better yet?

I think it unlikely the committee ever goes to KRACH, so I don't count that. How about, "Whenever RPI with within 1%, then look at other criteria?"

I would say no only because the current formula for RPI is flawed. Fix that and we can look at tweaking the pairwise a little.

Edit: I'm on my Kindle so I won't be expanding that answer for a while but it is cool that my Kindle offered up "pairwise" as an option. Guess it knows me well.

I have to completely agree on this. RPI is a useful tool, but it already counts more since it's used as the tiebreaker, and there's no way that RPI should count more when comparing two teams against each other than their actual head-to-head, on-ice performance. As the process stands right now, each head-to-head game counts for one point, the same as the other comparison criteria, and I think this is the proper weighting.

The main issue that I think needs to be addressed is the TUCliff. I'm not sure what the proper method of dealing with this is, and I know that none of these calculations are officially done until the last game of the conference tournaments is done, but the fact is, everyone knows the formula, it's fairly easy to program a computer to calculate the values, so people CAN look ahead to predict the outcome of games, so that you get to a situation that HAS come up before where it's beneficial for NCAA tournament purposes to LOSE a game at the end of the year, in order to ensure that your opponent moves above the TUCliff line, thus increasing your TUC number.

The easiest way would be to just remove the TUC component, but I'm not necessarily advocating that. That may be the best way, but I'm not necessarily opposed to measuring performance against the "top" teams, I just know that the current Cliff problems really should be dealt with.

[kingdobbs]

Re: Pairwise and Bracketology 2013 Edition

I have to completely agree on this. RPI is a useful tool, but it already counts more since it's used as the tiebreaker, and there's no way that RPI should count more when comparing two teams against each other than their actual head-to-head, on-ice performance. As the process stands right now, each head-to-head game counts for one point, the same as the other comparison criteria, and I think this is the proper weighting.

The main issue that I think needs to be addressed is the TUCliff. I'm not sure what the proper method of dealing with this is, and I know that none of these calculations are officially done until the last game of the conference tournaments is done, but the fact is, everyone knows the formula, it's fairly easy to program a computer to calculate the values, so people CAN look ahead to predict the outcome of games, so that you get to a situation that HAS come up before where it's beneficial for NCAA tournament purposes to LOSE a game at the end of the year, in order to ensure that your opponent moves above the TUCliff line, thus increasing your TUC number.

The easiest way would be to just remove the TUC component, but I'm not necessarily advocating that. That may be the best way, but I'm not necessarily opposed to measuring performance against the "top" teams, I just know that the current Cliff problems really should be dealt with.

The TUC cliff is a manufactured issue, as far as I'm concerned. I do not realistically see a coach telling his team to tank a game if it means it would improve their tournament standing. Badmintoneers got disqualified in this past Olympiad for doing that exact thing.

That being said, I am not sure, in a theoretical or practical sense, that there can be an elimination of the problem short of eliminating the criterion, only a mitigation of its effect. The best way, as far as I'm concerned, is to either add more criteria, thus diminishing the relative weight of the criterion compared to the others (that is to say, decreasing the likelihood that the TUC criterion is the one that changes a 2-1 comparison win for one team into a 1-2 comparison loss); another possibility is changing the points a given criterion is worth (this is already accomplished in the H2H criterion, as teams are simply awarded one point per win, instead of "one point for the best record"; in theory, someone could win a comparison with another team even though they lost in the RPI, TUC and COps criteria by beating that team four times in head to head competition).

[Numbers]

Re: Pairwise and Bracketology 2013 Edition

Concerning the TUC Cliff, this would be complicated math, but I think we all understand the problem is that 2-0 v Holy Cross does not have the same meaning as 2-0 v Quinn, Minn, or Miami.

One way to fix that would be: For every win against a TUC, you get their RPI on your "TUC score" (ties count half). After adding up your TUC score, divide by the # of games played. I actually think this still undervalues results against the very strongest teams, but it mitigates the effect a little. Granted, using the other teams' RPI has no theoretical backing. It's just the easiest mitigating factor i could think of.

[komey1]

Re: Pairwise and Bracketology 2013 Edition

I think they definately needs to be some sort of criteria based on how well you do against better teams. When you look at the current TUC list, it's about half the teams. I think that's too many. Drop the number to top 20 or so in RPI and I think you might get a better feel for how teams do against quality competition.

[Numbers]

Re: Pairwise and Bracketology 2013 Edition

I think they definately needs to be some sort of criteria based on how well you do against better teams. When you look at the current TUC list, it's about half the teams. I think that's too many. Drop the number to top 20 or so in RPI and I think you might get a better feel for how teams do against quality competition.

Komey,

I just tried it out on Whelan's site. I chose .515 in the RPI as the cutoff, because Wisco at 20 has RPi of .5152. Only difference in the field is that Omaha replaces Dartmouth. I believe because CC drops out of TUC status. Everyone else even ends up in the same band.

[MavHockey14]

Re: Pairwise and Bracketology 2013 Edition

Minnesota State from 7th to 15th. Do they come right back up with a win tonight? (Maybe not to 7 but 10 or so?)