2019-20 National Rankings

[Cdizzle]

But you're arguing that

A) Losing to the #3 team in the country at home by 12 points

is better than

B) Beating the #2 team in the country on the road by 12 points

[Kel Varnsen]

How about a 1 point win over Duke and a 1 point loss to TAMCC? Asking for a friend.

https://kenpom.com/team.php?team=Stephen+F.+Austin

The beauty of these ratings systems is that they've created a way to cut through the noise and ambiguity in instances like these, where a team has one great performance and one awful one.

On another note, how do you win at Duke, then manage to lose to #291 TA&MCC? At HOME?

[Stickboy46]

But wouldn't A) losing to the #10 team in the country on the road by two points be a better indication of a team being good than B) beating the #200 team in the country at home by 10 points? If I'm trying to choose the better team between the team that did A and the team that did B (assuming those are the only data points we have), I'd go with team A every day of the week.

To make this even close to relevant you need to add a third piece of info. Team B beat Team A by double digits on Team As floor.

You still going with team A?

[wichshock65]

For those hating on my take on metric use, I would point out that I also called Palm an idiot and terrible at his job for having WSU a 2 seed a week ago, when the NCAA Selection Committee's own metric supported his position. It's OK to have nuance. Metrics can be really useful for comparing unknowns over significant gaps, and for evaluating HOW a team is doing things.

Since 2016 keeps being brought up, that's a good example where we're not talking about flip-flopping #2 and #3 in a poll. We're talking about a team rated as roughly the 10th best being treated as roughly the 45th best. That kind of disparity is also difficult to reconcile.

I find it interesting that some folks are against a FAN of a team wanting to have it both ways, but are for supposedly impartial evaluators having it both ways. Fans are irrational. It's in the name. When 'impartial' people are also irrational, don't hate the fan for calling them on it.

We are all fans of the team and i have enjoyed the banter on this thread. What started all of this was the KU ass hat who ranked his hometown team higher than the team that just whooped their ass on their home turf. If calling him out makes me irrational, so be it.

WINNING MATTERS!

[Kel Varnsen]

But you're arguing that

A) Losing to the #3 team in the country at home by 12 points

is better than

B) Beating the #2 team in the country on the road by 12 points

If that's the only data point we have, no. But it's not the only data point. Each team has played 14-15 games up to this point.

[Cdizzle]

The beauty of these ratings systems is that they've created a way to cut through the noise and ambiguity in instances like these, where a team has one great performance and one awful one.

On another note, how do you win at Duke, then manage to lose to #291 TA&MCC? At HOME?

I don't have any idea. But I'm glad the poster posed the question, glad I have kenPom, and glad I immediately thought of SFA. Cuz that shite is hilarious.

[Kel Varnsen]

To make this even close to relevant you need to add a third piece of info. Team B beat Team A by double digits on Team As floor.

You still going with team A?

Great, each team has played 13-14 games prior to that. We have that data. Why do all of the prior games not matter?

[Cdizzle]

If that's the only data point we have, no. But it's not the only data point. Each team has played 14-15 games up to this point.

That's a little convoluted, given that the metrics gave us those ratings to begin with.

Which might mean what that game really did is create a glitch in the matrix, I mean metrics.

[WstateU]

…

[Kel Varnsen]

That's a little convoluted, given that the metrics gave us those ratings to begin with.

They did, but they also used data from those other games (and as a result, 13-14x as many possessions) in order to come up with those ratings.

Anyways, no one's gonna give here, so I'm done on the subject. Work has been slow today.

[Cdizzle]

They did, but they also used data from those other games (and as a result, 13-14x as many possessions) in order to come up with those ratings.

Anyways, no one's gonna give here, so I'm done on the subject. Work has been slow today.

So the model was invalidated. Time to tweak it.

[Kel Varnsen]

So the model was invalidated. Time to tweak it.

There are somewhere around 6000 college basketball games played per year. If we have to change the model based on getting one prediction wrong, then we can't have anything.

Okay, now I'm done.

[AZ Shocker]

With all these teams getting ranked so high with multiple quote/unquote "bad losses"...perhaps the voters should rank the #1 team in the country to a team that has played the most games against the Top-25. I'd like to have seen who that would've of been before conference schedules started. Lol

[Stickboy46]

Great, each team has played 13-14 games prior to that. We have that data. Why do all of the prior games not matter?

You are a lost cause

[Cdizzle]

There are somewhere around 6000 college basketball games played per year. If we have to change the model based on getting one prediction wrong, then we can't have anything.

Okay, now I'm done.

So if we don't tweak the model, what you're saying is that perhaps instead we should use our own skills to apply logical adjustments to the models rating system based on additional information when creating our own ranking?

Wonder who could do that and what it might look like.