NCAA Selection Discussion - Containment Thread

[RURICK]

If anyone thinks they’re not a lock they should have to present the scenario where Hopkins doesn’t get in.

Hop loses to Maryland, loses to Rutgers in the first round of the BT tournament (I don't know if this match-up is possible), Michigan or Ohio State wins the BT tournament, Cornell wins out except for losing the ILT championship to Yale, Georgetown wins out but loses the BE championship to Denver. Eight at-larges are Duke, UVa, ND, UMd, RU, PSU, Cornell, GT (not in order).

Hopkins resume is better than Rutgers, Cornell, and Georgetown. Rutgers and Georgetown resumes aren’t even in the same universe to Hopkins. Plus they lost to them H2H.

Ahh, the old H2H argument, yet in your 2019 scenario of Cornell NOT being a lock, they beat Notre Dame AT NOTRE DAME, yet ND was in and Cornell out. Does H2H mean anything or nothing? Which is it?
Gobigred
Joewillie78

If you go back and look at my post, here was the Cornell and Notre Dame comparison:Serious question that I don't have the answer for: When calculating top 5,10, 20 wins, is it based on when you played the team or based on the standings late in the season. Your figures suggest that Rutgers has NO top ten wins. Earlier in the season, Rutgers beat both Loyola and Princeton when they were both in the top ten. Please explain.

Notre Dame

RPI: 9
Top 5 wins: 1
Top 10 wins: 3
Top 20 wins: 3
SOS: 2
Bad losses: None

Cornell

RPI: 12
Top 5 wins: 0
Top 10 wins: 2
Top 20 wins: 0
SOS: 8
Bad losses: None

If you want to argue Cornell should have got in over Notre Dame because the RPI margin is close and therefore indistinguishable and the Top 10 wins are close and Cornell won H2H, go ahead. Again, I'm not saying you can't present the argument. But this is what the whole picture looks like and the whole picture pretty clearly favored Notre Dame.

Here is what the whole picture looks like right now between Johns Hopkins, Rutgers, and Georgetown

Hopkins:

RPI: 4
Top 5 wins: 0
Top 10 wins: 1
Top 20 wins: 5
SOS: 3
Bad losses: 1

Rutgers
RPI: 11
Top 5 wins: 0
Top 10 wins: 0
Top 20 wins: 3
SOS: 16
Bad losses: 1

Georgetown
RPI: 7
Top 5 wins: 0
Top 10 wins: 0
Top 20 wins: 2
SOS: 13
Bad losses: 1

If you look at those whole resumes and think there is any conceivable world in which case either of those teams gets in the tournament over Hopkins when you ALSO add in that Hopkins beats them H2H, then my only question to you is what possible reasoning are you using to reach that conclusion because I can't find any possible facts or logic that could in any way support it whatsoever.

[RURICK]

If anyone thinks they’re not a lock they should have to present the scenario where Hopkins doesn’t get in.

Hop loses to Maryland, loses to Rutgers in the first round of the BT tournament (I don't know if this match-up is possible), Michigan or Ohio State wins the BT tournament, Cornell wins out except for losing the ILT championship to Yale, Georgetown wins out but loses the BE championship to Denver. Eight at-larges are Duke, UVa, ND, UMd, RU, PSU, Cornell, GT (not in order).

Hopkins resume is better than Rutgers, Cornell, and Georgetown. Rutgers and Georgetown resumes aren’t even in the same universe to Hopkins. Plus they lost to them H2H.

Ahh, the old H2H argument, yet in your 2019 scenario of Cornell NOT being a lock, they beat Notre Dame AT NOTRE DAME, yet ND was in and Cornell out. Does H2H mean anything or nothing? Which is it?
Gobigred
Joewillie78

If you go back and look at my post, here was the Cornell and Notre Dame comparison:Serious question that I don't have the answer for: When calculating top 5,10, 20 wins, is it based on when you played the team or based on the standings late in the season. Your figures suggest that Rutgers has NO top ten wins. Earlier in the season, Rutgers beat both Loyola and Princeton when they were both in the top ten. Please explain.

Notre Dame

RPI: 9
Top 5 wins: 1
Top 10 wins: 3
Top 20 wins: 3
SOS: 2
Bad losses: None

Cornell

RPI: 12
Top 5 wins: 0
Top 10 wins: 2
Top 20 wins: 0
SOS: 8
Bad losses: None

If you want to argue Cornell should have got in over Notre Dame because the RPI margin is close and therefore indistinguishable and the Top 10 wins are close and Cornell won H2H, go ahead. Again, I'm not saying you can't present the argument. But this is what the whole picture looks like and the whole picture pretty clearly favored Notre Dame.

Here is what the whole picture looks like right now between Johns Hopkins, Rutgers, and Georgetown

Hopkins:

RPI: 4
Top 5 wins: 0
Top 10 wins: 1
Top 20 wins: 5
SOS: 3
Bad losses: 1

Rutgers
RPI: 11
Top 5 wins: 0
Top 10 wins: 0
Top 20 wins: 3
SOS: 16
Bad losses: 1

Georgetown
RPI: 7
Top 5 wins: 0
Top 10 wins: 0
Top 20 wins: 2
SOS: 13
Bad losses: 1

If you look at those whole resumes and think there is any conceivable world in which case either of those teams gets in the tournament over Hopkins when you ALSO add in that Hopkins beats them H2H, then my only question to you is what possible reasoning are you using to reach that conclusion because I can't find any possible facts or logic that could in any way support it whatsoever.

[RURICK]

My post was somehow incorporated into your post. Serious question: When determining top ten wins, why does it show that Rutgers has NO top ten wins. When they played Loyola and Princeton early in the year, both those teams were in the top ten. So, Rutgers should show to have 2 top ten wins. Please explain.

[pcowlax]

Most selection committees look at ranking of opponents, national or regional, based on final rankings. This makes complete sense, you shouldn’t get credit for a “top 10 win” based on an early season win over a team with a bogus pre-season ranking. This is why teams root for teams they have already beat to do well after that game. If the win over a ranked team was already banked, all your opponents could lose out after you played them and it wouldn’t affect your quality wins.

[wgdsr]

coaches and media polls are one of the few things that are deliberately spelled out as not to be used in selection. in season or end of season. "rankings" are derived from rpi. and tournament teams.

there are regional advisory boards (rep from every aq conference) that can provide input outside the committee, but it's vague as to what that input is.

[pcowlax]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.

[wgdsr]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.

correct. in fairness, football in particular does a lot of marketing around their teams throughout the year with this.

i don't think lacrosse fans would want to see how this looks in practice, the slew of old rankings wins vs. teams that didn't maintaiin.

but.... it'd be interesting to see (hint, hint, #s crunchers).

[Catbird]

Sorry laxreference, I don't think your scenario makes any sense...

PSU beats RU, MD beats JHU -> -> JHU is #3, Rutgers is #4 -> #3 JHU hosts OSU or Michigan, Rutgers #4 hosts OSU or Michigan in the first round. The two teams cannot meet until the B10 championship. Is there a scenario where Rutgers ends up #6 in the B10?

[laxreference]

Sorry laxreference, I don't think your scenario makes any sense...

PSU beats RU, MD beats JHU -> -> JHU is #3, Rutgers is #4 -> #3 JHU hosts OSU or Michigan, Rutgers #4 hosts OSU or Michigan in the first round. The two teams cannot meet until the B10 championship. Is there a scenario where Rutgers ends up #6 in the B10?

Very possible. I was playing a bit fast and loose with which seeds could theoretically end up with.

[10stone5]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.

correct. in fairness, football in particular does a lot of marketing around their teams throughout the year with this.

i don't think lacrosse fans would want to see how this looks in practice, the slew of old rankings wins vs. teams that didn't maintaiin.

but.... it'd be interesting to see (hint, hint, #s crunchers).

Does anyone really want that ?
You’d have to maintain point in time RPIs, poll rankings, schedule strength, head to head, quality wins and whatever other measures.
What’s the upside ?

[RURICK]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.
[/qu
I call BS.

[Farfromgeneva]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.
[/qu
I call BS.

It’s just wrong to use spot ratings with incomplete information. Point in time rankings or ratings nothing. No BS in what Pcow wrote he’s correct it’s kiddie mathematical understanding.

[10stone5]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.

I call BS.

It’s just wrong to use spot ratings with incomplete information. Point in time rankings or ratings nothing. No BS in what Pcow wrote he’s correct it’s kiddie mathematical understanding.

The one area that it would get gnarly in any assessment would be in the category of bad losses,
who you’ve beaten is another matter, its RPIs 1 through 20,
but bad losses is an odd measure - Loyola beat two top ten teams early, but they could end up with an RPI in the 25-30 range. Is that really a bad loss, teams in the 20-30 range aren’t really “bad” teams, they can and do beat top 20 teams.

[laxreference]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.

I call BS.

It’s just wrong to use spot ratings with incomplete information. Point in time rankings or ratings nothing. No BS in what Pcow wrote he’s correct it’s kiddie mathematical understanding.

The one area that it would get gnarly in any assessment would be in the category of bad losses,
who you’ve beaten is another matter, its RPIs 1 through 20,
but bad losses is an odd measure - Loyola beat two top ten teams early, but they could end up with an RPI in the 25-30 range. Is that really a bad loss, teams in the 20-30 range aren’t really “bad” teams, they can and do beat top 20 teams.

The problem with bad losses is the way the tiers are used in practice. A loss to the #21 team should be slightly more damaging than a loss to the #20 team. And the difference between a loss to #22 should be slightly more damaging than a loss to the #21 team. But the way the tiers are used sometimes, that's not the case.

It seems very silly to me that we should assign so much more weight when a loss comes against a team that is just barely #21 in the RPI vs #20. In the same way, why is a win vs a top-5 RPI team so much more valuable than a win vs the number 6 RPI team. The 1-5 vs 6-10 vs 11-20 tiers muddle everything up so much that they should be scrapped immediately.

(Also, just throwing in my two cents; the idea of looking at your opponent's ranking at the time of the game is a very crazy idea and we should not do that.)

[wgdsr]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.

I call BS.

It’s just wrong to use spot ratings with incomplete information. Point in time rankings or ratings nothing. No BS in what Pcow wrote he’s correct it’s kiddie mathematical understanding.

The one area that it would get gnarly in any assessment would be in the category of bad losses,
who you’ve beaten is another matter, its RPIs 1 through 20,
but bad losses is an odd measure - Loyola beat two top ten teams early, but they could end up with an RPI in the 25-30 range. Is that really a bad loss, teams in the 20-30 range aren’t really “bad” teams, they can and do beat top 20 teams.

The problem with bad losses is the way the tiers are used in practice. A loss to the #21 team should be slightly more damaging than a loss to the #20 team. And the difference between a loss to #22 should be slightly more damaging than a loss to the #21 team. But the way the tiers are used sometimes, that's not the case.

It seems very silly to me that we should assign so much more weight when a loss comes against a team that is just barely #21 in the RPI vs #20. In the same way, why is a win vs a top-5 RPI team so much more valuable than a win vs the number 6 RPI team. The 1-5 vs 6-10 vs 11-20 tiers muddle everything up so much that they should be scrapped immediately.

(Also, just throwing in my two cents; the idea of looking at your opponent's ranking at the time of the game is a very crazy idea and we should not do that.)

tiers is the dumbest thing about the entire "system".

[laxreference]

tiers is the dumbest thing about the entire "system".

And it's so easy to fix! 75 points for beating the best team. -75 for losing to the worst. 1 point for beating the worst, -1 for losing to the best team.

[wgdsr]

tiers is the dumbest thing about the entire "system".

And it's so easy to fix! 75 points for beating the best team. -75 for losing to the worst. 1 point for beating the worst, -1 for losing to the best team.

my system would give some credit for wins and losses near the bottom and top beyond 1. and more spread for top 20.

start at 170, drop 5+ for top 20, drop 2 for next 10, drop 1 down to 10 or 15. reverse for losses but start it higher (-20 pts for a loss to #1), climb less geometrical... start @ 2 pt gaps, but then 3... etc.

and there's plenty of opportunity cost in winning against low point squads, and losing to anyone.

[laxreference]

tiers is the dumbest thing about the entire "system".

And it's so easy to fix! 75 points for beating the best team. -75 for losing to the worst. 1 point for beating the worst, -1 for losing to the best team.

my system would give some credit for wins and losses near the bottom and top beyond 1. and more spread for top 20.

start at 170, drop 5+ for top 20, drop 2 for next 10, drop 1 down to 10 or 15. reverse for losses.

and there's plenty of opportunity cost in winning against low point squads, and losing to anyone.

I do a similar calculation, but instead of tiering, I use a decay factor that means that when comparing two resumes, both teams best wins get the most weight. Their next best wins get a bit less weight. You do it for every team in the same way and it emphasizes the most notable results without having to use any tiers at all.

[runrussellrun]

Right, and RPI as it is at the moment for current resumes about who teams have beaten and, at the end of the year when it is used for selection, the final RPIs. You don't get credit for beating a "top 10" RPI team because they were #6 when you played them in week 2 but end up at #27.

I call BS.

It’s just wrong to use spot ratings with incomplete information. Point in time rankings or ratings nothing. No BS in what Pcow wrote he’s correct it’s kiddie mathematical understanding.

The one area that it would get gnarly in any assessment would be in the category of bad losses,
who you’ve beaten is another matter, its RPIs 1 through 20,
but bad losses is an odd measure - Loyola beat two top ten teams early, but they could end up with an RPI in the 25-30 range. Is that really a bad loss, teams in the 20-30 range aren’t really “bad” teams, they can and do beat top 20 teams.

The problem with bad losses is the way the tiers are used in practice. A loss to the #21 team should be slightly more damaging than a loss to the #20 team. And the difference between a loss to #22 should be slightly more damaging than a loss to the #21 team. But the way the tiers are used sometimes, that's not the case.

It seems very silly to me that we should assign so much more weight when a loss comes against a team that is just barely #21 in the RPI vs #20. In the same way, why is a win vs a top-5 RPI team so much more valuable than a win vs the number 6 RPI team. The 1-5 vs 6-10 vs 11-20 tiers muddle everything up so much that they should be scrapped immediately.

(Also, just throwing in my two cents; the idea of looking at your opponent's ranking at the time of the game is a very crazy idea and we should not do that.)

tiers is the dumbest thing about the entire "system".

"...I see, no method, at all........sir"


WHO, or WHOM.....does the "system" work for? (year, after year, after year )

Hope Bergeron IS ok and skates tonight, but professional sports that use winning percentages as a "system" are just dumb.

....if only someone had a formula that looked at OOC games, included in the rpi

[wgdsr]

tiers is the dumbest thing about the entire "system".

And it's so easy to fix! 75 points for beating the best team. -75 for losing to the worst. 1 point for beating the worst, -1 for losing to the best team.

my system would give some credit for wins and losses near the bottom and top beyond 1. and more spread for top 20.

start at 170, drop 5+ for top 20, drop 2 for next 10, drop 1 down to 10 or 15. reverse for losses.

and there's plenty of opportunity cost in winning against low point squads, and losing to anyone.

I do a similar calculation, but instead of tiering, I use a decay factor that means that when comparing two resumes, both teams best wins get the most weight. Their next best wins get a bit less weight. You do it for every team in the same way and it emphasizes the most notable results without having to use any tiers at all.

the coaches have never been incented to make major changes. fans and mouthpieces are the loudest. it's weird.