2019 Bracketology Thread

[44WeWantMore]

Copying from the Big Red forum:

The key appears to be adding the RPI ranking to the Quality Win ranking plus half the strength of schedule ranking. The lower the number the better. At this point, after PSU defeated JHU, the numbers for the final 2 at large spots are:

Cornell: 20.50
Notre Dame: 21.00
JHU: 24.50
Maryland: 24.50
Syracuse: 27.00

Seems to match LaxBytes.

Code: Select all

  13 Cornell               94.42
  14 Notre Dame            87.52
  15 Johns Hopkins         68.83
  16 Maryland              68.83
  18 Syracuse              53.02

Apparent separation between the top two and bottom three. Cornell and ND seem safer, and JHU, UMD, and SU need to worry. If it is strictly by the numbers above, JHU and UMD are safe. If the committee says the numbers of the bottom three are within the margin of error, and goes to H2H, then SU claims the H2H over JHU, and JHU claims the H2H over UMD, leaving UMD out.

Final LaxBytes predictions do not seem to change my guesses above:

Code: Select all

  13 Cornell               86.25
  14 Loyola                86.25
  15 Notre Dame            79.94
  16 Johns Hopkins         68.26
  17 Maryland              57.78
  18 Syracuse              48.43

Pure numbers leave Syracuse out; H2H (SU>JHU>UMD) leaves UMD out.

[keno in reno]

There's no way Hopkins does not get in. They look like the 2016 Tar Heels right now.

[44WeWantMore]

Their waterbugs were quicker than our munchkins are.

[calourie]

As for seeding I see:

1. Penn State
2. Virginia
3. Penn
4. Duke
5. Yale
6. Towson
7. Loyola
8. Somebody

[Catbird]

My amateur attempt at a bracket:

1. Penn State (B1G)
Winner: UMBC (AE) @ Robert Morris (NEC)

8. Notre Dame (At Large)
Cornell (At Large)


2. Virginia (At Large - ACC Champ)
Marist (MAAC)

7. Loyola (At Large)
Syracuse (At Large)



3. Penn (Ivy)
Richmond (SOCON)

6. Towson (CAA)
Johns Hopkins (At Large)


4. Duke (At Large)
Georgetown (BE)

5. Yale (At Large)
Army (Patriot)

Syracuse squeaks in thanks to their Top 10 RPI. Then it's down to Maryland versus Cornell, I'm saying Cornell gets in for now due to severe SOS advantage, the other major metrics (such as QW) are very close but that one stands way out.

We'll see what happens.

[calourie]

Looks like a good shot, Catbird. It might be a sort of off year in terms of strength at the top of the bracket, but it looks strong on the bottom half.

[Big Dog]

I believe it is going to come down to Maryland vs. Cornell. Been saying that all along and nothing has changed my mind. Hope I'm right.

That probably would not be good for Cornell. Maryland has 5 top 20 RPI wins to Cornell's 2. Maryland has 4 losses to Cornell's 5, in either case all to top 10 RPI times. Maryland's RPI is slightly higher.

The Hopkins-Cornell comparision is closer, and in my opinion more likely to happen. Hopkins has 3 top 20 RPI wins to Cornell's 2, but both of Cornell's are in the top 10, while none of Hopkins' 3 wins are top 10. Cornell has 5 losses to Hopkins 7, and Cornell's losses are all to top 10 RPI teams, while one of Hopkins' losses is to #14 OSU. Hopkins does have a higher RPI.
I think Cornell wins this match-up.

All these teams have their little flaws, but the thing I think Hopkins' fans like you tend to overlook is that Hopkins is 8-7, and if this glaring void of being 0-6 against top 10 RPI teams.

That being said, as we all know Cornell would not be selected, because it's Hopkins.

Disagree with you on this, cooter. IMO, the intangible for the selection committee is to get new faces in. Of course, Big Red is not exactly a newbie, but still, being a blue blood is definitely not a help. Again, IMO.

[admin]

What the heck are "quality losses?"

When a lower-ranked team loses to a higher-ranked team. e.g. St. Bonaventure losing to PSU is a good loss. PSU losing to St. Bonaventure is a bad loss.

[stupefied]

What the heck are "quality losses?"

When a lower-ranked team loses to a higher-ranked team. e.g. St. Bonaventure losing to PSU is a good loss. PSU losing to St. Bonaventure is a bad loss.

In other words never have a team on schedule that could turn out to have a mediocre rank because little benefit and alot of downside.

Does the deficit in a loss to a ranked team even matter ? Issue if it doesn't because game should be at least be competitive if losing team is getting credit

[admin]

The Pre-Tournament Computer Rankings look very similar to the pre-Ivy and Patriot Finals but some differences. Besides some teams moving spots, we've added asterisks to signify that a team is still playing. Also, Winning %, Strength of Schedule, Best Win, and Simplified RPI. And, after looking at the chart...

1. Automatic Qualifiers are a good thing and... It's also a not so good thing. Hard to look at the entire list and see who's actually playing.
2. Based on Ws and Ls, Syracuse is out. But it's doubtful that the committee will base their decision on Ws and Ls.
3. Based on SRPI (Simplified RPI - No adjustments for home and away, Wins against Top 5, Top 10, Top 20, Etc.), Cornell is out.
4. Based on Best Wins, Cornell is out.

In conclusion, from where we're sitting, Cornell is out.

[Hawkeye]

In conclusion, from where we're sitting, Cornell is out.

Page 17, today, 12:36am:
"Penn/Yale loser, Duke, Cornell, Virginia, Maryland, and Loyola are in"

Page 19, today, 12:15pm:
"From where we're sitting, from our logical perspective, Cornell and UMD are the safest, JHU is relatively safe, and... ND and SU are fighting for that last spot."

What changed for you to join those of us on the "Cornell is out" side?

[houndace1]

wouldn't the penn win today, which would help MD slide in to the field, also concurrently help hopkins?

The penn win moves MD's rpi to i believe top 10 which further strengthens Hopkins case for the two MD wins

[admin]

In other words never have a team on schedule that could turn out to have a mediocre rank because little benefit and alot of downside.

Does the deficit in a loss to a ranked team even matter ? Issue if it doesn't because game should be at least be competitive if losing team is getting credit.

Or, worded differently, have a difficult schedule. Lots of high-ranked teams (relative to your team). But no so many that you're sub-.500. So, if you're the #15 team in the country with a 12 game schedule, play 6 games vs. Teams #1-#14 and 6 games vs. #16-#30. Which seems like a good policy to me.

Game score matters for some rankings, not at all for others, somewhere in between for others. Same is true for losing teams "getting credit". Many rankings, including rankings that the committee likes, give a losing team a boost even if they lose. Keep in mind, with RPI, 75% of your Rating is based on your opponents's record.

In terms of the committee, as they're a human element involved, scores matter. e.g. JHU lost yesterday. But I'd be astounded if the committee saw this loss to the #1 team in the country as an L and nothing else.

[Hawkeye]

The penn win moves MD's rpi to i believe top 10

Unless something has been updated that I haven't seen, Syracuse held on to 10th. Maryland finished 11th.

[admin]

What changed for you to join those of us on the "Cornell is out" side?

Based on Ws and Ls, We'd still say SU and Notre Dame, in that order. Today's games created more distance between them. Based on RPI (actually SRPI, but whatever), it's Cornell. We don't like RPI. Logic and the universe doesn't like RPI. But the committee and group-think within the lacrosse world likes RPI. Similarly, the mind-meld likes "Best Wins". Almost as much as they like head-to-head. So... we've said our piece about who we think, in a perfect world, should be in the tournament (i.e. Penn State through Army with OSU playing Army in the Play-In game) and, now, it's time to say what's going to actually happen.

[admin]

Hawk, I'd really like to see the actual formulas for each of the relevant RPIs that we put stock into. At one time, an RPI was an RPI. Ws and Ls and 25%, 50%, 25%. Today, very different.

[Hawkeye]

What changed for you to join those of us on the "Cornell is out" side?

Based on Ws and Ls, We'd still say SU and Notre Dame, in that order. Today's games created more distance between them. Based on RPI (actually SRPI, but whatever), it's Cornell. We don't like RPI. Logic and the universe doesn't like RPI. But the committee and group-think within the lacrosse world likes RPI. Similarly, the mind-meld likes "Best Wins". Almost as much as they like head-to-head. So... we've said our piece about who we think, in a perfect world, should be in the tournament (i.e. Penn State through Army with OSU playing Army in the Play-In game) and, now, it's time to say what's going to actually happen.

In this perfect world you've constructed, there is no play-in game, because there are no AQs. The play-in game literally exists because the field size is 16 (8 guaranteed at-larges) and there are 9 AQs. But that's another discussion.

[Hawkeye]

Hawk, I'd really like to see the actual formulas for each of the relevant RPIs that we put stock into. At one time, an RPI was an RPI. Ws and Ls and 25%, 50%, 25%. Today, very different.

Yes, I agree, I would prefer increased transparency.

[stupefied]

In other words never have a team on schedule that could turn out to have a mediocre rank because little benefit and alot of downside.

Does the deficit in a loss to a ranked team even matter ? Issue if it doesn't because game should be at least be competitive if losing team is getting credit.

Or, worded differently, have a difficult schedule. Lots of high-ranked teams (relative to your team). But no so many that you're sub-.500. So, if you're the #15 team in the country with a 12 game schedule, play 6 games vs. Teams #1-#14 and 6 games vs. #16-#30. Which seems like a good policy to me.

Game score matters for some rankings, not at all for others, somewhere in between for others. Same is true for losing teams "getting credit". Many rankings, including rankings that the committee likes, give a losing team a boost even if they lose. Keep in mind, with RPI, 75% of your Rating is based on your opponents's record.

In terms of the committee, as they're a human element involved, scores matter. e.g. JHU lost yesterday. But I'd be astounded if the committee saw this loss to the #1 team in the country as an L and nothing else.

Believe that the human element /eye test should always factor but any of these computer models should also assign credits to a loss by its competitiveness. Most basic input is score . A loss of eight is different than a loss of five which is different than a loss of one .

[CU77]

Hawk, I'd really like to see the actual formulas for each of the relevant RPIs that we put stock into. At one time, an RPI was an RPI. Ws and Ls and 25%, 50%, 25%. Today, very different.

While RPI is computed differently in different sports, mens lax RPI is indeed 25% your own record, 50% average record of your opponents (excluding games against you), and 25% the average record of each of your opponents (computed excluding games against the opponent, but not against you).

But RPI is then just the starting point for a confusing list of selection criteria.