PDA

View Full Version : RPI 2010-2011 V: The "Chase" to AC & Beyond



Pages : 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 [46] 47 48 49 50 51

DrDemento
03-18-2011, 12:15 PM
If we work on it-we could finish this thread just in time for the Sunday NCAA announcement show. Then we can have the new thread for the Tournament (hoping).

engineerhockeyfan
03-18-2011, 12:21 PM
I'm sorry to be argumentative but your entire premise is wrong. If Colgate beats Yale then we still have a 67.8% chance of getting in.

Don't be sorry, that's what this is all about. I am sure that your math is correct, but if Colgate goes all the way, then we have Yale, Union and Colgate going to the show.
I have seen in the past when the powers that be ignored the math and went another way. I will stick with Yale and let the rest take care of itself. JMO.

GRussinko
03-18-2011, 12:31 PM
Don't be sorry, that's what this is all about. I am sure that your math is correct, but if Colgate goes all the way, then we have Yale, Union and Colgate going to the show.
I have seen in the past when the powers that be ignored the math and went another way. I will stick with Yale and let the rest take care of itself. JMO.

There are scenarios that have 5 ECAC teams in the tournament too, and that includes RPI. When has the hockey selection committee gone contrary to what the pairwise says? We're not talking basketball here. The reason all this discussion is going on is that we're down to a small enough number of games left where we can quantitatively look at every result and what it does to RPI. We're several months past "just win and it'll work out" or "Yale just needs to win and it'll work out".

burgie12
03-18-2011, 12:35 PM
I have seen in the past when the powers that be ignored the math and went another way. I will stick with Yale and let the rest take care of itself. JMO.
When? Provide an example when the PWR did not properly pick the 16 teams that made the tournament. You can probably find an example of when #13 and #14 got switched for seemingly no reason, or got placed in a regional that wasn't expected. But, I don't know of any time when the PWR did not correctly pick the teams that made up each of the seeding bands.

It really is just simple mathematics and a purely objective process. Those who claim that the Engineers won't make the tournament just because of their play in February are lying to themselves.

FlagDUDE08
03-18-2011, 12:35 PM
There are scenarios that have 5 ECAC teams in the tournament too, and that includes RPI. When has the hockey selection committee gone contrary to what the pairwise says? We're not talking basketball here. The reason all this discussion is going on is that we're down to a small enough number of games left where we can quantitatively look at every result and what it does to RPI. We're several months past "just win and it'll work out" or "Yale just needs to win and it'll work out".

It must have been back when the last 16 games mattered more. However, you are correct in that the PWR has correctly predicted every team in the tournament in the past years; there's still some question regarding tiebreakers and whether it goes to RPI or H2H, or even something else. We may just find out that answer this year.

burgie12
03-18-2011, 12:47 PM
There are scenarios that have 5 ECAC teams in the tournament too, and that includes RPI.
Here's one example that puts 5 teams in:

Atlantic Hockey
Semifinal #2: Air Force defeats Holy Cross
Semifinal #1: RIT defeats Connecticut
Championship game: RIT defeats Air Force

CCHA
Semifinal #2: Miami defeats Notre Dame
Semifinal #1: Michigan defeats Western Michigan
Championship game: Michigan defeats Miami
Consolation game: Notre Dame defeats Western Michigan

ECAC
Semifinal #2: Cornell defeats Dartmouth
Semifinal #1: Colgate defeats Yale
Championship game: Colgate defeats Cornell
Consolation game: Dartmouth defeats Yale

Hockey East
Semifinal #2: Merrimack defeats New Hampshire
Semifinal #1: Boston College defeats Northeastern
Championship game: Boston College defeats Merrimack

WCHA
Semifinal #2: Denver defeats Bemidji State
Semifinal #1: Colorado College defeats North Dakota
Championship game: Colorado College defeats Denver
I'm amused that 3 of the ECAC teams are #4 seeds, meaning that Yale would be stuck playing RIT.

there's still some question regarding tiebreakers and whether it goes to RPI or H2H, or even something else. We may just find out that answer this year.
It's doubtful that we'll find out unless Rensselaer gets tied in a critical position (12/13 or in/out). There are few comparisons that do not involve the Engineers which break opposite of the RPI. And, if the Tute is tied for 13th or similar, then the NCAA could have always switched match-ups based on attendance or saving flights. So, it's possible that we'll answer that question, but it's unlikely.

jmhusker
03-18-2011, 12:51 PM
Cornell - RPI at 69.8805%
http://uploads.cuttlefishtech.com/cornell.html
Dartmouth - RPI at 67.5558%
http://uploads.cuttlefishtech.com/dartmouth.html

Colgate - RPI at 67.8593%
http://uploads.cuttlefishtech.com/colgate.html
Yale - RPI at 68.5779%
http://uploads.cuttlefishtech.com/yale.html

Great going on the effort, Reilly. The Cornell/DC game is a puzzler for me. I don't think we should want Cornell to win and here's why:

I ran a what-if on all 16 permutations for the ECAC tournament (I did not include the tie possibility). Of those 16 outcomes, we still rank out of the tournament in all 4 instances of a Cornell win, in 3 instances of a Colgate win, 2 times when DC wins and once when Yale wins. This downside alone makes the 2.3% delta between DC and Cornell not seem worth it. Beyond that, there are some other areas where the Krach model falls short as well as some positives that come from a possible Dartmouth sweep. As far as the Krach model goes, should Cornell meet Colgate in the consolation game it assumes that Colgate only has a 29% chance of winning. I hold that the chance is at least 50% given how the teams are playing now and I'd even favor Colgate. Krach also assumes that the D/C loser plays Colgate 90% of the time (while the winner plays Yale). In other words, the model is telling us to hope for Cornell to win because there's only a 22% chance of them beating Yale in the final while there is a 22.5% chance that Dartmouth loses to Colgate in the consolation - and 2 DC losses are worth more than 2 Cornell losses.

That might be true but I hold that two Cornell losses are still quite valuable. On top of that, a final with DC and Yale keeps the cutoff for the tournament at 15. Furthermore, 2 DC wins will flip some of the DC PWR comparisons with the others on the bubble.

BTW... out of the 16 scenarios, our two best outcomes put us tied for 12th. Those two outcomes are Yale-Dart-Gate-Cor and Dart-Yale-Gate-Cor.

So call it the conservative strategy if you like but we should be pulling for Dartmouth. If the Krach ratings were based on recent play only then I'm confident it would say as much.

FlagDUDE08
03-18-2011, 12:53 PM
Here's one example that puts 5 teams in:

Atlantic Hockey
Semifinal #2: Air Force defeats Holy Cross
Semifinal #1: RIT defeats Connecticut
Championship game: RIT defeats Air Force

CCHA
Semifinal #2: Miami defeats Notre Dame
Semifinal #1: Michigan defeats Western Michigan
Championship game: Michigan defeats Miami
Consolation game: Notre Dame defeats Western Michigan

ECAC
Semifinal #2: Cornell defeats Dartmouth
Semifinal #1: Colgate defeats Yale
Championship game: Colgate defeats Cornell
Consolation game: Dartmouth defeats Yale

Hockey East
Semifinal #2: Merrimack defeats New Hampshire
Semifinal #1: Boston College defeats Northeastern
Championship game: Boston College defeats Merrimack

WCHA
Semifinal #2: Denver defeats Bemidji State
Semifinal #1: Colorado College defeats North Dakota
Championship game: Colorado College defeats Denver
I'm amused that 3 of the ECAC teams are #4 seeds, meaning that Yale would be stuck playing RIT.

It's doubtful that we'll find out unless Rensselaer gets tied in a critical position (12/13 or in/out). There are few comparisons that do not involve the Engineers which break opposite of the RPI. And, if the Tute is tied for 13th or similar, then the NCAA could have always switched match-ups based on attendance or saving flights. So, it's possible that we'll answer that question, but it's unlikely.

It's not only us, there are a couple close call situations across the bubble teams at the in/out line. I just hope, come Monday, we aren't posting in the MTGS thread.

jmhusker
03-18-2011, 12:55 PM
Don't be sorry, that's what this is all about. I am sure that your math is correct, but if Colgate goes all the way, then we have Yale, Union and Colgate going to the show.
I have seen in the past when the powers that be ignored the math and went another way. I will stick with Yale and let the rest take care of itself. JMO.
If Colgate goes all the way we still do pretty well. If Colgate and NE and BSU do then we've got problems. Root against NE, BSU and Cornell!

knsmith
03-18-2011, 01:01 PM
Great going on the effort, Reilly. The Cornell/DC game is a puzzler for me. I don't think we should want Cornell to win and here's why:

I ran a what-if on all 16 permutations for the ECAC tournament (I did not include the tie possibility). Of those 16 outcomes, we still rank out of the tournament in all 4 instances of a Cornell win, in 3 instances of a Colgate win, 2 times when DC wins and once when Yale wins. This downside alone makes the 2.3% delta between DC and Cornell not seem worth it. Beyond that, there are some other areas where the Krach model falls short as well as some positives that come from a possible Dartmouth sweep. As far as the Krach model goes, should Cornell meet Colgate in the consolation game it assumes that Colgate only has a 29% chance of winning. I hold that the chance is at least 50% given how the teams are playing now and I'd even favor Colgate. Krach also assumes that the D/C loser plays Colgate 90% of the time (while the winner plays Yale). In other words, the model is telling us to hope for Cornell to win because there's only a 22% chance of them beating Yale in the final while there is a 22.5% chance that Dartmouth loses to Colgate in the consolation - and 2 DC losses are worth more than 2 Cornell losses.

That might be true but I hold that two Cornell losses are still quite valuable. On top of that, a final with DC and Yale keeps the cutoff for the tournament at 15. Furthermore, 2 DC wins will flip some of the DC PWR comparisons with the others on the bubble.

BTW... out of the 16 scenarios, our two best outcomes put us tied for 12th. Those two outcomes are Yale-Dart-Gate-Cor and Dart-Yale-Gate-Cor.

So call it the conservative strategy if you like but we should be pulling for Dartmouth. If the Krach ratings were based on recent play only then I'm confident it would say as much.

I agree with your basic challenge to premise of using raw KRACH for weighting probabilities (especially in the ECAC consolation game), but how can you purport to come to these conclusions without accounting for the goings on in other conferences? For example, my understanding is that if we were to get perfect results in the CCHA, we would likely be in even with an upset ECAC winner (Cornell or Colgate). Indeed, a Cornell win tonight helps us in that scenario.

burgie12
03-18-2011, 01:09 PM
It's not only us, there are a couple close call situations across the bubble teams at the in/out line.
It still has to be a seed boundary (4/5, 8/9, 12/13, in/out) and a 2-team tie (because past precedent shows that 3+-team ties get broken by RPI). And, the reason that I said that it is likely to be Rensselaer-related is because the Engineers are the only team with a PWR so far above their RPI. It's certainly not exclusive to the Tute, but they definitely have the most examples.

burgie12
03-18-2011, 01:12 PM
but how can you purport to come to these conclusions without accounting for the goings on in other conferences?
Winner! There are 13 games left to be played outside of the ECAC tournament this weekend. Each of them will have an impact on our chances.

DrDemento
03-18-2011, 01:24 PM
Now I'm even more confused on who to root for in the Dartmouth/Cornell game. I know every game has an effect but for some reason that is the one I just cannot figure out. I have read everyone's posts on it-and still find myself not knowing.

Senna
03-18-2011, 01:26 PM
The many posts over the last several days remind me of a little-used First Church cheer:
We like math,
we like beer.
We are the Engineers!

Reilly, you're the man.

Couldnt agree with you more. Im not a math guy or an engineer....and now I know why. This is some fantastic stuff.

FlagDUDE08
03-18-2011, 01:26 PM
Now I'm even more confused on who to root for in the Dartmouth/Cornell game. I know every game has an effect but for some reason that is the one I just cannot figure out. I have read everyone's posts on it-and still find myself not knowing.

Take a wall. Put some duct tape down the center. Label one side Dartmouth, label one side Cornell. Blindfold yourself. Spin around a few times. Fling something dirty at the wall. Wherever it sticks, there's your answer. :D

knsmith
03-18-2011, 01:35 PM
Now I'm even more confused on who to root for in the Dartmouth/Cornell game. I know every game has an effect but for some reason that is the one I just cannot figure out. I have read everyone's posts on it-and still find myself not knowing.

It's one of those things that depends on what happens elsewhere. I believe that if we generally get a lot of help elsewhere (i.e. ND, WMU losses; CC win [these are all pretty huge; I speculate that if/when RHamilton runs his "lock" scenarios, getting these three results will put us at 95%+ chance of getting in) then a Cornell WIN is better for us (because we don't need them to drop as a TUC). If we don't get help elsewhere, we absolutely need Cornell to lose because them dropping out as a TUC is more or less our best chance of flipping things.

burgie12
03-18-2011, 01:38 PM
Now I'm even more confused on who to root for in the Dartmouth/Cornell game. I know every game has an effect but for some reason that is the one I just cannot figure out. I have read everyone's posts on it-and still find myself not knowing.
It doesn't matter too much. The bigger concern is that whomever loses that game today loses again tomorrow. Two Dartmouth losses puts them solidly below us in RPI and PWR. Two Cornell losses puts them very, very close to the TUC line (or just plain out).

ETA: And... What knsmith said.

DrDemento
03-18-2011, 01:46 PM
Thanks guys. It is still up in the air for me but at least i am looking at it the way you guys are. Elsewhere really does matter.

engineerhockeyfan
03-18-2011, 02:01 PM
When? Provide an example when the PWR did not properly pick the 16 teams that made the tournament. You can probably find an example of when #13 and #14 got switched for seemingly no reason, or got placed in a regional that wasn't expected. But, I don't know of any time when the PWR did not correctly pick the teams that made up each of the seeding bands.

It really is just simple mathematics and a purely objective process. Those who claim that the Engineers won't make the tournament just because of their play in February are lying to themselves.

I distinctly remember the last time RPI got into the tournament, Colorado College having a fit that they were left out. I can't say for sure if they had a better PWR, but they certainly felt that they got screwed.

supe
03-18-2011, 02:21 PM
I distinctly remember the last time RPI got into the tournament, Colorado College having a fit that they were left out. I can't say for sure if they had a better PWR, but they certainly felt that they got screwed.

Didn't that year lead to the now gone "CC Rule," where regular season champs got an auto bid along with conf tourney winners?

CC felt screwed, but it wasn't because the committee did anything shady. In fact it was the committee's strict adherence to the numbers that led to CC complaining.