Bracketology 2023

Article from new ECH blog start of Feb.:
www.echlocker.com/blog/ech-weekend-review-january2022-ep9zm-lzctd-gb939

Bridgeport, CT: #2 Quinnipiac vs. #15 Omaha, #7 Denver vs. #11 Cornell

Recent CHN:
www.collegehockeynews.com/news/2023/02/14_Bracket-ABCs-First-Look.php

2. Quinnipiac vs. 15. Northeastern
7. St. Cloud State vs. 10. Cornell


TLD(wanna)R: Second seed and Bridgeport regional sounds good to us!

That's a 3 seed and wholly dependent on QU being 2 and Cornell being 10. Anything else, and all bets are off. Long way to go.

adamw
That's a 3 seed and wholly dependent on QU being 2 and Cornell being 10. Anything else, and all bets are off. Long way to go.

As you know, a lot of people who post on this forum like to post speculation like this even though it's meaningless this far ahead of time. For example, if Cornell loses one of its four remaining regular season games, which is completely plausible, and doesn't win a home quarterfinal series, which also is completely plausible, there won't be an NCAA bid to talk about. Just like some eLynah members had us winning the Ivy League mens basketball title this winter after our fast start.

Can we please not quote any pairwise probability matrices this year

BearLover
Can we please not quote any pairwise probability matrices this year

I pulled up the pairwise probability matrix and my dog licked my phone. Pretty sure that means six more weeks of speculation.

Because you asked nicely, here are the www.playoffstatus.com probabilities for now:

ECAC Tourney:
.85 Advance to SF
.41 Advance to F
.18 ECAC Champion

NCAA Tourney:
.90 Qualification
.40 Advance to QF
.17 Advance to SF
.08 Advance to F
.03 NCAA Champion

Dafatone

BearLover
Can we please not quote any pairwise probability matrices this year

I pulled up the pairwise probability matrix and my dog licked my phone. Pretty sure that means six more weeks of speculation.

Ha!!

After the Brown win, we are 13 in PWR.

and Mich tech could hold on and get us back to 12.

could also use ND losing to mich tomorrow just to make their path harder to .500

The playoff probability models continue to yield absurd results. If “weighted by KRACH” indeed means abmarks’ interpretation thereof (see discussion in other thread that I’m too lazy to dig up in which abmarks accused me of not reading something I did read), then the model on CHN gives Q a 64% chance of winning the ECAC tournament (ridiculous even if Q is the clear best team). But even more absurd is that it gives the bottom four seeds (Dartmouth, Yale, Brown, and Princeton) collectively a .1% chance of winning the ECAC. That is to say, one of those four teams would win the ECAC one out of one thousand times. The model gives the bottom eight seeds a collective
2.1% chance of winning the ECAC. StL has a 3% chance of winning, which means, per the model, someone other than Q/H/Cornell wins the ECAC 5% of the time.

Let’s please stop citing these prediction models.

www.playoffstatus.com

ECAC Tournament

        SF    F   Ch
Qpc    .99  .82  .51
Hvd    .97  .62  .30
Cor    .84  .34  .13
SLU    .63  .13  .04
Field  .57  .07  .02

Trotsky
www.playoffstatus.com

ECAC Tournament

        SF    F   Ch
Qpc    .99  .82  .51
Hvd    .97  .62  .30
Cor    .84  .34  .13
SLU    .63  .13  .04
Field  .57  .07  .02

Models like this are what happens when you use whatever inputs you want without backwards-testing the methodology whatsoever.

KRACH is a decent model for ranking teams from top to bottom, but it hugely overestimates the probability of a higher-ranked team beating a lower-ranked team.

They aren't predictive models. They are descriptive analytics which summarize past performance. But conditions in the playoffs aren't like the range of season results, so it is not theoretically justified to interpret them as predictive.

The reason to cite them is they drive a certain personality type insane, and that is funny.

Trotsky
They aren't predictive models. They are descriptive analytics which summarize past performance. But conditions in the playoffs aren't like the range of season results, so it is not theoretically justified to interpret them as predictive.

The reason to cite them is they drive a certain personality type insane, and that is funny.

They’re predictive models. They literally provide a probability for each team which is supposed to represent its chances of making each round of the NCAAs. KRACH is not a predictive model, if that’s what you mean—which is part of the reason why it’s absurd to base a predictive model off of KRACH. The problem with these predictive models is not that “conditions in the playoffs aren't like the range of season results,” it’s that they’re extrapolating future probabilities from a system of ranking teams based on past performance.

BearLover
The problem with these predictive models is not that “conditions in the playoffs aren't like the range of season results,” it’s that they’re extrapolating future probabilities from a system of ranking teams based on past performance.

That problem is because the future games aren't enough like the past games to justify extrapolation.

I'm not sure if you are being deliberately dense to have fun trolling (in which case, OK, I get that) or you honestly do not understand this.

the issue is mostly that probabilities like this are unfalsifiable unless the projection is 100/0. i call this the Nate Silver Annoyance. the solution to this is to not care that much.

ugarte
the issue is mostly that probabilities like this are unfalsifiable unless the projection is 100/0.

Whether and under what interpretation that would constitute falsifiability would be a really interesting philosophical question.

But it illustrates the problem perfectly. If Cornell and Harvard were the only two teams, and met once in the RS to determine the home team for the NCAA title, and Cornell won in the RS, then our descriptive analytic for the RS would be 1.00 and Harvard would be .00. It would not be theoretically justifiable to extrapolate this to a 100% prediction of Cornell winning the next meeting in the NCAA final. That, in microcosm, is what is happening here when we take these numbers, or for that matter KRACH or PWR, as predictive of the tournament result.

But it's fun to spin up the trogs.

Also good call on the Silver Slugger. If I hear one more idiot say Silver got 2016 "wrong"... No, you morons. He said a ten percent chance of dumbfuckery. 1 in 10. And that's what we got: the 1 in 10.

Trotsky
Also good call on the Silver Slugger. If I hear one more idiot say Silver got 2016 "wrong"... No, you morons. He said a ten percent chance of dumbfuckery. 1 in 10. And that's what we got: the 1 in 10.

The same people who say he got 2016 wrong would not, if they were told by a statistician that there's a roughly 16.7% chance of rolling a six on a single die roll then shown a single die roll that lands on six, claim the statistician got that "wrong." But in politics, dumb gotta dumb.

Trotsky
Also good call on the Silver Slugger. If I hear one more idiot say Silver got 2016 "wrong"... No, you morons. He said a ten percent chance of dumbfuckery. 1 in 10. And that's what we got: the 1 in 10.

Wasn't his call closer to 3 in 10?

I will say that you have to differentiate between 538 the election prediction site and Nate Silver the twitter bad take generator.

Dafatone

Trotsky
Also good call on the Silver Slugger. If I hear one more idiot say Silver got 2016 "wrong"... No, you morons. He said a ten percent chance of dumbfuckery. 1 in 10. And that's what we got: the 1 in 10.

Wasn't his call closer to 3 in 10?

I will say that you have to differentiate between 538 the election prediction site and Nate Silver the twitter bad take generator.

19 sticks in my head actually.

Nate was wonderful until he wasn't. But if you speak in public that many times, a few times are going to be awful.

That's just probability.

Trotsky
Also good call on the Silver Slugger. If I hear one more idiot say Silver got 2016 "wrong"... No, you morons. He said a ten percent chance of dumbfuckery. 1 in 10. And that's what we got: the 1 in 10.

i definitely don't want to go on a tangent about silver but ... my point is that silverism is mostly bad not mostly good. the unfalsifiability of "1 in 10" and the general focus on the horse race over the substance (and silver constantly acting like the pundit he once claimed to be the antidote for is what drives me hold on i think i just went on a tangent please someone stop me from posting i don't want to do th

For those questioning the Krach based predictions, particularly the chance of winning the ECAC tournament, here's a comparable analysis.

I was curious about the Tournament Champ distribution by seed but couldn't find the data easily; and even if I found it there'd be a small sample size. What I did find was data from the 1985 to 2019 NCAA men's hoop tournaments.

With 64 team fields, split into 4 regions and teams seeded 1-16 in each, there are essentially four 16 teamers each year to reach the Final Four. Look at 35 years of data and we can examine actual results from 140 of these 16-team tournaments.

Basketball and hockey are of course not the same, and I'm looking at 16 team fields rather than 12, but I think the distribution is informative nonetheless.

Teams that made the Final Four are effectively the champion of a 16 team field. Looking at this it makes the 0% probability of winning for the bottom half of the ecac field look a lot more realistic since the bottom half in the hoop data shows zero to one percent for all but the 11-seed's three percent.





NCAA Men's Basketball tournament: 1985-2019
Final Four team seeds (or winners of 16 te.tournaments)

Seed	MadeFF: Pct
        (Champ)
1	57	41%
2	29	21%
3	17	12%
4	13	9%
5	7	5%
6	3	2%
7	3	2%
8	5	4%
9	1	1%
10	1	1%
11	4	3%
12	0	0%
13	0	0%
14	0	0%
15	0	0%
16	0	0%
Total	140

ugarte

Trotsky
Also good call on the Silver Slugger. If I hear one more idiot say Silver got 2016 "wrong"... No, you morons. He said a ten percent chance of dumbfuckery. 1 in 10. And that's what we got: the 1 in 10.

i definitely don't want to go on a tangent about silver but ... my point is that silverism is mostly bad not mostly good. the unfalsifiability of "1 in 10" and the general focus on the horse race over the substance (and silver constantly acting like the pundit he once claimed to be the antidote for is what drives me hold on i think i just went on a tangent please someone stop me from posting i don't want to do th

I don't know what Silverism is, but 538 was an excellent source of information on statistical analysis for lofo voters and bettors, er, sports fans, and it went the extra mile in demythologizing much of probability theory, in a way that still wasn't entirely wrong from a quant methods POV. I give that an A plus. I lay most of the blame for 538 being taken wrongly to other media who didn't know or care about nuance. That wasn't the Two Nates' problem.

abmarks
For those questioning the Krach based predictions, particularly the chance of winning the ECAC tournament, here's a comparable analysis.

I was curious about the Tournament Champ distribution by seed but couldn't find the data easily; and even if I found it there'd be a small sample size. What I did find was data from the 1985 to 2019 NCAA men's hoop tournaments.

With 64 team fields, split into 4 regions and teams seeded 1-16 in each, there are essentially four 16 teamers each year to reach the Final Four. Look at 35 years of data and we can examine actual results from 140 of these 16-team tournaments.

Basketball and hockey are of course not the same, and I'm looking at 16 team fields rather than 12, but I think the distribution is informative nonetheless.

Teams that made the Final Four are effectively the champion of a 16 team field. Looking at this it makes the 0% probability of winning for the bottom half of the ecac field look a lot more realistic since the bottom half in the hoop data shows zero to one percent for all but the 11-seed's three percent.





NCAA Men's Basketball tournament: 1985-2019
Final Four team seeds (or winners of 16 te.tournaments)

Seed	MadeFF: Pct
        (Champ)
1	57	41%
2	29	21%
3	17	12%
4	13	9%
5	7	5%
6	3	2%
7	3	2%
8	5	4%
9	1	1%
10	1	1%
11	4	3%
12	0	0%
13	0	0%
14	0	0%
15	0	0%
16	0	0%
Total	140

Really good stuff, thanks. I think the 11 seed is a decent spot b/c in chalk you get a 5 then a 4 and by then maybe some chaos has eliminated some of 1-3 top teams for an easier path. Or it could just be a random coincidence. Of course the ECAC reseeds so there’s a big difference Vs the basketball bracket approach.

CU2007

abmarks
For those questioning the Krach based predictions, particularly the chance of winning the ECAC tournament, here's a comparable analysis.

I was curious about the Tournament Champ distribution by seed but couldn't find the data easily; and even if I found it there'd be a small sample size. What I did find was data from the 1985 to 2019 NCAA men's hoop tournaments.

With 64 team fields, split into 4 regions and teams seeded 1-16 in each, there are essentially four 16 teamers each year to reach the Final Four. Look at 35 years of data and we can examine actual results from 140 of these 16-team tournaments.

Basketball and hockey are of course not the same, and I'm looking at 16 team fields rather than 12, but I think the distribution is informative nonetheless.

Teams that made the Final Four are effectively the champion of a 16 team field. Looking at this it makes the 0% probability of winning for the bottom half of the ecac field look a lot more realistic since the bottom half in the hoop data shows zero to one percent for all but the 11-seed's three percent.





NCAA Men's Basketball tournament: 1985-2019
Final Four team seeds (or winners of 16 te.tournaments)

Seed	MadeFF: Pct
        (Champ)
1	57	41%
2	29	21%
3	17	12%
4	13	9%
5	7	5%
6	3	2%
7	3	2%
8	5	4%
9	1	1%
10	1	1%
11	4	3%
12	0	0%
13	0	0%
14	0	0%
15	0	0%
16	0	0%
Total	140

Really good stuff, thanks. I think the 11 seed is a decent spot b/c in chalk you get a 5 then a 4 and by then maybe some chaos has eliminated some of 1-3 top teams for an easier path. Or it could just be a random coincidence. Of course the ECAC reseeds so there’s a big difference Vs the basketball bracket approach.

There is way less variance in a game of basketball than a game of hockey. I honestly don’t think this is a very helpful comparison.

BearLover

CU2007

abmarks
For those questioning the Krach based predictions, particularly the chance of winning the ECAC tournament, here's a comparable analysis.

I was curious about the Tournament Champ distribution by seed but couldn't find the data easily; and even if I found it there'd be a small sample size. What I did find was data from the 1985 to 2019 NCAA men's hoop tournaments.

With 64 team fields, split into 4 regions and teams seeded 1-16 in each, there are essentially four 16 teamers each year to reach the Final Four. Look at 35 years of data and we can examine actual results from 140 of these 16-team tournaments.

Basketball and hockey are of course not the same, and I'm looking at 16 team fields rather than 12, but I think the distribution is informative nonetheless.

Teams that made the Final Four are effectively the champion of a 16 team field. Looking at this it makes the 0% probability of winning for the bottom half of the ecac field look a lot more realistic since the bottom half in the hoop data shows zero to one percent for all but the 11-seed's three percent.





NCAA Men's Basketball tournament: 1985-2019
Final Four team seeds (or winners of 16 te.tournaments)

Seed	MadeFF: Pct
        (Champ)
1	57	41%
2	29	21%
3	17	12%
4	13	9%
5	7	5%
6	3	2%
7	3	2%
8	5	4%
9	1	1%
10	1	1%
11	4	3%
12	0	0%
13	0	0%
14	0	0%
15	0	0%
16	0	0%
Total	140

Really good stuff, thanks. I think the 11 seed is a decent spot b/c in chalk you get a 5 then a 4 and by then maybe some chaos has eliminated some of 1-3 top teams for an easier path. Or it could just be a random coincidence. Of course the ECAC reseeds so there’s a big difference Vs the basketball bracket approach.

There is way less variance in a game of basketball than a game of hockey. I honestly don’t think this is a very helpful comparison.

It's not a single game comparison. you have to win 4 games to win a 16-team tournament/bracket.

Also, in the ECAC tournament you've got to win a best of 3 in the second round. Best of 3 should significantly reduce variance and reduce upsets versus a single game round.

BearLover
There is way less variance in a game of basketball than a game of hockey. I honestly don’t think this is a very helpful comparison.

Depends on whether you run up against a hot goal tender.

Two months ago on this site the practice of using records of past performance to predict winners of future games was reviled. Probability Puritans must be deeply disappointed by the total dominance of past performance used to select and seed teams in the finals.

osorojo
Two months ago on this site the practice of using records of past performance to predict winners of future games was reviled. Probability Puritans must be deeply disappointed by the total dominance of past performance used to select and seed teams in the finals.

Your mom's past performance was reviled by all of the longshoremen who paid her the nickel.

French Rage

osorojo
Two months ago on this site the practice of using records of past performance to predict winners of future games was reviled. Probability Puritans must be deeply disappointed by the total dominance of past performance used to select and seed teams in the finals.

Your mom's past performance was reviled by all of the longshoremen who paid her the nickel.

Wow, Just wow.

F.R. - I trust you did not study logic or rhetoric [or college ice hockey] - and I DO hope you did not attend C.U.

Jeff Hopkins '82

French Rage

osorojo
Two months ago on this site the practice of using records of past performance to predict winners of future games was reviled. Probability Puritans must be deeply disappointed by the total dominance of past performance used to select and seed teams in the finals.

Your mom's past performance was reviled by all of the longshoremen who paid her the nickel.

Wow, Just wow.

I had to try to suppress a laugh. But I'm not sure that's something even Trotsky would have posted.

osorojo
F.R. - I trust you did not study logic or rhetoric [or college ice hockey] - and I DO hope you did not attend C.U.

Your mental diarrhea is being given all of the consideration it deserves. Lick balls.

French Rage

osorojo
F.R. - I trust you did not study logic or rhetoric [or college ice hockey] - and I DO hope you did not attend C.U.

Your mental diarrhea is being given all of the consideration it deserves. Lick balls.

osorojo is posting dumb stuff but this level of nastiness is completely uncalled for IMO

BearLover

French Rage

osorojo
F.R. - I trust you did not study logic or rhetoric [or college ice hockey] - and I DO hope you did not attend C.U.

Your mental diarrhea is being given all of the consideration it deserves. Lick balls.

osorojo is posting dumb stuff but this level of nastiness is completely uncalled for IMO

At least one moderator agrees.

scoop85

Jeff Hopkins '82

French Rage

osorojo
Two months ago on this site the practice of using records of past performance to predict winners of future games was reviled. Probability Puritans must be deeply disappointed by the total dominance of past performance used to select and seed teams in the finals.

Your mom's past performance was reviled by all of the longshoremen who paid her the nickel.

Wow, Just wow.

I had to try to suppress a laugh. But I'm not sure that's something even Trotsky would have posted.

While FR goes out of his way to be cruel to me for some reason or other, I defend him here. It was obviously an over-the-top joke, where any actual virulence is detoxified by the ridiculousness of the language. Y'all act like you never heard Chris Rock before, jaws all on the floor, like Redd Foxx just walked through the door.

I'd like to add that 2 bits will get your platoon a discount on the waterfront. I've heard.

I freaking love the bye week. Honestly.

RichH
I freaking love the bye week. Honestly.

Same here. 4/5 is the largest gap in the ECAC.

#49 St. Thomas holding #11 Michigan Tech scoreless in the second in Houghton. I assume if MTU doesn't win that one it will fuck them hard.

Jeff Hopkins '82

French Rage

osorojo
Two months ago on this site the practice of using records of past performance to predict winners of future games was reviled. Probability Puritans must be deeply disappointed by the total dominance of past performance used to select and seed teams in the finals.

Your mom's past performance was reviled by all of the longshoremen who paid her the nickel.

Wow, Just wow.

Wow indeed. I mean, they overpaid..

Nate Silver had Ms. Clinton at 70% in 2016. By far the lowest of any prognostication - most of which had her at around 97% - thanks to a logical fallacy that only Silver had right.

The farther we fall, the more likely all the conference champions will be above us!

Just eyeballing it, to qualify for the NCAAs I’d guess Cornell needs to either win their quarterfinals series 2-0 or win it 2-1 and then beat their semifinals opponent. Given the quarterfinals opponent is probably Clarkson and the semifinals opponent is probably Harvard, that’s going to be very difficult. The BU game and senior weekend failures loom large now as Cornell is really teetering on the edge.

It would help if ND were to lose to MSU tonight and better yet twice. But they could easily lose tonight, then win then lose again
Minn St hopefully loses to Mtech but doubtful before that

Alaska pretty much locked in I think

Vermont beating Merrimack would be nice

Maybe Omaha can lose to ND again

Northeastern losing again would also be good.

Then 2-0 next week might be enough Even with a loss.

We do have decent RPI leads over the teams below us other than Minn St.

The home/away weighting has me wishing we were on the road next weekend.

Which is dumb.

upprdeck
It would help if ND were to lose to MSU tonight and better yet twice. But they could easily lose tonight, then win then lose again
Minn St hopefully loses to Mtech but doubtful before that

Alaska pretty much locked in I think

Vermont beating Merrimack would be nice

Maybe Omaha can lose to ND again

Northeastern losing again would also be good.

Then 2-0 next week might be enough Even with a loss.

We do have decent RPI leads over the teams below us other than Minn St.

Alaska would be out if they lost tonight to Lindenwood (highly unlikely)

BearLover

upprdeck
It would help if ND were to lose to MSU tonight and better yet twice. But they could easily lose tonight, then win then lose again
Minn St hopefully loses to Mtech but doubtful before that

Alaska pretty much locked in I think

Vermont beating Merrimack would be nice

Maybe Omaha can lose to ND again

Northeastern losing again would also be good.

Then 2-0 next week might be enough Even with a loss.

We do have decent RPI leads over the teams below us other than Minn St.

Alaska would be out if they lost tonight to Lindenwood (highly unlikely)

I think they were briefly down 1-0 last night. Here's hoping.

Does a win over the weakest team always give you more in PWR than a loss to the strongest team?

www.playoffstatus.com

ECAC Tournament

        SF    F   Ch
Qpc    .99  .82  .51
Hvd    .96  .61  .30
Cor    .80  .32  .12
SLU    .61  .13  .04
Field  .64  .12  .03

Trotsky
www.playoffstatus.com

ECAC Tournament

        SF    F   Ch
Qpc    .99  .82  .51
Hvd    .96  .61  .30
Cor    .80  .32  .12
SLU    .61  .13  .04
Field  .64  .12  .03

lol wonderful

Trotsky
www.playoffstatus.com

ECAC Tournament

        SF    F   Ch
Qpc    .99  .82  .51
Hvd    .96  .61  .30
Cor    .80  .32  .12
SLU    .61  .13  .04
Field  .64  .12  .03

I was actually about to post this, as I’d checked it this morning for a laugh. Imagine creating a predictive model this bad, and then having people around the internet cite to your model in earnest. I wonder if the people publishing the model know it is complete garbage, or if they don’t care and simply don’t mind the total lack of academic rigor and accuracy?

BearLover

Trotsky
www.playoffstatus.com

ECAC Tournament

        SF    F   Ch
Qpc    .99  .82  .51
Hvd    .96  .61  .30
Cor    .80  .32  .12
SLU    .61  .13  .04
Field  .64  .12  .03

I was actually about to post this, as I’d checked it this morning for a laugh. Imagine creating a predictive model this bad, and then having people around the internet cite to your model in earnest. I wonder if the people publishing the model know it is complete garbage, or if they don’t care and simply don’t mind the total lack of academic rigor and accuracy?

I'm not sure there is such a thing as rigor in an instance like this. The sample size is never going to be large enough to test out what the percentages should have been.

I'll keep muttering this until the cows come home, but it's not predictive, no matter what anybody says. It extrapolates prior performance to future results, but that is only justifiably predictive if the conditions of the future match the conditions of the past. In the case of the ECAC tournament that is categorically false. If nothing else, the final two rounds are played at a neutral site where literally 0% of past results were generated.

I cite these numbers because (1) beats workin', and (2) it's fun for a variety of reasons we have touched on previously. There is no dependable predictive metric.

if there was a predictive model Vegas would be hurting. But we here would all be rich.

BearLover

Trotsky
www.playoffstatus.com

ECAC Tournament

        SF    F   Ch
Qpc    .99  .82  .51
Hvd    .96  .61  .30
Cor    .80  .32  .12
SLU    .61  .13  .04
Field  .64  .12  .03

I was actually about to post this, as I’d checked it this morning for a laugh. Imagine creating a predictive model this bad, and then having people around the internet cite to your model in earnest. I wonder if the people publishing the model know it is complete garbage, or if they don’t care and simply don’t mind the total lack of academic rigor and accuracy?

Isn't there a story about the engineering students who designed the suspension bridge refusing to walk over it because they knew how they designed it?

Swampy
Isn't there a story about the engineering students who designed the suspension bridge refusing to walk over it because they knew how they designed it?

No. That was capitalism.

Trotsky
I'll keep muttering this until the cows come home, but it's not predictive, no matter what anybody says. It extrapolates prior performance to future results, but that is only justifiably predictive if the conditions of the future match the conditions of the past. In the case of the ECAC tournament that is categorically false. If nothing else, the final two rounds are played at a neutral site where literally 0% of past results were generated.

I cite these numbers because (1) beats workin', and (2) it's fun for a variety of reasons we have touched on previously. There is no dependable predictive metric.

The model is intended to be predictive, even if some people recognize it as not doing a good job at being predictive. The model gives probabilities of future events. Therefore, in layman’s terms, it is intended to be predictive.

The issue isn’t really the neutral site. The issue is that the model relies on KRACH, which is meant to rank teams based on their record, not on how good they actually are. (I assume it relies on KRACH like the CHN predictor does, but I don’t know that for sure.) KRACH has no reason to (and does not) take into account random chance. It has no reason to (and does not) account for the fact that a team may be better or worse than its record. KRACH doesn’t bother with this because records are all that matters for ranking teams based on past performance. But a predictive model needs to account for the many random and flukish results over a hockey season. One way for a predictive model to do this would be to consider statistics beyond record such as possession numbers. Another simpler way would be take the rankings of teams straight from KRACH, but rather than weighting the teams by KRACH for purposes of comparing them, instead use historical data to determine how often a team of KRACH ranking X beats a team of KRACH ranking Y.

if you were making a model of 2 teams that play the same shedule.

1 team plays 20 games and wins 10 and loses next 10
1 team plays 20 games and loses 10 and wins the next 10

some models would say the teams are equal and some would not.

the NCAA selection committee for some sports say it doesn't matter since they dropped the last x game criteria.

but if you were betting in Vegas I think would care..

The model also doesn't take into account how bad the refs are.

ursusminor
The model also doesn't take into account how bad the refs are.

It also doesn't take into account the entertainment value for those of us watching BL's response. Past performance in some cases does predict future outcomes.

Been waiting 10 years for someone to work with me on a better model. I ask every year. Every time someone makes the slightest gesture toward that end, they disappear again in a day or two. Meanwhile, just complaints. ... I don't see the same amount of righteous outrage directed at polls. Talk about something truly meaningless.

Please by all means - let's write a better model.

Better models already exist, eg Massey ratings: [masseyratings.com]

His method is proprietary but the results seem reasonable.

CU77
Better models already exist, eg Massey ratings: [masseyratings.com]

His method is proprietary but the results seem reasonable.

Where do you see predictive models there?

Also Cornell is 20th there - be careful what you wish for.

adamw

CU77
Better models already exist, eg Massey ratings: [masseyratings.com]

His method is proprietary but the results seem reasonable.

Where do you see predictive models there?

Also Cornell is 20th there - be careful what you wish for.

I do see a page on there where you can "sim" a best-of-3 series ... so I simmed one between Cornell and Quinnipiac - and Cornell won 13 out of 15 best-of-3 series between the teams. So ... I'd say he's got some work to do.

adamw

adamw

CU77
Better models already exist, eg Massey ratings: [masseyratings.com]

His method is proprietary but the results seem reasonable.

Where do you see predictive models there?

Also Cornell is 20th there - be careful what you wish for.

I do see a page on there where you can "sim" a best-of-3 series ... so I simmed one between Cornell and Quinnipiac - and Cornell won 13 out of 15 best-of-3 series between the teams. So ... I'd say he's got some work to do.

sounds right to me

CU77
Better models already exist, eg Massey ratings: [masseyratings.com]

His method is proprietary but the results seem reasonable.

The probabilities on this page seem eminently reasonable to me.

adamw
Been waiting 10 years for someone to work with me on a better model. I ask every year. Every time someone makes the slightest gesture toward that end, they disappear again in a day or two. Meanwhile, just complaints. ... I don't see the same amount of righteous outrage directed at polls. Talk about something truly meaningless.

Please by all means - let's write a better model.

I think we all recognize that it takes some level of mathematical ability to be able to improve upon the current model in a way that is theoretically defensible, and not based on some kind of fudgery. My only suggestion would be to treat each team’s KRACH value as a random variable (likely non-Gaussian) with the current KRACH representing the median of that variable. Then you could probably run a smaller Monte Carlo simulation and get better numbers.

BearLover

CU77
Better models already exist, eg Massey ratings: [masseyratings.com]

His method is proprietary but the results seem reasonable.

The probabilities on this page seem eminently reasonable to me.

Extrapolating those probabilities out over best of 3 likely gives you nearly the same probabilities as the playoffstatus page.

KRACH probabilities can be improved by adding to every team's record some number of ties against a fictitious "average" team; Ken Butler's original method added one such game, but more gives a better distribution of predicted results.

I've heard all of these suggestions before. On this forum, and elsewhere. I've encouraged anyone with ideas to reach out to me and please feel free to improve the algorithm. A couple people in the last 10 years have gestured to take me up on that offer, only to fade away pretty quickly.

So again - if anyone has an idea - please feel free and come to me. The door is open. I only went to that school in the ghetto part of Ithaca, not your fancy Ivy League place filled with math and physics PhDs. So instead of complaints, please come to me with a better system. Not a vague sentence or two -- the actual math. That I can code.

Well, I would start with KRACH, add some number x of ties against a fictitious team to each team's record, compute KRACH ratings at the end of a regular season for various values of x, then compute predicted records of each team in an all-against-all set of games, and then compare the distribution of the computed winning percentages against the actual distribution of winning percentages in the games actually played, and then adjust x to get the best match.

But this would be a LOT of work for a marginal improvement of predictions that no one cares about anyway.

Better: a ton of work has gone into making such predictions for basketball. Find out the the state of the art there, and adapt for hockey.

Yes, a better model would be a ton of work. What if Wayne Gretzky, Jr was on a team, but injured for the first 20 games of the season and the team went .500. Since he got back, he’s averaging 2.5 points per game and the team is undefeated. Where would your “better” model rank them headed into the playoffs? Maybe the team hired a better strength and conditioning coach halfway through the season? What if your star players also care about academics and the team always tanks during exam weeks….and there are midterms the week of the quarterfinals?

Models will never capture the richness and complexity of life. Once you accept that fact, you can paradoxically enjoy looking at models again, because you are content in your certainty that it only tells a part of the story and just agree with yourself to leave it at that.

the model might capture the change in stuff/roster but if they really were playing better then they should be scoring more or winning more and it should see that?

The next level of modeling is figuring out the hot teams and how long they stay hot vs play at the level they should be at.

Robb
Yes, a better model would be a ton of work. What if Wayne Gretzky, Jr was on a team, but injured for the first 20 games of the season and the team went .500. Since he got back, he’s averaging 2.5 points per game and the team is undefeated. Where would your “better” model rank them headed into the playoffs? Maybe the team hired a better strength and conditioning coach halfway through the season? What if your star players also care about academics and the team always tanks during exam weeks….and there are midterms the week of the quarterfinals?

Models will never capture the richness and complexity of life. Once you accept that fact, you can paradoxically enjoy looking at models again, because you are content in your certainty that it only tells a part of the story and just agree with yourself to leave it at that.

My problem with the models at issue is that they don’t even accurately tell the part of the story they claim to tell. Yes, they don’t account for injuries, or changes in personnel, or exams. But they don’t claim to do those things. Rather, these models claim to predict future likelihoods based entirely on a team’s past success. But they don’t even do that properly, because the probabilities the models spit out do not accurately reflect the actual probability that one should get from looking purely at past success.

As to adamw’s point—I want to preface this by saying I love CHN and go there 30 times a day. There are many awesome features of the site. But when it comes to the model, my view is that if it isn’t doing what it claims to do (provide realistic probabilities of future results), and we know it does not do this, then it would be better to not publish the model at all. Otherwise, publishing it just misleads people.

I don’t have the training necessary to help improve the model. I hope someone does who is willing to assist. But until that happens, better not to publish the model at all, IMO.

KRACH could be adjusted to do better, as I outlined above. KRACH is a very simple model.

The first question would be on what context are the models not working? As a whole or for a team?

Look at the probablility matrix thing

Cornell is like 60% to get in.. the model could have us at 90% of the time winning in a sweep and losing to Harvard. That would get us in. Maybe

But it also has to account that there could be 3 upsets someplace else and we wouldnt get in.

The first thing you have to do is eyeball what you thing will happen and then compare it to the model. Then you can decide if its wrong. You also have to do that across 100s of games.

And if a model is successful more than X% that makes it a good model. What do we consider that X to be

Look at horse racing. Favs win like less than 30% of the time.. So that means across millions of peoples models they can only get the fav right 30% and that includes tons of races where there really is no toss up choice.. What is the stat where races have favs that are not clear favs like less than 10% or lower? And thats in races with clear things to measure and known past results and 1000s of histories to compare to and training notes to look at.

what if i told you that all of what has come before and all that comes after has been foretold. *turns chair around* *flips hat around* and that one man gave his life...

I'm sure we'll solve this problem here.

Nature will always maintain her rights, and prevail in the end over any abstract reasoning whatsoever. Though we should conclude, for instance, as in the foregoing section, that, in all reasonings from experience, there is a step taken by the mind, which is not supported by any argument or process of the understanding; there is no danger, that these reasonings, on which almost all knowledge depends, will ever be affected by such a discovery.

Shorter Hume: We make shit up and try our best.

Shorter Shorter Hume: Drop the puck.

CU77
KRACH could be adjusted to do better, as I outlined above. KRACH is a very simple model.

The door is open ...

BearLover
As to adamw’s point—I want to preface this by saying I love CHN and go there 30 times a day. There are many awesome features of the site. But when it comes to the model, my view is that if it isn’t doing what it claims to do (provide realistic probabilities of future results), and we know it does not do this, then it would be better to not publish the model at all. Otherwise, publishing it just misleads people.

I don’t have the training necessary to help improve the model. I hope someone does who is willing to assist. But until that happens, better not to publish the model at all, IMO.

I generally agree with this philosophy. Which is why I cannot stand publishing "Bracketology" articles that tell you what the Brackets would be like if the season ended today - because they don't (especially when said articles don't acknowledge that point). It's also why I can't stand polls.

However, I disagree that the CHN Probability Krach-based Matrix thingy is doing that.

The Matrix is intended to play out the schedule and see who finishes where. Automatically. It's like "You Are the Committee" but done for you, because you can't eyeball and figure it out, and because it's a pain to manually input results - and also then proceed forward with brackets based on the first set of results you put in.

I don't think anyone in their right mind really believes we're predicting the future. It's meant to extrapolate probable Pairwise scenarios - not "predict winners" per se.

adamw

CU77
KRACH could be adjusted to do better, as I outlined above. KRACH is a very simple model.

The door is open ...

I don't care enough to do it myself. In any case Massey probabilities are likely better, as they use more input (scores, locations, times).

CU77

adamw

CU77
KRACH could be adjusted to do better, as I outlined above. KRACH is a very simple model.

The door is open ...

I don't care enough to do it myself. In any case Massey probabilities are likely better, as they use more input (scores, locations, times).

This comic says it well: [www.gocomics.com]

David Harding

CU77

adamw

CU77
KRACH could be adjusted to do better, as I outlined above. KRACH is a very simple model.

The door is open ...

I don't care enough to do it myself. In any case Massey probabilities are likely better, as they use more input (scores, locations, times).

This comic says it well: [www.gocomics.com]

Cloudy with a high wind warning.

CU77

adamw

CU77
KRACH could be adjusted to do better, as I outlined above. KRACH is a very simple model.

The door is open ...

I don't care enough to do it myself.

Congrats - you're in a growing club.

CU77
In any case Massey probabilities are likely better, as they use more input (scores, locations, times).

Times? We do have a home/away version of KRACH. It's not much different. I should use it though.

Scores have always been an extremely dubious and debatable thing to use in hockey. I go back and forth on it.

few sports have things like goals scored in the end of hockey games.

What other sport has a mechanism where teams can totally change the way they play to try and score goals at the end of games, that can lead to the other team scoring like ENGs in hockey.. even Soccer where teams throw everything forward seldom does it mean the other team scores.. But in hockey a 3-1 game can turn into a 4-3 game that had little to do with how the 95% of the game was played.

There are metrics such as Corsi and Fenwick. But those annoy me because "what if we measured all shots and not just shots on goal" isn't a big enough breakthrough to deserve slapping someone's name on it.

adamw

CU77
In any case Massey probabilities are likely better, as they use more input (scores, locations, times).

Times? We do have a home/away version of KRACH. It's not much different. I should use it though.

Scores have always been an extremely dubious and debatable thing to use in hockey. I go back and forth on it.

Times = early season vs late season results.

Scores are data, completely ignoring that data is almost surely subobtimal.

The fancy-stats people in pro hockey likely have good ideas about rankings.

As it stands now, Massey's match-up tool can be used to predict win probabilities for any possible game. I doubt that it's possible to significantly improve on those.

upprdeck
few sports have things like goals scored in the end of hockey games.

What other sport has a mechanism where teams can totally change the way they play to try and score goals at the end of games, that can lead to the other team scoring like ENGs in hockey.. even Soccer where teams throw everything forward seldom does it mean the other team scores.. But in hockey a 3-1 game can turn into a 4-3 game that had little to do with how the 95% of the game was played.

Basketball is like that when the two-minute foul line parade starts for the team that played best for the first 38 minutes. Unwatchable now for me.

Al DeFlorio

upprdeck
few sports have things like goals scored in the end of hockey games.

What other sport has a mechanism where teams can totally change the way they play to try and score goals at the end of games, that can lead to the other team scoring like ENGs in hockey.. even Soccer where teams throw everything forward seldom does it mean the other team scores.. But in hockey a 3-1 game can turn into a 4-3 game that had little to do with how the 95% of the game was played.

Basketball is like that when the two-minute foul line parade starts for the team that played best for the first 38 minutes. Unwatchable now for me.

That plus the incessant time outs.

CU77
The fancy-stats people in pro hockey likely have good ideas about rankings.

I don't know what makes you think that. I've seen NHL rankings -- I work for a site that has those too -- and they essentially the same as KRACH.

adamw

CU77
The fancy-stats people in pro hockey likely have good ideas about rankings.

I don't know what makes you think that. I've seen NHL rankings -- I work for a site that has those too -- and they essentially the same as KRACH.

I just assumed they would do something fancy because that's what they do ... also I would want to dig into what "essentially" means; any kind of ranking based of results of paired comparisons likely has some foundation in Bradley-Terry (aka KRACH), but adjusting based on scores, home v away v neutral, and time since game played can all be done. The issue is deciding, for example, by how much a game played a month ago should be discounted from a game played yesterday. This can only be done on an empirical basis, and it's a ton of work to get and analyze the data needed to do that.

upprdeck
few sports have things like goals scored in the end of hockey games.

What other sport has a mechanism where teams can totally change the way they play to try and score goals at the end of games, that can lead to the other team scoring like ENGs in hockey.. even Soccer where teams throw everything forward seldom does it mean the other team scores.. But in hockey a 3-1 game can turn into a 4-3 game that had little to do with how the 95% of the game was played.

When basketball turns into “foulpalooza” and it’s just 20 minutes of exciting foul shooting.

Edit: Al beat me to it.

You could fix basketball by giving extra foul shots for multiple fouls within a certain interval. Also, basketball should get one time out per half, period.

Trotsky
You could fix basketball by giving extra foul shots for multiple fouls within a certain interval. Also, basketball should get one time out per half, period.

The TV networks rely on the timeouts for commercial breaks, especially in the NCAA Tournament when the timeouts seem to last forever.

But if you are going to use scores, changing from a 75-65 game to an 82-77 game is vastly different than being up 2-0 and the goalie us pulled and winning 4-0..

The best proposal for fixing basketball I've seen is to play to a score. Play 35 minutes, then if the score is Team 1 X, Team 2 Y (assume X > Y), first team to X+10 wins.

Dafatone
The best proposal for fixing basketball I've seen is to play to a score. Play 35 minutes, then if the score is Team 1 X, Team 2 Y (assume X > Y), first team to X+10 wins.

that's how they play the ASG

Trotsky
You could fix basketball by giving extra foul shots for multiple fouls within a certain interval. Also, basketball should get one time out per half, period.

Just give the fouled team the option of taking the ball again. The fouls will stop.