Normalized ACC Standings

[uh_no]

There are lots of complaints about unbalanced schedules in leagues, and how that affects standings. These bother me, not because of the unbalanced schedules, but because the solution is so simple. The concern is that a team who doubles up against cupcakes may have a few free wins someone with two games against UNC and UL may take some extra losses.

The solution, in my mind, is obvious. Simply divide the record vs any given team by the number of games against that team.

You won twice against BC? still only counts in your record as 1-0. Lost twice to duke? 0-1. Split with UNC? .5-.5

There is no benefit or injury from multiple games against a team, but yet every game is still reflected in the record. Does this mean that a game against a team you play twice only counts for half a game in the record? Yep. But so does any game that's part of a multi-game series. Why don't we use this? Because it would confuse people and require the league to do something intelligent. I also have a very low opinion of your average sports fan, and am skeptical of the population's ability to understand a system other than "every game counts the same"...basically "why can't we have nice things?"

Anyway, so lets look at what the standings would be right now. They will be somewhat similar, just because the schedule will be inevitably backloaded, but what is interesting right now is how much ground teams can make up or lose, which is more limited in duplicate games.

First up is just a list of teams and their records in home/away games. This is simply convenience for me to calculate the adjusted records
http://www.theacc.com/news/acc-releases-men-conference-basketball-matchups-02-12-2016
BC: cuse 1-1 ND 0-1 WFU 0-2 VT 0-1
Clemson: FSU 0-1 GT 1-1 WF 2-0 VT 0-2
Duke: UNC 1-0 WF 2-0 FSU 0-1 UM 1-0
FSU: UM 1-0 clemson 1-0 Duke 1-0 ND 1-1
GT: clemson 1-1 ND 1-0 NCSU 1-1 cuse 1-0
UL: pitt 2-0 uva 0-2 cuse 1-0 nd 0-1
UM: fsu 0-1 vt 1-0 duke 0-1 ncsu 2-0
unc: duke 0-1 ncsu 2-0 uva 1-0 pitt 1-0
ncsu: unc 0-2 wfu 0-2 um 0-2 gt 1-1
nd: bc 1-0 gt 0-1 fsu 1-1 ul 1-0
pitt: ul 0-2 cuse 1-1 unc 0-1 uva 1-0
cuse: bc 1-1 pitt 1-1 gt 0-1 ul 0-1
uva: ul 2-0 vt 1-1 unc 0-1 pitt 0-1
vt: um 0-1 uva 1-1 clemson 2-0 bc 1-0
wf: duke 0-2 ncsu 2-0 clemson 0-2 bc 2-0


Here are the adjusted standings. The biggest changes are
1. The stratification of the 10-5 block, where we are on the losing side of that since our record is nearly perfect in home/away games.
2. NCSU jumping up the standings. a couple of spots. they've lost nearly all their double games.
3. with only a game and a half left on the schedule, Duke is eliminated from regular season contention. ND and UL are the only two with a chance to unseat the holes.


team
record
adjusted


UNC
12-3
10-2.5


UL
10-5
8.5-3.5


ND
10-5
8.5-4


Duke
10-5
8-4.5


FSU
10-5
8-4.5


cuse +1
9-7
8-5


UM -1
9-6
7.5-5


uva +1
8-7
6.5-5.5


vt -1
8-7
6-6


GT
7-8
5-7


wf
7-9
5-7


NCSU +2
4-12
3.5-8.5


pitt -1
4-11
3-9


Clemson -1
4-11
2.5-9


BC
2-13
1.5-11.5



I'll try to keep this at least somewhat updated the rest of the way. Also considered making standings that ignore big east teams...a true balanced ACC title :)

[rasputin]

There are lots of complaints about unbalanced schedules in leagues, and how that affects standings. These bother me, not because of the unbalanced schedules, but because the solution is so simple. The concern is that a team who doubles up against cupcakes may have a few free wins someone with two games against UNC and UL may take some extra losses.

The solution, in my mind, is obvious. Simply divide the record vs any given team by the number of games against that team.

You won twice against BC? still only counts in your record as 1-0. Lost twice to duke? 0-1. Split with UNC? .5-.5

There is no benefit or injury from multiple games against a team, but yet every game is still reflected in the record. Does this mean that a game against a team you play twice only counts for half a game in the record? Yep. But so does any game that's part of a multi-game series. Why don't we use this? Because it would confuse people and require the league to do something intelligent. I also have a very low opinion of your average sports fan, and am skeptical of the population's ability to understand a system other than "every game counts the same"...basically "why can't we have nice things?"

Anyway, so lets look at what the standings would be right now. They will be somewhat similar, just because the schedule will be inevitably backloaded, but what is interesting right now is how much ground teams can make up or lose, which is more limited in duplicate games.

First up is just a list of teams and their records in home/away games. This is simply convenience for me to calculate the adjusted records
http://www.theacc.com/news/acc-releases-men-conference-basketball-matchups-02-12-2016
BC: cuse 1-1 ND 0-1 WFU 0-2 VT 0-1
Clemson: FSU 0-1 GT 1-1 WF 2-0 VT 0-2
Duke: UNC 1-0 WF 2-0 FSU 0-1 UM 1-0
FSU: UM 1-0 clemson 1-0 Duke 1-0 ND 1-1
GT: clemson 1-1 ND 1-0 NCSU 1-1 cuse 1-0
UL: pitt 2-0 uva 0-2 cuse 1-0 nd 0-1
UM: fsu 0-1 vt 1-0 duke 0-1 ncsu 2-0
unc: duke 0-1 ncsu 2-0 uva 1-0 pitt 1-0
ncsu: unc 0-2 wfu 0-2 um 0-2 gt 1-1
nd: bc 1-0 gt 0-1 fsu 1-1 ul 1-0
pitt: ul 0-2 cuse 1-1 unc 0-1 uva 1-0
cuse: bc 1-1 pitt 1-1 gt 0-1 ul 0-1
uva: ul 2-0 vt 1-1 unc 0-1 pitt 0-1
vt: um 0-1 uva 1-1 clemson 2-0 bc 1-0
wf: duke 0-2 ncsu 2-0 clemson 0-2 bc 2-0


Here are the adjusted standings. The biggest changes are
1. The stratification of the 10-5 block, where we are on the losing side of that since our record is nearly perfect in home/away games.
2. NCSU jumping up the standings. a couple of spots. they've lost nearly all their double games.
3. with only a game and a half left on the schedule, Duke is eliminated from regular season contention. ND and UL are the only two with a chance to unseat the holes.


team
record
adjusted


UNC
12-3
10-2.5


UL
10-5
8.5-3.5


ND
10-5
8.5-4


Duke
10-5
8-4.5


FSU
10-5
8-4.5


cuse +1
9-7
8-5


UM -1
9-6
7.5-5


uva +1
8-7
6.5-5.5


vt -1
8-7
6-6


GT
7-8
5-7


wf
7-9
5-7


NCSU +2
4-12
3.5-8.5


pitt -1
4-11
3-9


Clemson -1
4-11
2.5-9


BC
2-13
1.5-11.5



I'll try to keep this at least somewhat updated the rest of the way. Also considered making standings that ignore big east teams...a true balanced ACC title :)

This just creates other anomalies. We only had one shot against Louisville, in their gym, and came up short. It doesn't make sense to me to count that equally with the scenario where we lost to Louisville both home and away.

[CDu]

But this isn't really balanced. In fact, it is LESS balanced, because it places double the value on the games that don't involve home-and-home. So the team that plays the tougher set of "only away" opponents is even more disadvantaged. This is why we see that UNC is even further ahead in the adjusted standings than in the standard standings.

[uh_no]

This just creates other anomalies. We only had one shot against Louisville, in their gym, and came up short. It doesn't make sense to me to count that equally with the scenario where we lost to Louisville both home and away.

then maybe looking at won/loss at all is not your way to go.

For the record, i'm one for ignoring wins and losses altogether in favor of efficiency, but that doesn't seem to be popular. If we're going to insist on using a silly metric to evaluate teams play such as win and loss, we might as well make some basic adjustment for multiple opponents.

[uh_no]

So the team that plays the tougher set of "only away" opponents is even more disadvantaged. This is why we see that UNC is even further ahead in the adjusted standings than in the standard standings.

This is true regardless of how you adjust. It's true even in a balanced schedule that if you only have tough teams on the road, then you may lose to all of them, but if you only have easy teams, you may beat all of them. If we're going to try to adjust for home/away as well, we can't use win/loss as our base metric.

[flyingdutchdevil]

There are lots of complaints about unbalanced schedules in leagues, and how that affects standings. These bother me, not because of the unbalanced schedules, but because the solution is so simple. The concern is that a team who doubles up against cupcakes may have a few free wins someone with two games against UNC and UL may take some extra losses.

The solution, in my mind, is obvious. Simply divide the record vs any given team by the number of games against that team.

You won twice against BC? still only counts in your record as 1-0. Lost twice to duke? 0-1. Split with UNC? .5-.5

There is no benefit or injury from multiple games against a team, but yet every game is still reflected in the record. Does this mean that a game against a team you play twice only counts for half a game in the record? Yep. But so does any game that's part of a multi-game series. Why don't we use this? Because it would confuse people and require the league to do something intelligent. I also have a very low opinion of your average sports fan, and am skeptical of the population's ability to understand a system other than "every game counts the same"...basically "why can't we have nice things?"

Anyway, so lets look at what the standings would be right now. They will be somewhat similar, just because the schedule will be inevitably backloaded, but what is interesting right now is how much ground teams can make up or lose, which is more limited in duplicate games.

First up is just a list of teams and their records in home/away games. This is simply convenience for me to calculate the adjusted records
http://www.theacc.com/news/acc-releases-men-conference-basketball-matchups-02-12-2016
BC: cuse 1-1 ND 0-1 WFU 0-2 VT 0-1
Clemson: FSU 0-1 GT 1-1 WF 2-0 VT 0-2
Duke: UNC 1-0 WF 2-0 FSU 0-1 UM 1-0
FSU: UM 1-0 clemson 1-0 Duke 1-0 ND 1-1
GT: clemson 1-1 ND 1-0 NCSU 1-1 cuse 1-0
UL: pitt 2-0 uva 0-2 cuse 1-0 nd 0-1
UM: fsu 0-1 vt 1-0 duke 0-1 ncsu 2-0
unc: duke 0-1 ncsu 2-0 uva 1-0 pitt 1-0
ncsu: unc 0-2 wfu 0-2 um 0-2 gt 1-1
nd: bc 1-0 gt 0-1 fsu 1-1 ul 1-0
pitt: ul 0-2 cuse 1-1 unc 0-1 uva 1-0
cuse: bc 1-1 pitt 1-1 gt 0-1 ul 0-1
uva: ul 2-0 vt 1-1 unc 0-1 pitt 0-1
vt: um 0-1 uva 1-1 clemson 2-0 bc 1-0
wf: duke 0-2 ncsu 2-0 clemson 0-2 bc 2-0


Here are the adjusted standings. The biggest changes are
1. The stratification of the 10-5 block, where we are on the losing side of that since our record is nearly perfect in home/away games.
2. NCSU jumping up the standings. a couple of spots. they've lost nearly all their double games.
3. with only a game and a half left on the schedule, Duke is eliminated from regular season contention. ND and UL are the only two with a chance to unseat the holes.


team
record
adjusted


UNC
12-3
10-2.5


UL
10-5
8.5-3.5


ND
10-5
8.5-4


Duke
10-5
8-4.5


FSU
10-5
8-4.5


cuse +1
9-7
8-5


UM -1
9-6
7.5-5


uva +1
8-7
6.5-5.5


vt -1
8-7
6-6


GT
7-8
5-7


wf
7-9
5-7


NCSU +2
4-12
3.5-8.5


pitt -1
4-11
3-9


Clemson -1
4-11
2.5-9


BC
2-13
1.5-11.5



I'll try to keep this at least somewhat updated the rest of the way. Also considered making standings that ignore big east teams...a true balanced ACC title :)

Unlike the other posters, let me start by saying, excellent work! You clearly spent time on this.

And it's really interesting. I think someone can poke holes into the analysis no matter what way you go (wins/losses, efficiency, spread, multiplier on away games, etc).

Personally, with unbalanced schedules, you give the regular season to whoever won it and put more equity into the tournament.

And right now, UNC is playing the best basketball in the ACC. So good for them.

[CDu]

This is true regardless of how you adjust. It's true even in a balanced schedule that if you only have tough teams on the road, then you may lose to all of them, but if you only have easy teams, you may beat all of them. If we're going to try to adjust for home/away as well, we can't use win/loss as our base metric.

Yes, but your way is only accentuating the problem, which is my point. It isn't actually a better solution, because it further emphasizes the games against teams you only play once.

There isn't a simple solution that is good. It would require something a bit more like RPI to properly adjust, where you take into account your opponents' conference record and your opponents' opponents' conference record, as well as some sort of weighting of home vs road.

I'm not suggesting that we should do that either. I just don't think the simple adjustment solution presented in this thread is better than what we currently have. In fact, I think it is worse.

[BandAlum83]

But this isn't really balanced. In fact, it is LESS balanced, because it places double the value on the games that don't involve home-and-home. So the team that plays the tougher set of "only away" opponents is even more disadvantaged. This is why we see that UNC is even further ahead in the adjusted standings than in the standard standings.

What is for the teams that only get one game, a win counts as .75 if won on the road, but only .25 if won at home. Conversely, a loss at home adds .75 to your loss total, but a loss away only counts .25.

So in essence, .5 is a standard vaule for a game, but it is adjusted for home/away.

[Newton_14]

Yes, but your way is only accentuating the problem, which is my point. It isn't actually a better solution, because it further emphasizes the games against teams you only play once.

There isn't a simple solution that is good. It would require something a bit more like RPI to properly adjust, where you take into account your opponents' conference record and your opponents' opponents' conference record, as well as some sort of weighting of home vs road.

I'm not suggesting that we should do that either. I just don't think the simple adjustment solution presented in this thread is better than what we currently have. In fact, I think it is worse.

CDu, this is right up your alley friend. (Sagegrouse too for that matter). So I'm challenging you to come up with a fair measure to determine the true performance based ranking of the ACC teams currently. Not because I think you can't do it or to want you to fail, but quite the opposite. There is a way to determine fair standings and I believe you my friend can come up with it

Are you game or what?

[CDu]

CDu, this is right up your alley friend. (Sagegrouse too for that matter). So I'm challenging you to come up with a fair measure to determine the true performance based ranking of the ACC teams currently. Not because I think you can't do it or to want you to fail, but quite the opposite. There is a way to determine fair standings and I believe you my friend can come up with it

Are you game or what?

I always like a challenge. :)

I'm happy to give it some thought. Given time constraints, it will need to be simple. But that's probably preferable anyway as anything too complex loses credibility as "black boxy".

[Ballboy1998]

Thanks for putting this together. It is an interesting thought experiment if nothing else. I think this would be a perfect resolution for imbalanced schedules where all games were played at neutral sites, but as others have noted, it doesn't really account for home/away imbalance, though I'm not able to suggest an easy fix for that.

The conference standings may have changed this, but at one point earlier in the year, someone noted that if you looked at the top 8 teams in the conference, most played 4 or 5 road games against the other 7, while Duke played all 7 and unc played 2.

It is what it is, so not worth dwelling on, but I do miss the full round-robin.

[uh_no]

I always like a challenge. :)

I'm happy to give it some thought. Given time constraints, it will need to be simple. But that's probably preferable anyway as anything too complex loses credibility as "black boxy".

I have an idea:

reigning champion: UNC

12/31: GT beats UNC....new champion GT
1/4 duke beats GT
1/10: FSU beats duke
1/25: GT beats FSU
2/1: cleanson beats GT
2/5: FSU beats clemson
2/11: ND beats FSU


CURRENT CHAMPION: ND.

You can't argue with that logic.

[Kedsy]

This is true regardless of how you adjust. It's true even in a balanced schedule that if you only have tough teams on the road, then you may lose to all of them, but if you only have easy teams, you may beat all of them. If we're going to try to adjust for home/away as well, we can't use win/loss as our base metric.

Perhaps I don't understand what you're saying, but in a balanced schedule wouldn't you play everyone both home and away?

[rsvman]

I have an idea:

reigning champion: UNC

12/31: GT beats UNC...new champion GT
1/4 duke beats GT
1/10: FSU beats duke
1/25: GT beats FSU
2/1: cleanson beats GT
2/5: FSU beats clemson
2/11: ND beats FSU


CURRENT CHAMPION: ND.

You can't argue with that logic.

Awesome. It's the Walton metric.

[Mtn.Devil.91.92.01.10.15]

I have come up with the normalized rankings that reflect fairness, competition, and common sense:

Duke #1
Other ACC Teams #2 - #14
UNC #15

I could bore you all with my methodology, but I would rather not. PM me if you'd like details

[Kishiznit]

I get the 16th team (most likely WVA) and split conference into 2 divisions. Within each division, a true home and away schedule. Divisions rotate every year and complete visibility 5 years in advance of where you will be. Divisional winners play for championship game in a 1 game championship; replacing the tourney. This makes regular season more important and eliminates burden of extended tourney before the big dance. You have a true conference champ, and many years, two #1 seeds.

[UrinalCake]

I'm not a math person, but I don't understand the reasoning behind this metric at all. Why would you normalize playing a team twice and have it count as if you only played them once? The whole point is that some teams play hard opponents twice while other teams only have to play those hard opponents once. Dividing by the number of times you play the team will effectively remove that difference.

Is there some way to perform an RPI or KenPom calculation using only conference games? Those metrics account for strength of schedule, so that could boil down a team's value to a single number. Or you could look at just the SOS ranking and figure out who really had the hardest schedules. I believe the metrics account for road vs. home difference, though that is a really hard thing to quantify. Playing on the road vs. BC is probably not that big a deal, whereas playing in Cameron is a huge deal.

I really believe that if we played UNC's schedule while they played ours, we would be tied in the standings at the very least. But I have no proof. I need numbers!

[Wheat/"/"/"]

All the courts at every ACC venue are the same size and the baskets are the same height...the Hoosier effect we'll call it. Home teams have a slight advantage, it's generally accepted, but the game still has to be played well to win and in my book, things are pretty even no matter where you play once that ball goes up.

Play whoever you're told to by the league and put a team on the floor, then look at the scoreboard at the end...add a check mark to the win or loss column then move on to the next one.

At the end of the regular season, the team that wins the most games is your regular season champion. The league has to keep it that simple, there is never going to be a way with a conference as large as the ACC to have a perfectly balanced schedule, or to see the future as to who is going to be strong or weak when making up the schedules.

This year, in hindsight, yes, UNC might have had the good fortune of a slight "advantage" when their home games happened to work out against some of their stronger opponents, but they still had to win those games against some really good teams. Nobody rolled over. I've seen other teams have scheduling advantages over the years, too. It's all noise.

There was no conspiracy in the making of that UNC schedule this season, didn't happen, and it seems unbecoming to keep harping on it, to me.

It's not over, yet. UNC may look like the best team in the conference now, but they have to close it out or they won't be. Again, it's that simple.

[UrinalCake]

So I found a site that provides in-conference SOS rankings. They are the rankings to-date and thus do not include the remaining games on the schedule. Here are the ACC teams with their rank among all teams in parenthesis.

Virginia (1)
Louisville (5)
Clemson (7)
Duke (9)
FSU (15)
VA Tech (16)
ND (17)
Pitt (18)
UNC (19)
Miami (20)
Syracuse (22)
Wake (24)
BC (25)
GT (26)
NC State (30)

link (https://www.teamrankings.com/ncaa-basketball/ranking/in-conference-sos-by-other)

Not surprising that all of the ACC teams are in the top 30 nationally, since the conference is so brutal.

Wheat if you are actually arguing that home team advantage does not play a factor in college basketball, I'm not really sure what to tell you. There is a plenty of evidence this year that playing at home makes an enormous difference. Claiming that playing at home doesn't matter is something a team would say when they're the ones playing at home.

And if you consider games against the top teams in the conference (UNC, Duke, Louisville, FSU, ND) who have separated themselves, UNC plays four games at home and one on the road (@Duke, which they lost). Meanwhile Duke plays two at home and four on the road. If you were to add UVA into the conversation, UNC gets them home and away while Duke only gets them away.

[BandAlum83]

All the courts at every ACC venue are the same size and the baskets are the same height...the Hoosier effect we'll call it. Home teams have a slight advantage, it's generally accepted, but the game still has to be played well to win and in my book, things are pretty even no matter where you play once that ball goes up.

Play whoever you're told to by the league and put a team on the floor, then look at the scoreboard at the end...add a check mark to the win or loss column then move on to the next one.

At the end of the regular season, the team that wins the most games is your regular season champion. The league has to keep it that simple, there is never going to be a way with a conference as large as the ACC to have a perfectly balanced schedule, or to see the future as to who is going to be strong or weak when making up the schedules.

This year, in hindsight, yes, UNC might have had the good fortune of a slight "advantage" when their home games happened to work out against some of their stronger opponents, but they still had to win those games against some really good teams. Nobody rolled over. I've seen other teams have scheduling advantages over the years, too. It's all noise.

There was no conspiracy in the making of that UNC schedule this season, didn't happen, and it seems unbecoming to keep harping on it, to me.

It's not over, yet. UNC may look like the best team in the conference now, but they have to close it out or they won't be. Again, it's that simple.

I don't believe anyone suggested a conspiracy in place.

[BandAlum83]

So I found a site that provides in-conference SOS rankings. They are the rankings to-date and thus do not include the remaining games on the schedule. Here are the ACC teams with their rank among all teams in parenthesis.

Virginia (1)
Louisville (5)
Clemson (7)
Duke (9)
FSU (15)
VA Tech (16)
ND (17)
Pitt (18)
UNC (19)
Miami (20)
Syracuse (22)
Wake (24)
BC (25)
GT (26)
NC State (30)

link (https://www.teamrankings.com/ncaa-basketball/ranking/in-conference-sos-by-other)

Not surprising that all of the ACC teams are in the top 30 nationally, since the conference is so brutal.

Wheat if you are actually arguing that home team advantage does not play a factor in college basketball, I'm not really sure what to tell you. There is a plenty of evidence this year that playing at home makes an enormous difference. Claiming that playing at home doesn't matter is something a team would say when they're the ones playing at home.

And if you consider games against the top teams in the conference (UNC, Duke, Louisville, FSU, ND) who have separated themselves, UNC plays four games at home and one on the road (@Duke, which they lost). Meanwhile Duke plays two at home and four on the road. If you were to add UVA into the conversation, UNC gets them home and away while Duke only gets them away.


Reposted from the Syracuse post game thread. It has some interesting stuff on the site I link.


http://www.playoffstatus.com/accbasketball/accsosag.html#sflx

Unless I am misunderstanding something, I believe this is in-conference SOS sortable by full schedule, games played so far and games remaining.

If I am reading this correctly, Duke has the most difficult schedule for the year, UNC has the 11th.

From a games played standpoint, Duke has the 8th most difficult thus far, and UNC has the 12th hardest.

Remaining schedule, Duke has the most difficult remaining games by far, and UNC has the 9th most difficult remaining games.

BTW, there is a probability section as well that gives Duke a 15% probability of winning the ACCT (UNC 21%, Louisville 20%) with a 27% chance of participating in the Championship game (UNC 35%, Louisville 35%, FSU 27%)

[BandAlum83]

Reposted from the Syracuse post game thread. It has some interesting stuff on the site I link.


http://www.playoffstatus.com/accbask...osag.html#sflx

Unless I am misunderstanding something, I believe this is in-conference SOS sortable by full schedule, games played so far and games remaining.

If I am reading this correctly, Duke has the most difficult schedule for the year, UNC has the 11th.

From a games played standpoint, Duke has the 8th most difficult thus far, and UNC has the 12th hardest.

Remaining schedule, Duke has the most difficult remaining games by far, and UNC has the 9th most difficult remaining games.

BTW, there is a probability section as well that gives Duke a 15% probability of winning the ACCT (UNC 21%, Louisville 20%) with a 27% chance of participating in the Championship game (UNC 35%, Louisville 35%, FSU 27%)

Also, according to the site, Duke is in control of its destiny for a #3 seed, meaning, if we win out, we are assured of a double bye.

ETA: link to seeding probabilities. http://www.playoffstatus.com/accbasketball/accstandings.html

[Wheat/"/"/"]

So I found a site that provides in-conference SOS rankings. They are the rankings to-date and thus do not include the remaining games on the schedule. Here are the ACC teams with their rank among all teams in parenthesis.

Virginia (1)
Louisville (5)
Clemson (7)
Duke (9)
FSU (15)
VA Tech (16)
ND (17)
Pitt (18)
UNC (19)
Miami (20)
Syracuse (22)
Wake (24)
BC (25)
GT (26)
NC State (30)

link (https://www.teamrankings.com/ncaa-basketball/ranking/in-conference-sos-by-other)

Not surprising that all of the ACC teams are in the top 30 nationally, since the conference is so brutal.

Wheat if you are actually arguing that home team advantage does not play a factor in college basketball, I'm not really sure what to tell you. There is a plenty of evidence this year that playing at home makes an enormous difference. Claiming that playing at home doesn't matter is something a team would say when they're the ones playing at home.

And if you consider games against the top teams in the conference (UNC, Duke, Louisville, FSU, ND) who have separated themselves, UNC plays four games at home and one on the road (@Duke, which they lost). Meanwhile Duke plays two at home and four on the road. If you were to add UVA into the conversation, UNC gets them home and away while Duke only gets them away.

I'm arguing that it looks like you guys are searching for some justification to make UNC's lead in the conference easier to swallow, let's just play ball...:)

[UrinalCake]

.BTW, there is a probability section as well that gives Duke a 15% probability of winning the ACCT (UNC 21%, Louisville 20%) with a 27% chance of participating in the Championship game (UNC 35%, Louisville 35%, FSU 27%)

Thanks for the link! Those numbers for the ACCT seem way to high for us. They might not be factoring in our lingering injuries or lack of depth, which are going to seriously handicap us in the three-games-in-three-days (or possibly four) type of format. I would give FSU a really high chance of winning based on their depth.

Wheat, you're probably right that Duke fans are searching for alternate reasons to explain why UNC is in first place. But if that's true, then why do UNC fans spend so much time complaining about the supposed easy paths we got to our 2010 and 2015 titles?

[jcastranio]

The unbalanced schedule has given UNC a tremendous advantage this year. On the one hand, that is just the way the dice fall. On the other ...

The unbalanced schedule makes it difficult to determine just how strong teams are in relation to one another. UNC (who I admit I despise) is the top team in the conference and a potential #1 seed. However, they lost their two non-conference games of note (Indiana and Kentucky). They lost three of their away games (G Tech, Miami, and Duke). Their away games in conference are all against the weakest teams in the conference (except for Duke and VA). So - they look good. They are good. Are they head and shoulders above all the other teams in the conference? Who knows?

[jv001]

The unbalanced schedule has given UNC a tremendous advantage this year. On the one hand, that is just the way the dice fall. On the other ...

The unbalanced schedule makes it difficult to determine just how strong teams are in relation to one another. UNC (who I admit I despise) is the top team in the conference and a potential #1 seed. However, they lost their two non-conference games of note (Indiana and Kentucky). They lost three of their away games (G Tech, Miami, and Duke). Their away games in conference are all against the weakest teams in the conference (except for Duke and VA). So - they look good. They are good. Are they head and shoulders above all the other teams in the conference? Who knows?

Like you said, it's hard to tell who's the best team. I say that there are some real good teams: Louisville, VA, Duke, ND, FSU and the Cheats. But there are some teams that have proven they can step up and beat those very good teams: VT, Wake, GT, Cuse, Clemson and Miami. The in-balanced schedule leaves a lot questions to be answered. I guess we'll get some answers in the ACCT. GoDuke!

[House P]

Here is one way you can attempt to "normalize" unbalanced schedules.


Use KenPom or another system to estimate how many games an average ACC team would be expected to win against each team's schedule (taking into account opponent strength and the location of each game).
Compare how many games a team actually won vs how many an average team would be expected to win.
Rank the teams according to how many games each team won above or below what would be expected by an average ACC team playing the team's schedule.

For example, let's look at Duke's last 3 games. Using the KenPom ratings, an "average" ACC team with a Adjusted Expected Margin of 17 would be expected to be a 1.9 point underdog at Syracuse, a 3.9 point favorite at home against Wake, and a 10 point underdog at UVA. This translates to a 43% likelihood of winning at Syracuse, a 64% likelihood of winning at home vs Wake and an 18% likelihood of winning at UVA. Thus, an average ACC team would be expected to win an average of 1.25 games over this 3 game stretch (0.43+0.64+0.18=1.25). If a hypothetical team actually won 2 of these 3 games, this would count at winning 0.75 games above expected.

Similarly, an average ACC team would be expected to win 1.81 games over a 3-game stretch at home against Va Tech and Pitt and an away vs Miami. If a hypothetical team actually won 2 of these 3 games, this would count at winning 0.19 games above expected.

Using this method, Duke's winning 1 of 3 @FSU, @Louisville, and @UVA is more of an accomplishment than winning 2 of 3 at home vs Ga Tech, BC, and NC St.

Here is a look at the "normalized" ACC standing based on this approach. While some teams have easier schedules than others, the difference isn't all that big and the normalization mostly serves to break ties. So, while UNC does appear to have the easiest ACC schedule (the average ACC team would be expected to win 8.1 games vs UNC's schedule), the normalization doesn't eliminate their two game lead in the standings.




Games
Actual Wins
Expected Wins
Win over Expected
Expected Win%
Actual Win%


UNC
15
12
8.1
3.9
0.539
0.800


N Dame
15
10
7.4
2.6
0.490
0.667


Louisville
15
10
7.6
2.4
0.504
0.667


Fla St.
15
10
7.6
2.4
0.508
0.667


Duke
15
10
7.7
2.3
0.514
0.667


Miami FL
15
9
7.8
1.2
0.521
0.600


Va Tech
15
8
7.1
0.9
0.473
0.533


Virginia
15
8
7.2
0.8
0.480
0.533


Syracuse
16
9
8.4
0.6
0.524
0.563


Ga Tech
15
7
8.0
-1.0
0.534
0.467


Wake
16
7
8.5
-1.5
0.533
0.438


Clemson
15
4
6.8
-2.8
0.450
0.267


Pittsburgh
15
4
7.2
-3.2
0.480
0.267


NC St.
16
4
7.9
-3.9
0.496
0.250


BC
15
2
7.3
-5.3
0.484
0.133



For kicks, I looked at Gonzaga's conference schedule and estimated that the average ACC team would be expected to win 13.5 games vs this schedule. Thus, Gonzaga has won 3.5 games above what would be expected by an average ACC team. If you consider Gonzaga's entire schedule, they have won 6.65 games more than would be expected of an average ACC team.

DISCLAIMER: All the calculations above were made in a custom spreadsheet, so it is entirely possible that errors were made.

[pfrduke]

(bunch of really great stuff)

This was excellent and I like this method of looking at the schedules. Very minor quibble - it looks like Syracuse and Wake have had easier schedules based on this analysis than UNC.

And poor Clemson - not only did they lose a ton of heartbreakingly close games, they did so against what has been, to date, the toughest schedule.

Will be interesting to see the final numbers in a little over a week.

[BandAlum83]

This was excellent and I like this method of looking at the schedules. Very minor quibble - it looks like Syracuse and Wake have had easier schedules based on this analysis than UNC.

And poor Clemson - not only did they lose a ton of heartbreakingly close games, they did so against what has been, to date, the toughest schedule.

Will be interesting to see the final numbers in a little over a week.

You are correct that Syracuse and Wake have had easier schedules played thus far as compared to UNC. Based on the website I linked (http://www.playoffstatus.com/accbasketball/accsospg.html#sflx) above, to date (going into this week - it appears it is updated weekly, not as games are played) here is the strength of in-conference schedule played to date. UVA's schedule has been slightly stronger than Clemson's:

Virginia
Clemson
N.C. State
Pittsburgh
Florida St.
Notre Dame
Boston College
Duke
Virginia Tech
Louisville
Miami
North Carolina
Georgia Tech
Syracuse
Wake Forest

Projected strength of conference schedule when all is said and done is below. Of course this could change based on performance in the remaining games. For instance, if we lose all the remaining games, then the strength of schedule for all who played us drops.

Note that by the end of the season, Duke moves to the top of the strength of schedule list. 3 of our last 4 games are on the road, and all are against upper tier teams. UNC moves up one spot in SOS, probably because of the @uva game


Duke
Pittsburgh
Florida St.
Virginia
N.C. State
Miami
Clemson
Boston College
Louisville
Notre Dame
North Carolina
Georgia Tech
Virginia Tech
Syracuse
Wake Forest

[BandAlum83]

The site I linked has a bunch of really interesting off shoots if you play with it.

Check out this Duke "What if" Scenarios:

http://www.playoffstatus.com/accbasketball/dukewhatif.html

If we win all 3 remaining games, we have the following probabilities:

#1 seed for ACCT: 56%
#2 seed for ACCT: 31%
#3 seed for ACCT: 13%

If we win 2 of 3, we are 90% likely to get a double bye.

[CDu]

The site I linked has a bunch of really interesting off shoots if you play with it.

Check out this Duke "What if" Scenarios:

http://www.playoffstatus.com/accbasketball/dukewhatif.html

If we win all 3 remaining games, we have the following probabilities:

#1 seed for ACCT: 56%
#2 seed for ACCT: 31%
#3 seed for ACCT: 13%

If we win 2 of 3, we are 90% likely to get a double bye.

Yes, by winning all 3, we'd be guaranteed to be ahead of FSU (by beating them) and Notre Dame or Louisville (one of them has to lose head to head). That would lock us into the top 3. The scenarios then would be as follows:

a. UNC wins at Pitt and at UVa, Louisville wins out (we end up the #3 seed)
b. UNC wins at Pitt and at UVa, Louisville loses a game or more (we get the #2 seed)
c. UNC loses one, either at Pitt or at UVa, Louisville wins out (we get the #1 seed in a 3-way tie with Louisville #2 and UNC #3)
d. UNC loses one, either at Pitt or at UVa, Louisville loses a game or more (we get the #1 seed in a tie with UNC #2)
e. UNC loses out, Louisville wins out (we get the #2 seed in a tiebreaker with Louisville #1)
f. UNC loses out, Louisville loses one or more (we get the #1 seed)

Note: a couple of those scenarios have been simplified because we would hold the tiebreaker over Notre Dame regardless.

[Newton_14]

The unbalanced schedule has given UNC a tremendous advantage this year. On the one hand, that is just the way the dice fall. On the other ...

The unbalanced schedule makes it difficult to determine just how strong teams are in relation to one another. UNC (who I admit I despise) is the top team in the conference and a potential #1 seed. However, they lost their two non-conference games of note (Indiana and Kentucky). They lost three of their away games (G Tech, Miami, and Duke). Their away games in conference are all against the weakest teams in the conference (except for Duke and VA). So - they look good. They are good. Are they head and shoulders above all the other teams in the conference? Who knows?

They said on the radio this morning that uncCheat does not have a road win so far against any of the Top 10 teams in the ACC standings. That's bizarre....

[-bdbd]

Was surprised during the L'ville-NC telecast them mentioning that that was going to be their only meeting of the season, AT Chapel Hill. Whereas Duke only saw them AT L'ville. That is huge, since they are arguably the third best team in the league. So, yes, certainly "unbalanced."

[CDu]

They said on the radio this morning that uncCheat does not have a road win so far against any of the Top 10 teams in the ACC standings. That's bizarre...

Well, to be fair, they have only played 3 of the top 10 ACC teams on the road. They are 0-3 against the top 10 ACC teams on the road. They have played Clemson, BC, and State (the three worst teams record-wise in the ACC) on the road, and are 3-0 against those guys. They also beat Wake in Winston-Salem. They still have road games against Pitt (#12) and UVa (#9).

So they will have played 4 road games against top-10 teams and road games against each of the #11-15 teams. Conversely, they have home games against all of the top-10 teams except Miami. And of those 9, 5 of them they don't have to travel for a road matchup.

That's not to blame UNC. They didn't write up the schedule. But they almost couldn't have written up a more favorable schedule this year.

[UrinalCake]

Thank you HouseP for that excellent analysis. This was exactly what I was looking for. I remember taking a closer look at UVA's schedule back in 2014 when they won the regular season but appeared to have a very favorable schedule. My analysis was not nearly as detailed, but I basically concluded that the schedule might have netted them one additional win, whereas they won the conference by two games, so I had to give them credit. They also went on to beat us in the ACCT Final, so they were clearly deserving of the "best team in the ACC" title.

I guess the other quirk to consider is that we have in many cases played better on the road than at home. Narrow wins over Wake, Clemson, Pitt, and even Miami (which was closer than the final score indicates) as well as the loss to State suggest that maybe we're actually better off playing some of the easier teams at home and the tougher teams on the road. The Syracuse game was one that I felt we probably would have won had it been played at home, but @VT, @FSU, and @Louisville I don't think would have made a difference.

[Skydog]

This just creates other anomalies. We only had one shot against Louisville, in their gym, and came up short. It doesn't make sense to me to count that equally with the scenario where we lost to Louisville both home and away.

I don't know if it is helpful but I would point out that there are two distinct problems with unbalanced schedules. Uh-no's system adjusts for one of them but not the other.

First problem - Which teams do you get to (or have to) play twice? The "lucky schedule" teams get to play weaker teams twice while the unlucky schedule teams play the tougher teams twice. Uh-no's system adjusts beautifully for that problem. Say UNC gets to play BC twice - they still can only amass one win under this system, so no real schedule advantage. Vice versa - if GT has to play Louisville twice they can still only be saddled with a max of one of loss so no schedule disadvantage.

Second problem- For the teams you play only once - is the home/away distribution favorable or unfavorable for your particular team? Top tier teams benefit if they face other top tier teams at home and their weaker opponents on the road. That distribution gives them the best chance of sweeping or coming close to it. Interestingly the very bottom tier teams benefit from the exact opposite type of schedule - toughest teams on the road, weakest opponents played at home. Teams like BC generally aren't going to beat top teams wherever they play so they may as well play them on the road. Their best hope is if they face weaker teams at home and rack up a few wins that way. For middling teams I think the home away distribution will tend to matter less, but will still matter. For example every middling team would want to play BC on the road and save their home games for tougher opponents.

Uh-no's system fixes the first problem but not the second. So his systems conference rankings are an improvement (fairness wise) but not a cure. The only way to adjust for the 2nd issue is using more advanced/complicated SOS metrics as some of you have already pointed out.