Which team has a higher PR rank?

Team A

Notable Wins:
Home RPI 3
Road RPI 17
Home RPI 27
Road RPI 47
Home RPI 84
Road RPI 92

Losses:
Home RPI 6
Road RPI 28
Road RPI 30


Team B

Notable Wins:
Home RPI 3
Home RPI 32
Home RPI 50
Road RPI 71
Road RPI 86
Road RPI 90

Losses:
Road RPI 58
Road RPI 84

Assume really similar margins. Most wins and all losses on here are single digits.

This all depends on what all the other games are that you left out. Does one team play 10 games against 200+ rpi teams? While the other team players 0 games against 200+ rpi teams?

Or are these games listed the ONLY season games so far?

Posted by cubcub113 on 5/16/2020 2:21:00 PM (view original):
Which team has a higher PR rank?

Team A

Notable Wins:
Home RPI 3
Road RPI 17
Home RPI 27
Road RPI 47
Home RPI 84
Road RPI 92

Losses:
Home RPI 6
Road RPI 28
Road RPI 30


Team B

Notable Wins:
Home RPI 3
Home RPI 32
Home RPI 50
Road RPI 71
Road RPI 86
Road RPI 90

Losses:
Road RPI 58
Road RPI 84

Assume really similar margins. Most wins and all losses on here are single digits.

7.0.1

This makes my brain hurt.

Posted by cubcub113 on 5/16/2020 2:21:00 PM (view original):
Which team has a higher PR rank?

Team A

Notable Wins:
Home RPI 3
Road RPI 17
Home RPI 27
Road RPI 47
Home RPI 84
Road RPI 92

Losses:
Home RPI 6
Road RPI 28
Road RPI 30


Team B

Notable Wins:
Home RPI 3
Home RPI 32
Home RPI 50
Road RPI 71
Road RPI 86
Road RPI 90

Losses:
Road RPI 58
Road RPI 84

Assume really similar margins. Most wins and all losses on here are single digits.

7.0.1

my brain isn't working well enough yet this morning to think about all that but if you are asking this question, ideally, you'd be organizing the data in a way more consistent with how the PR works. what is the record vs top 25, 50, 75, 100? also rpi, record, average margin of victory? margin of victory is a factor but usually the heavy stuff all can be summed up with rpi and the record vs top X.

i'd recommend you do so for your own edification, im assuming whatever the answer is surprises you, and hopefully looking at it in a more similar way to the engine will reveal what it is thinking differently than you. you might not agree but at least you may understand? i actually like the PR - mostly because i compare it to the old scheme - its *so* much better, and i know no exception to that statement. but i never really considered it from the 'would i do better' type of angle. that said, i'll be happy to go down the rabbit hole with you, but preferably in a way consistent with how the PR report actually works.

Team A (8-3, 3-1 Home, 5-2 Away, 5 RPI, 3 SOS)
2-1 vs Top 25
2-2 vs 25-50
2-0 vs 51-100
1-0 vs 101-200
1-0 vs 201+

7 Average Margin


Team B (9-2, 3-0 Home, 6-2 Away, 11 RPI, 31 SOS)
1-0 vs Top 25
2-0 vs 25-50
3-2 vs 51-100
3-0 vs 101-200
0-0 vs 200+

11.5 Average Margin

So Team A has played a blisteringly hard schedule and Team B has played a schedule with 8 games against middling teams and done 4 ppg better in 1 more away game.

But doesn't HD (foolishly) rate based on the Top 100 matchups primarily? By that stat it's pretty even 6-3 vs 6-2

Posted by cubcub113 on 5/16/2020 3:23:00 PM (view original):
Team A (8-3, 3-1 Home, 5-2 Away, 5 RPI, 3 SOS)
2-1 vs Top 25
2-2 vs 25-50
2-0 vs 51-100
1-0 vs 101-200
1-0 vs 201+

7 Average Margin


Team B (9-2, 3-0 Home, 6-2 Away, 11 RPI, 31 SOS)
1-0 vs Top 25
2-0 vs 25-50
3-2 vs 51-100
3-0 vs 101-200
0-0 vs 200+

11.5 Average Margin

So Team A has played a blisteringly hard schedule and Team B has played a schedule with 8 games against middling teams and done 4 ppg better in 1 more away game.

its top X... 1-25, 1-50, 1-75, 1-100. those are the four metrics specifically added to the PR report (per seble when he did the work). note that home away *do not* factor in to the record vs top X.

so looks like:
A:
2-1 top 25
4-3 top 50
? top 75
6-3 top 100

B:
1-0 top 25
3-0 top 50
? top 75
6-2 top 100

even though A has better RPI, its looking like a meaningful advantage for team B in the top X metrics (top 50 difference is huge!). he also has the better record and margin. my money is on B higher in the PR report, and i'm pretty confident in that answer.

time to make me look the fool? :)

the top 75 looks overwhelming too... 4-3 team A vs 4-1 team B. with that additional evidence im virtually positive B has a significant PR advantage.

now, should they? different question. of course a limited heuristic like record vs top 25/50/75/100 has limits. it can be exploited just like the old scheme (RPI based), but IMO its a lot better in most cases, and also less prone to manipulation. still quite prone though. i think i've said scheduling is about 20% of the success of a program, i'd probably still stand by that in today's day and age.

before 7 years ago, the manipulation game was simple - road wins against teams with good records. implementation was as simple as finding the best teams in trash conferences and beating up on them (they are still nothing compared to any decent human team).

now, its more complicated. you need to decide a bit which you need more - record vs top X or rpi help. usually its a bit of both. some road wins against high record opponents to pad the RPI a bit, but also, making sure to get a bunch of easy top 100 rpi wins. for higher teams, easy top 50 or even top 25 rpi wins - to be had at home, if necessary. if you can easily beat team, still go do that at their place. but home wins against the 25 rpi team are just as valuable as road wins vs the #1 rpi in the top X bit, while being massively easier. so if you need some top 25/50 wins but aren't sure you can get them, playing at home usually helps. i'd say most folks could follow a general model where their 10 games feature a roughly a 7-3 split with most of those 7 road wins against easier top 100 rpi teams, and with the 3 games confidently winnable games against higher rpi teams.

I'd say Team B is higher on the PR by about two spots. Both are probably top 10 though. Maybe something like 8 on the PR for Team B and 10 on the PR for Team A.

Posted by emy1013 on 5/16/2020 4:53:00 PM (view original):
I'd say Team B is higher on the PR by about two spots. Both are probably top 10 though. Maybe something like 8 on the PR for Team B and 10 on the PR for Team A.

Gold star.

Edit: You're in Wooden, so I'm guessing you did some research. haha

Posted by emy1013 on 5/16/2020 4:53:00 PM (view original):
I'd say Team B is higher on the PR by about two spots. Both are probably top 10 though. Maybe something like 8 on the PR for Team B and 10 on the PR for Team A.

Gil said it best, and I think you're right. The answer has to be a bit of a shock at first glance, compared to the obvious choice. Or we would discuss it most likely.

I'd say B is higher as well based off that. And the fact that HD doesn't put enough emphasis on "big time" wins, compared to above average bodies of work.

I don't have exacts. But I feel like in 10 game measurements, HD values a 10-0 record with 10 wins against rpi's 11-100 is valued higher than a 10-0 record with 5 wins against top 10 rpis and 5 wins against 101+ rpis. And I don't really feel like it should be that way (I left a lot of holes there with the numbers. Like were those +101 rpi teams in the 350s? But I'm just speaking vaguely about the numbers in general. I just feel that top 5 and top 10 wins should have a LOT more weight than they do)

I think its the RPI bag. My local college is having really trouble Net problems which I do not see fixing after the next Century nobody wants to build here with ACC right around the corner. We can barely schedule State which we are 1-1 in this era. I don't see anything changing though.

Posted by gillispie1 on 5/16/2020 4:06:00 PM (view original):
the top 75 looks overwhelming too... 4-3 team A vs 4-1 team B. with that additional evidence im virtually positive B has a significant PR advantage.

now, should they? different question. of course a limited heuristic like record vs top 25/50/75/100 has limits. it can be exploited just like the old scheme (RPI based), but IMO its a lot better in most cases, and also less prone to manipulation. still quite prone though. i think i've said scheduling is about 20% of the success of a program, i'd probably still stand by that in today's day and age.

before 7 years ago, the manipulation game was simple - road wins against teams with good records. implementation was as simple as finding the best teams in trash conferences and beating up on them (they are still nothing compared to any decent human team).

now, its more complicated. you need to decide a bit which you need more - record vs top X or rpi help. usually its a bit of both. some road wins against high record opponents to pad the RPI a bit, but also, making sure to get a bunch of easy top 100 rpi wins. for higher teams, easy top 50 or even top 25 rpi wins - to be had at home, if necessary. if you can easily beat team, still go do that at their place. but home wins against the 25 rpi team are just as valuable as road wins vs the #1 rpi in the top X bit, while being massively easier. so if you need some top 25/50 wins but aren't sure you can get them, playing at home usually helps. i'd say most folks could follow a general model where their 10 games feature a roughly a 7-3 split with most of those 7 road wins against easier top 100 rpi teams, and with the 3 games confidently winnable games against higher rpi teams.

Let's unpack the meaning of using just top X metrics...

Let's say two teams have wins against 1,5,10,15,45,65,85 and Team A has losses against 6 and 12 and Team B has losses against 60 and 80

Then:

Team A
Top 25 4-2
Top 50 5-2
Top 75 6-2
Top 100 7-2

Team B
Top 25 4-0
Top 50 5-0
Top 75 6-1
Top 100 7-2

Team A has better records in each category despite their record being SUPER worse.

As Emy said, B is 8th and A is 10th which strikes me as ridiculous. A has better wins and far better losses.

Nearly twice as hard of a schedule (.67 vs .59) and is only doing 4 points worse.

Posted by topdogggbm on 5/16/2020 5:24:00 PM (view original):

Posted by emy1013 on 5/16/2020 4:53:00 PM (view original):
I'd say Team B is higher on the PR by about two spots. Both are probably top 10 though. Maybe something like 8 on the PR for Team B and 10 on the PR for Team A.

Gil said it best, and I think you're right. The answer has to be a bit of a shock at first glance, compared to the obvious choice. Or we would discuss it most likely.

I'd say B is higher as well based off that. And the fact that HD doesn't put enough emphasis on "big time" wins, compared to above average bodies of work.

I don't have exacts. But I feel like in 10 game measurements, HD values a 10-0 record with 10 wins against rpi's 11-100 is valued higher than a 10-0 record with 5 wins against top 10 rpis and 5 wins against 101+ rpis. And I don't really feel like it should be that way (I left a lot of holes there with the numbers. Like were those +101 rpi teams in the 350s? But I'm just speaking vaguely about the numbers in general. I just feel that top 5 and top 10 wins should have a LOT more weight than they do)

i hear you, from the top X standpoint, there is no gradient. a road win over #1 rpi is equal to a home win against #100 rpi, in terms of the impact on the record vs top 100 component. obviously there is additional value in that #1 win in terms of rpi, record vs top 25/50/75 - but still. the fact remains that the victory over the #1 team gets its due in exactly 0 of the metrics.

in rpi, a road win over a #1 rpi team and a home loss to the #200 team is equivalent to a road win over the #200 team and a home loss to the #1 rpi team. a home over #1 rpi and a road win over #200 is also equal to a road win over #1 rpi and a home win over #200. so playing really tough games, it never really got its due in rpi. obviously with a flat 'record vs top X', beating the hardest teams in those brackets is going to be underrepresented value wise while beating the weaker teams is over represented. so its really a systemic issue. that said, if you want to win lots of titles, you should play those toughest teams anyway, so i don't think its a massive imbalance or anything.