Manipulating the NET | Syracusefan.com

Manipulating the NET

In the end it seems that NET and RPI wind up with the same inherent flaw. The calculations put too much value on something other than winning/losing.

RPI: you benefit too much from playing good teams even if they destroyed you, just playing them helped you.

NET: you benefit too much from margin against bad teams and that disproportionally boosts your conference.

EDIT: I do understand that both NET and RPI were trying to weight the value of the teams you won and lost to. Obviously that is necessary but not at the expense of undervaluing the most important part of a game result which is of course getting the W. In one thread jncuse talked about some tweaks that would give us something that landed in-between NET/RPI which on its face made a lot of sense to me.
 
Last edited:
Cusefan0307 said:
I have a feeling this is going to lead to some really garbage buy games in the next couple years that will not be entertaining to fans at all.
Expand...
Only if the committee takes the net rankings at face value. If the committee dives into the data like Danny. They should be able balance it out better. The Big 12 may end up performing well in the dance and may prove to be a better conference this year. But the Big 12 is certainly not 5 more tourney bids better than the ACC. Bids among the power conference should be fairly consistent and proportional, with less bids in years the mid majors are strong (this year) and more bids when the mid majors are weak.
 
Last edited:
jordoo said:
In the end it seems that NET and RPI wind up with the same inherent flaw. The calculations put too much value on something other than winning/losing.

RPI: you benefit too much from playing good teams even if they destroyed you, just playing them helped you.

NET: you benefit too much from margin against bad teams and that disproportionally boosts your conference.

EDIT: I do understand that both NET and RPI were trying to weight the value of the teams you won and lost to. Obviously that is necessary but not at the expense of undervaluing the most important part of a game result which is of course getting the W. In one thread jncuse talked about some tweaks that would give us something that landed in-between NET/RPI which on its face mad e a lot of sense to me.
Expand...
Averaging your net and rpi ratings could balance this out maybe?
 
This is why I love analytics, but also understand analytics is not the end all be all. You can do a lot with numbers. But it cant/wont tell the entire story because sports and humans perform differently in different situations. Their are no rules like there are in math.

Not to mention, once you understand how things are calculated, numbers can be fudged as the above shows. Numbers are strict. Human intervention needs to be able to utilize different metrics to make an informed decision.
 
I also think others forget that 6 20point losses also hurts our NET just as much as beating up on bad teams helps other teams’ NET.
 
I like the Oklahoma example since we both crossed paths with UNC.

Is the 8th place of 12 big12 team with a losing conference record and best win over Providence OOC and IowaSt really an 8 Seed?

Then we pull the thread and say ok the ACC is 9-3 against the Big12. UNC Handled Oklahoma and a middling Virginia Tech team handled Iowa State, been awhile but still.

Let's be honest too Wake and NCState and Pitt all pulled the same crap as the Big12, 6 or 8 Home Q4 cupcakes, many of which in the 300's that shouldn't even count. Luckily most of the rest of the ACC attempted a real noncon.

The committee is not stupid, they know who performs in the tournament and that December isn't as important as it used to be. I think we'll see a big shift towards looking at recency and road wins hopefully this year but definitely next year.
 
I'm no math/stat mind but it just seems like common sense:
it ought to be as simple as who you beat, regardless of score, and where (i.e. road wins should weigh more than neutral wins which should weigh more than home wins)... you shouldn't get credit for close losses nor should you get (more) credit for blowout wins...
like, who tf cares about style and efficiency? that should have no part in the formula. again, who you beat and where. are we really rewarding teams losing games efficiently?
am I oversimplifying?
 
Last edited:
pearl31 said:
I'm no math/stat mind but it just seems like common sense:
it ought to be as simple as who you beat, regardless of score, and where (i.e. road wins should weigh more than neutral wins which should weigh more than home wins)... you shouldn't get credit for close losses nor should you get (more) credit for blowout wins...
like, who tf cares about style and efficiency? that should have no part in the formula. again, who you beat and where. are we really rewarding teams losing games efficiently?
am I oversimplifying?
Expand...
But the Buffalo Bills are the best team in the NFL. They’ve had the best point differential for five years!
 
pearl31 said:
I'm no math/stat mind but it just seems like common sense:
it ought to be as simple as who you beat, regardless of score, and where (i.e. road wins should weigh more than neutral wins which should weigh more than home wins)... you shouldn't get credit for close losses nor should you get (more) credit for blowout wins...
like, who tf cares about style and efficiency? that should have no part in the formula. again, who you beat and where. are we really rewarding teams losing games efficiently?
am I oversimplifying?
Expand...
You can try it and see what it gets you. I think it would be even worse. And then all of that can be blowed up once a team has a hot night and shoots 75% from three. A transitive property in college basketball means that SUNY-ESF could be better than UConn - we just can't know because they haven't played, but ESF has beaten teams that have beaten teams that beat UConn! Until there's a ~180 game season where every P6+ team plays every other team 2-3 times there'll always be tradeoffs and teams and fans whining about what's holding them back: We get screwed by the refs! Team W's coach was stealing signs! We were shorthanded! They shouldn't get credit for that win, Team Y had food poisoning! Team Z's guards hated each other ...etc ad nasueum.

The best way (so far) to measure a team's performance within the limited data we're stuck with is to measure how they performed against their opponents relative to how all of the other teams played them. You measure how a team scores vs how do they defend per possession versus opponents compared to other teams.

I don't care about NET in particular, but the idea behind the various efficiency metrics IS NOT to get precise and accurate results, it's about getting close enough to separate the most wheat from the chaff.
 
Fly Rodder said:
Sure, because that's exactly what it's used for.
Expand...
Listen I don’t have a big problem with the NET like some. I’m just trying to have some fun.

That said I do think the Big 12 has a built in advantage this year and I am concerned we may see worse OOC matchups moving forward.
 
Fly Rodder said:
You can try it and see what it gets you. I think it would be even worse. And then all of that can be blowed up once a team has a hot night and shoots 75% from three. A transitive property in college basketball means that SUNY-ESF could be better than UConn - we just can't know because they haven't played, but ESF has beaten teams that have beaten teams that beat UConn! Until there's a ~180 game season where every P6+ team plays every other team 2-3 times there'll always be tradeoffs and teams and fans whining about what's holding them back: We get screwed by the refs! Team W's coach was stealing signs! We were shorthanded! They shouldn't get credit for that win, Team Y had food poisoning! Team Z's guards hated each other ...etc ad nasueum.

The best way (so far) to measure a team's performance within the limited data we're stuck with is to measure how they performed against their opponents relative to how all of the other teams played them. You measure how a team scores vs how do they defend per possession versus opponents compared to other teams.

I don't care about NET in particular, but the idea behind the various efficiency metrics IS NOT to get precise and accurate results, it's about getting close enough to separate the most wheat from the chaff.
Expand...
but surely a formula can be tweaked so as to not let one-off anomaly upsets by otherwise bad teams weigh too much - like, you'd still have to accumulate "quality" wins and avoid many "bad" losses as a season progresses to achieve a good ranking...
I dunno, I just think the idea of a team with a bunch of close-ish losses playing a "tough" schedule being more highly regarded than a team with a better w/l record against a similar schedule but with a few of the losses being large margins - is not the way to go...
at the end of the day it is about winning games, establishing over the course of a season you're a team that wins games against quality teams. The efficiency and scoring margin metrics have got to go.

kenpom is useful and instructive as a certain ranking system - but I have argued and will continue to argue that the most efficient teams aren't necessarily the best teams. this type of ranking, I feel, should be a tool for the committee to refer to - but not a system by which to officially rank teams for post-season consideration
 
Last edited:
you will end up with just watch the games.

Winning and losing doesn't mean everything, neither does pt differential.

you really as the committee are talking about like 20-40 teams that are in that bubble range. with condensed games you could watch 3-4 teams a night play in about 5 games in a couple weeks and have a much better feel about them. Sure you might not watch at their best. but you would see the flaws and the style.

no matter how deep a conf is, if you cant finish .500 in it you are out as an at large.
the b10 has like 5 teams in the middlish range
the b12 has 4
the Acc has 6
the SEC has 4
the BE has 2


OC matters, but really what you do over 18-20 games should matter more than an OC game played 3 months ago.
 
If the NET is easily manipulated, why did the ACC not manipulate it? Did the B12 really outsmart the rest of the country? People with full time jobs in the ACC office couldn't think of this before the season and advise the programs how to make their schedule? Or is it not true?
 
upperdeck said:
you will end up with just watch the games.

Winning and losing doesn't mean everything, neither does pt differential.

you really as the committee are talking about like 20-40 teams that are in that bubble range. with condensed games you could watch 3-4 teams a night play in about 5 games in a couple weeks and have a much better feel about them. Sure you might not watch at their best. but you would see the flaws and the style.

no matter how deep a conf is, if you cant finish .500 in it you are out as an at large.
the b10 has like 5 teams in the middlish range
the b12 has 4
the Acc has 6
the SEC has 4
the BE has 2


OC matters, but really what you do over 18-20 games should matter more than an OC game played 3 months ago.
Expand...
Impactful OOC games shouldn’t be just relegated to the beginning of the season, particularly with the revolving rosters from year to year. Playing only conference games for the last 3 months of the season, with imbalanced schedules in new mega member conferences, doesn’t give a true comparative intra-conference base for evaluation and is getting more difficult for even true inter conference rankings.
 
jordoo said:
In the end it seems that NET and RPI wind up with the same inherent flaw. The calculations put too much value on something other than winning/losing.

RPI: you benefit too much from playing good teams even if they destroyed you, just playing them helped you.

NET: you benefit too much from margin against bad teams and that disproportionally boosts your conference.

EDIT: I do understand that both NET and RPI were trying to weight the value of the teams you won and lost to. Obviously that is necessary but not at the expense of undervaluing the most important part of a game result which is of course getting the W. In one thread jncuse talked about some tweaks that would give us something that landed in-between NET/RPI which on its face made a lot of sense to me.
Expand...
Just want to comment on a few things on the RPI.

a) The RPI is still about W/L's. You do need to win games. The gaming / punishing came as a result of who you played, I don't think the biggest weakness was losing to great teams, but better to lose to #5 than #50 obviously.
b) The RPI punishes/destroys you when you play sub 280/sub 300 teams, especially more than a few. That is because 50% of the formula is the opponent's win%.
c) The weakness of the RPI, and where it has been gamed, is teams (and specifically leagues) instructed their teams to try to find sweet spot wins and avoid the punishers. More in the 100-200 category at home. All basically as winnable as sub 300 games (all 95%+ chance of winning), but would be rewarded better.

But I highlighted b) above because this is the natural hedge to the NET manipulation by the Big12. Its a simple solution. You want to stop teams doing what the Big12 has been doing this year, make the "NEW NET" = 50% current NET + 50% RPI.

I did a calculation maybe 10 days ago, where I found the average NET of Big12 teams to be 47, and the average RPI to be 67.

They are getting rewarded in one system, and getting punished in the other. Their true worth is probably somewhere in the middle of the two.
 
Last edited:
The NET was invented to fkkkk mid-majors. College sports are a money grabbing joke and will only get worse.

With that said, how many days until conference tourneys start?
 
Cusefan0307 said:
Listen I don’t have a big problem with the NET like some. I’m just trying to have some fun.

That said I do think the Big 12 has a built in advantage this year and I am concerned we may see worse OOC matchups moving forward.
Expand...
Money to be made picking against them in the tourney then

That said, I'm not picking against Houston and their D
 
pearl31 said:
I'm no math/stat mind but it just seems like common sense:
it ought to be as simple as who you beat, regardless of score, and where (i.e. road wins should weigh more than neutral wins which should weigh more than home wins)... you shouldn't get credit for close losses nor should you get (more) credit for blowout wins...
like, who tf cares about style and efficiency? that should have no part in the formula. again, who you beat and where. are we really rewarding teams losing games efficiently?
am I oversimplifying?
Expand...
They spend too much time trying to identify the best team(s), when they should looking into which schools are the most deserving. I'd say the same about CFB.

The best teams will identify themselves IN THE TOURNAMENT ITSELF. Buy winning and advancing.

It's not that easy, but it is also not that hard either.
 
You must log in or register to reply here.

Similar threads

Sbucketball
    • Like
Strength of ACC; NET & KenPom Tracker (24-25)
Replies
0
Views
371
Sbucketball
Sbucketball

Forum statistics

Threads
170,310
Messages
4,884,072
Members
5,991
Latest member
Fowler

Share this page

Online statistics

Members online
51
Guests online
922
Total visitors
973
Totals may include hidden visitors.


...
Top Bottom