Professional and College Basketball Forums banner

41 - 60 of 85 Posts

·
Registered
Joined
·
2,316 Posts
I don't disagree with you. I'm not saying it's right, I'm explaining why.

I also think a lot of the NET system is ridiculously flawed in the sense that it REALLY hammers people for bad losses.

I desperately want to see what the Bonnies NET would be if it only included games Osun played in. Because the difference between RPI and NET is 46 spots: 81 RPI, 127 NET.

NET has the Bonnies behind Pacific, who is 3-8 vs Q1/2/3 and 13-0 vs Q4. SOS 333 OOC, 189 overall.
Bona is 6-7 vs Q1-2-3, and 8-1 vs Q4, SOS 226 OOC, 109 overall.

So I really do not understand that, or how anyone can say NET is somehow better than RPI, or WTF has happened to college basketball.
The Team Value Index seems to be the only component of NET that is skewed towards the P5 because they have more chances to win against “good teams”. It’s the most heavily weighted component and the algorithm used to calculate it Is not public knowledge so it can’t be manipulated like the RPI. Who knows how a loss impacts the calculation, but in Bona’s situation they have four losses against teams with NET rankings of 206, 187, 237 and 153. The win against Rutgers is a really good win but do the Bonnie’s have another win against a team with a better NET than their 119?

There is probably an inflection point where a good wining percentage won’t offset the lack of more than one quality win. This is where the NET is unfair to Bona, but A10 teams can’t afford to have four losses like that. To be honest, St. Bonaventure shouldn’t have those types of losses because the program is too good for that to happen. No doubt those games are more difficult than what they appear to be on paper, but if Bona wins those 4 games they are 11-1 going in to conference play with a marquee OOC win. The NET would reward that resume.
 

·
Registered
Joined
·
5,686 Posts
Man, we went around in circles about the RPI and how it was "unfair" or whatever.

Now we're going around in circles about the NET. I need more bourbon.
 

·
Registered
Joined
·
3,061 Posts
Man, we went around in circles about the RPI and how it was "unfair" or whatever.

Now we're going around in circles about the NET. I need more bourbon.
Well the RPI could be gamed. Power cons host Crap St. U. all OOC and everyone in the con is 12-0. If you notice, now that we've moved to all these showcase games, power conference challenges, etc, the RPI was less gameable. So they invent the NET and .. uhh .. "take care" of that little issue.
 

·
Registered
Joined
·
3,240 Posts
Discussion Starter #44
1. I'm not saying NET is "unfair" to the Bonnies in the sense that "they deserve a bid"
Although, if they basically win out until the A-10 final, then I will be all aboard the "Do the man with/without Osun and see that we're legit"

I was merely saying that I know Pacific isn't a good basketball team, they've simply played an easy schedule, and aside from a 4 OT home win over Saint Mary's, they've basically beat bad teams and lost to good teams; while the Bonnies have "bad losses" (again, because they were missing Osun for all but one loss), they're a decent team with blemishes. My brain cannot reconcile how extra bad losses outweigh going 1-8 vs the first THREE QUADS.

Basically it looks like a system designed to eliminate teams who suffer bad losses, and because "bad losses" are nothing more than "playing road conference games" + "Conference Effect" it appears designed to screw over non-P5 teams.


2. As to the whole "unfairness" of the system, all the inherent unfairness of the system is because it's impossible to build a system that can compensate for the "Conference Effect." The Conference Effect is how SOS is manipulated, and it's not an adequate measure of how hard a schedule actually is.

Now, as this pertains to the NET, the reason the NET is jacked up is because WHILE we don't know the formula, they've told us the 5 items...
1. Team Value Index
2. Net Efficiency
3. Win Percentage
4. Win Percentage (Adjusted)
5. Score Margin (capped at 10)

That's double-counting a ton of stuff multiple times, and the Conference Effect applies HERE, too. They say "SOS isn't included in Net Efficiency, but it kind of is. You're more likely to be efficient on Offense/Defense when the other team sucks and can't stop you/score on you. So the P7 racks up OOC wins at home against bad teams, then when they are less efficient in conference play, they're getting "Team Value Index" Wins and Losses.

If NET is so damned good of a formula... why have the committee? This is my problem with the entire process. Having a formula that you can try to manipulate isn't a problem if manipulation of the formula ALSO proves you're a good team. If you make a formula that actually works, then the Index it creates is going to look like what everyone thinks the bracket SHOULD BE, and you don't NEED a committee.

NET is a system designed to basically be a better baseline the committee can use to cherry pick whom they want in and whom they want out. We could use anything for that. If you're going to develop a system, develop a "Wins Above Lock" metric that actually assesses resumes based on historical inclusion and what actually IS the best resume.
 

·
Registered
Joined
·
1,720 Posts
All of these things about NET are true. What I dont think happens is the committee sits there and just uses NET to select teams. It is one tool they use. I give them lots of credit last year for putting Belmont in and UNCG was 1st out (probably should have been in but lets not get greedy). At some point you have to watch a team play. That goes back to JP's point. If you watch Pacific and Bona, you will see Bona is a way better team, yet Pacific is 7 spots ahead.
 

·
Registered
Joined
·
1,433 Posts
All of these things about NET are true. What I dont think happens is the committee sits there and just uses NET to select teams. It is one tool they use. I give them lots of credit last year for putting Belmont in and UNCG was 1st out (probably should have been in but lets not get greedy). At some point you have to watch a team play. That goes back to JP's point. If you watch Pacific and Bona, you will see Bona is a way better team, yet Pacific is 7 spots ahead.
The bottom line for Bona is that none of the formulas give you a break for injuries. They don't see that Osun was out, they just see that Bona lost a bunch of games to crap teams, thus to the computer, Bona isn't that good. If Osun had been healthy all year, and Bona still lost those games, no one would be making an argument for Bona. Unfortunately, that's exactly what the computer sees, thus Bona gets a low ranking. I guess if Bona were on the bubble, the committee could take Osun's injury into consideration, but because they lost so many games without him, they're not close enough to the bubble for human consideration to come into play. It sucks, but in reality, it works this way in all sports. If Patrick Mahomes gets hurt and misses 8 games next year, then gets healthy, and has the Chiefs rolling at the end of the year, everyone will know that the Chiefs are one of the best teams in the league. However, if they don't win enough games while he's out, they'll be watching the playoffs on tv. No break is given for the injury. The only difference is that the system in pro football is black and white, while college hoops is murky, due to the sometimes large difference between leagues, and the way the field is determined.
 

·
Registered
Joined
·
1,695 Posts
The committee used to always say "the RPI is just one tool of many", but there were so many habitual patterns of at-large selections over the years that followed very specific RPI ranking criteria that it was a fabrication to say it was just one tool of many. It was, but it was the primary tool. They even told us. When being interviewed after the Selection Show, the chairs would always start running off RPI and SOS stats. They never referenced KPI, Ken Pom, BPI, or whatever. They excused or substantiated selections by rattling off RPI metrics. So I doubt the NET is any less important.

Interesting that the RPI's SOS metric is still used, but not the RPI itself -- even though SOS math principles are second cousins to the RPI which was deemed unsatisfactory and needed replacing!
 

·
Registered
Joined
·
3,240 Posts
Discussion Starter #49
That was the whole problem to me... Back when Greg Shaheen was in charge, it would INSTRUCT the committee on HOW to do their jobs. And if you look at his regime, the RPI was basically followed much more strict adherence unless there was significant cause to deviate from it.

Before Shaheen left, when you were looking at who was safe, you would basically take the top 33 teams without looking at them and mark them as locks. Because only once had a team outside the Top 35 been left out. Since he was shoved out so the P5 ADs could stack the field with their members, it became "Well, your RPI is good, but your SOS isn't good so you're really not that good." Which was ridiculous because RPI literally IS "Your W-L vs your SOS."

And then everyone starting bring in those other metrics as a way to justify how they stack the deck. The NET system is designed to be a New Number to point to, one which looks more like their deck-stacking opinions than mathematically objective RPI. But they still use it the same way they used RPI post-Shaheen.

NCAA vs NET
1 Baylor NET 2
2 Kansas NET 4
3 Gonzaga NET 3
4 San Diego St NET 1
5 Duke NET 6
6 Dayton NET 5
7 Louisville NET 7
8 WVU NET 10
9 Maryland NET 9
10 Fla St NET 14
11 Seton Hall NET 13
12 Nova NET 15
13 Auburn NET 17
14 Oregon NET 19
15 Butler NET 11
16 Michigan St NET 12

Funny how Gonzaga, SDSU and Dayton are combined 4 spots below their NET, isn't it?
 

·
Registered
Joined
·
3,240 Posts
Discussion Starter #50
The NET, the AP voters, the coaches' poll voters, and Stephen F. Austin all know Dayton is better than Duke... but the Committee is like "Nah, Duke is better than an Atlantic 10 school"
 

·
Registered
Joined
·
6,646 Posts
Further to that NET/RPI comparison, as of this morning the Bonnies’ NET and RPI rankings are 113 and 61, respectively. And no, I’m not saying Bona’s should be in the bubble conversation.
 

·
Registered
Joined
·
3,240 Posts
Discussion Starter #52
Further to that NET/RPI comparison, as of this morning the Bonnies’ NET and RPI rankings are 113 and 61, respectively. And no, I’m not saying Bona’s should be in the bubble conversation.
I pretty much am. Their RPI puts them between:

55 Michigan, 9 Seed in Bracket Matrix, 6-8 vs Q1 (2-2 away from home, including W vs Rutgers)
56 Indiana, 10 seed in BM, 3-7 vs Q1 (0-5 away) loss to Rutgers.
58 Stanford, 10 seed in BM, 2-5 vs Q1 (1-4 away)
59 Purdue, 11 seed in BM, 4-7 vs Q1 (2-7 away) loss to Rutgers.
61 Bona, not even talked about, 2-2 vs Q1, (2-2 away). 1-0 vs Big Ten
62 Minnesota, First Four Out in BM, 3-10 Q1 (1-9 away). loss to Rutgers
63 Providence, On someone's bracket, 4-7 Q1 (3-5 away). Loss to URI by 14, we lost by 6. (They also have 4 losses vs Q3 and Q4, as do we).

So we have a better win pct, better win pct vs Q1, better away Q1, and better similar opponents than all those guys. And that doesn't event factor in that we were missing Osunniyi for 8 games and went 1-7.

Injuries ARE considered. Comparing the Bonnies Offensive/Defensive efficiency with and without him -- which is so important to the NCAA at-large bids that they incorporated it into NET, mind you! -- the Bonnies average 67 possessions per game... without Osun, the Bonnies are 7.5 points worse on offense and 13.4 points worse on defense.

That translates to a 21-point swing. Would the Bonnies lose to Canisius, Siena, Vermont, URI and Ohio with a healthy Osun? They'd be 18-3 or 17-4, which would put them 23 or 25 in the RPI, and make them a Q1 win for Dayton and VCU.
 

·
Registered
Joined
·
6,646 Posts
I pretty much am. Their RPI puts them between:

55 Michigan, 9 Seed in Bracket Matrix, 6-8 vs Q1 (2-2 away from home, including W vs Rutgers)
56 Indiana, 10 seed in BM, 3-7 vs Q1 (0-5 away) loss to Rutgers.
58 Stanford, 10 seed in BM, 2-5 vs Q1 (1-4 away)
59 Purdue, 11 seed in BM, 4-7 vs Q1 (2-7 away) loss to Rutgers.
61 Bona, not even talked about, 2-2 vs Q1, (2-2 away). 1-0 vs Big Ten
62 Minnesota, First Four Out in BM, 3-10 Q1 (1-9 away). loss to Rutgers
63 Providence, On someone's bracket, 4-7 Q1 (3-5 away). Loss to URI by 14, we lost by 6. (They also have 4 losses vs Q3 and Q4, as do we).

So we have a better win pct, better win pct vs Q1, better away Q1, and better similar opponents than all those guys. And that doesn't event factor in that we were missing Osunniyi for 8 games and went 1-7.

Injuries ARE considered. Comparing the Bonnies Offensive/Defensive efficiency with and without him -- which is so important to the NCAA at-large bids that they incorporated it into NET, mind you! -- the Bonnies average 67 possessions per game... without Osun, the Bonnies are 7.5 points worse on offense and 13.4 points worse on defense.

That translates to a 21-point swing. Would the Bonnies lose to Canisius, Siena, Vermont, URI and Ohio with a healthy Osun? They'd be 18-3 or 17-4, which would put them 23 or 25 in the RPI, and make them a Q1 win for Dayton and VCU.
I hear ya, jp. But, one of these things is not like the others.

RPI 55 Michigan - NET 31
56 Indiana - 61
58 Stanford - 30
59 Purdue - 26
61 Bona - 113
62 Minnesota - 40
63 Providence - 57
 

·
Registered
Joined
·
3,061 Posts
I hear ya, jp. But, one of these things is not like the others.

RPI 55 Michigan - NET 31
56 Indiana - 61
58 Stanford - 30
59 Purdue - 26
61 Bona - 113
62 Minnesota - 40
63 Providence - 57
This is true, and I also do not believe for a second that the Bonnies are an NCAAT caliber team this season, or right now. However, if you do some digging on the discrepancies in NET/RPI you'll notice that other than a few outliers, the big negatives are non-P5 and the big positives are P5. I'll keep pointing to the Iowa St/SLU discrepancy because it affects an A10 school who I happen to think is very good. I'm sure it isn't the only wacky scenario either, but as of today:

Iowa State: RPI 119, NET 69 RPI Q's (1-8, 2-4, 3-0, 4-1) NET Q's (2-11, 2-1, 1-0, 5-1)
SLU: RPI 48, NET 75 RPI Q's (1-4, 3-2, 5-1,7-0) NET Q's (1-4, 1-2, 8-1, 6-0)

And mind you, Iowa St. just moved up 5 places in the NET for beating K-State. Who is terrible. I'd still love an explanation that comes close to making sense of this. You cannot tell me there is not some bias in that NET formula that favors "power conference".
 

·
Registered
Joined
·
2,316 Posts
St. Bonaventure has 5 losses to Q3 and Q4 teams. I don’t understand the thought process that the NET is flawed because “it hammers” teams with bad losses. I can somewhat buy the argument that there is a discussion to be had regarding Osun’s potential impact on these games.

We don’t like the NET because it looks at the ranking of an opponent vs. the opponent’s winning percentage. This favors the P5, and is a different discussion.
 

·
Registered
Joined
·
3,240 Posts
Discussion Starter #56
Right, there are two different arguments.

1. NET sucks because it just sucks. You can look at it and see that it sucks when we all know better teams are getting placed behind teams we all know are worse teams.
And the majority of the problems with NET stem from the "Conference Effect" which is going to exist in any system that doesn't acknowledge there are two realities in college hoops and doesn't try to reconcile the two.

2. Where the Bonnies should be viewed on the national landscape in terms of their resume (and how that view should be adjusted considering Osun's injuries).


Let's not confuse the two. NET doesn't suck because it lists the Bonnies so low. NET sucks because the thing it's trying to achieve is not being achieved; AND that thing it's trying to achieve isn't actually what they should be trying to achieve anyway.

Providence and St. Bonaventure really have similar resumes.
PC: 6-7 Q1/Q2 (3-5 away) BONA: 2-3 (2-2 away)
PC: 7-4 Q3/Q4 BONA 13-5

Providence isn't seriously considered because they're 13-11 and have four bad losses. Bonaventure isn't seriously considered because they have four bad losses.

So I understand why Bonaventure's full resume isn't close to the bubble, it's an NIT resume. However, once you add in the injury factor, this is an NCAA team, period. They just don't have an NCAA resume due to injuries.
 

·
Registered
Joined
·
3,240 Posts
Discussion Starter #58
JP, what are your thoughts on how Iowa St. is ranked above SLU? And how in the absolute fuck thy moved up 5 spots in the NET for beating K-State.
I mean, the short answer is because it's rigged towards P5. The more you play "bad teams" the lower you are, even if you win; and the more you play good teams, the higher you are, even if you lose.

This is why I want a "Wins Above Bubble" stat, that takes the resumes of every NCAA team seeded from "last at-large in" to overall #1, generates their Win Pct vs each category of RPI, home and road, and assesses every teams' win pct vs each group compared to that NCAA team average, and compiles it into a nice handy number based on 30 (so it looks like a win total with decimal places. Take the highest scoring teams in order each season, and you're done.

Because what they're doing is saying 4-10 vs Q1, that's a good team! Non-P5 team only has TWO wins against Q1.

But if you looked at win percentage vs each group, then you'd see that the non-P5 time is taking care of business at a better clip.

They haven't found a way to "ignore" good losses so they don't hurt you instead of REWARDING good losses, which is stupid because anyone can lose to a top team. Why does having lost nine times to Maryland, Michigan State, Iowa, Illinois, Wisconsin, Indiana, Rutgers, Penn State, Purdue, Ohio State, Minnesota make you a good team? Pick nine of them, I'll play them all tomorrow 1 vs 5, and I guarantee you I can pull off losing to them. Doesn't make me an NCAA squad.
 

·
Registered
Joined
·
1,695 Posts
I think the NET is a bad system because they're afraid to share how it really works.
Basically. We have no real idea how much of the metrics are baked into the cake. They have given us a breakdown of the five basic ingredients into the NET, but we have no idea how much is salt and how much is pepper. From what I have tracked, it appears metrics are weighted more heavily than we even think. If you can leave the starters in and beat a bad team by 45 instead of 30 -- do it. It wont help your margin metric which is capped, but it will help your offensive and defensive efficiency metrics.

I think this is why Dayton is ranked so high. The Flyers havent beaten a Top-25 team all year. Can any other ranked team say that, nevermind ranked team in the Top-10? I dont know -- but its probably a short list even if it exists. So I think UD is ranked very high b/c their offense is Top-2 in efficiency and the defense is solid. Margin of victory is also among the leaders in the country. UD also has a good Q1/Q2 record, but in all honestly name one win that really stands out. St. Mary's? Georgia and Va Tech are wetting the bed of late. VCU, Duquesne, and SLU(2) are probably the next best in line and none of those are locks to make the tournament either. Many teams that will never make the NCAA tournament have a couple wins vs ranked teams this year. Playing KU and COL to the buzzer is nice but they were still losses.

But it helps to have a NPOY candidate (Obi) that can do something SportsCenter worthy every night. We all know having one great player DOES influence people. Look at MSU. How many more mulligans can they get before folks realize they are good but not great. But Cassius Winston sits in the back of the mind of every pollster. And it will sit in the back of their mind on Selection Sunday too.
 

·
Registered
Joined
·
5,686 Posts
Two things seem apparent to me:
1 - I'm surprised that the math wizards who run some of the metric sites haven't reverse engineered the NET to show exactly how it is being calculated. Would think they have enough data points by now to do this.
2 - In theory, if you were using NET, the at-large bids would end at NET 48. Right now that puts 2 at-large bids for the A10, and 1 for the AAC, and 2 for the ACC(!).

There will be a lot of nonsense (different nonsense every year) when they pick someone in the 50's, and leave out someone in the 40's. Hello this year's Syracuse.
 
41 - 60 of 85 Posts
Top