...in the poll that counts. And Michigan St is ranked ahead of us ( 7th ). This new " NET" system that replaces the RPI for ranking and selecting ncaa tournament teams is a complete train wreck. Check out their crack head top 20:
https://www.cbssports.com/college-basketball/news/ncaas-new-net-rankings-debut-ohio-state-shockingly-no-1-with-duke-no-6-and-kentucky-no-61/ ↗ ↗
The NCAA just can not help themselves. They have to continue to prove their ineptness to the sports world.
It seems pretty awful, but I'm going let the NET system collect more data before I get too mad. During conference play, if KU is still ranked in the Top 5 in the AP/Coaches Poll but ranked outside of the Top 10 in the NET system, then I'll start to gripe more.
It's worrying that a team that the Jayhawks beat head-to-head (Michigan St.) is currently ranked higher on NET rating.
@approxinfinity It simply makes no sense. Could you imagine the top rated team in the country getting a 3 seed?
It's not all bad. The entire Big 12 conference is rated in the Top 50 in the NET system and Kentucky is rated at 61. I think that's pretty cool.
Ah, you beat me to it. Came here to post the same! At least they got one thing right, Kentucky is #61!
They should use the same thing for football. Maybe our team would make the playoff!
All of the raw data-based rating systems look like garbage this early in the season. RPI looks worse than NET does right now; they have Kentucky ranked 81st and Georgia Southern ranked 2nd.
The entire Big 12 is ranked in the Top 50 in the NET system which means that during conference play Kansas will start looking pretty good. We just need to be patient. The NET rating system will start to normalize after it receives more data.
Wow they finally found a system worse than BPI lol
Like most computer systems, it gets better as the season moves on. It is the equivalent of a player going in at the end of the game for one minute and hitting a 3 point shot. If you project that number over 40 inured gives you a ridiculous 120 points but once that player has gone in and played more minutes and not scored the numbers will come down to earth. Right now there are too few games and the trends are not set; once teams have played 12-15 games the numbers will get better and by the end of the season the will be a lot more stable and representative.
Pomeroy has Duke at #1 followed by KU, Virginia, UNC, Nevada, Gonzaga, Michigan and Michigan State.
Sagarin has Duke as #1 followed by Michigan State, Gonzaga, Michigan, UNC, KU, Virginia and Texas Tech.
Neither of those rankings is is even close.
I'm worried about this system only because it fails (at least so far) to give credit for quality of wins, awarding quantity instead.
A 5 point win over Texas shouldn't be worth less than a 10 point win over Texas State. This is just weird. More thoughts tomorrow.
What they could have done is just wait a while until the system collected data before releasing the rankings. Release this going into conference season and I think this would make more sense.
@BeddieKU23 Good idea. Just as they wait until fb season is at least half over before releasing the NCAA "where the playoff contenders stand" ranking.
I've never understood why anyone cares about the rankings. Final Four decides everything and leaves no doubt who is #1 every year.
This is an entire season long system that won't be complete until the last game is played. I won't grade it until the last one is produced.
It's fair to take the stance of not grading it yet. So do no teams start with a preseason ranking? Is that the deal? If the system valued both Michigan State and Tennessee as top 10 opponents you'd think we would have to be top ten. Is the formula a secret sauce? Sorry so lazy. Lygtfm (let you Google that for me @JayHawkFanToo )? ;)
It's a way to gin up the results they want. It's been proven by their first result that it has no basis in reality. Michigan St 4 spots ahead of us after we beat them.
Here you go... ↗ this is the NCAA version by Andy Katz that now posts on the NCAA site.
Here is the Sports Illustrated ↗ explanation.
Okay, let's see if I can explain this. The NET has five components, ranked from highest weight to lowest:
1.Team Value Index, which is based on Game Results (i.e. wins or losses) factoring in the opponent, location and winner.
2. Net efficiency, which is just offensive efficiency minus defensive efficiency. I will guess that very strong defensive teams will benefit here.
3. Winning Percentage (this one is straight forward)
4. Adjusted Winning Percentage (a bit confusing, but somewhat easy to understand)
5. Scoring Margin (capped at 10 points).
The first red flag to me is that NET considers opponent, location and result to get its Team Value Index. But how do I rank teams in the first step? KU plays Tennessee on Friday at a neutral site. KU wins. So I know how to do the location and result. But what value do I give Tennessee? The NCAA says this is designed to give credit to teams that "beat other good teams." Tennessee is ranked 27th by NET. So does KU get credit for beating the 27th best team, or the 5th best team (according to AP) or something in between? And remember, this factor gets the MOST weight! Yet I can't figure out how they are even evaluating teams at the start of all of this. Moving on.
NET efficiency makes sense. It's something like the KenPom rankings. It will take some time to balance out, although it should be noted that right now KenPom's top ten is Duke, Kansas, Virginia, North Carolina, Gonzaga, Nevada, Michigan, Michigan State, Auburn and Tennessee. By the eye test, there are some misses (Duke ranked first, mostly on the strength of blowing out everyone, then losing a close game to Gonzaga), but I have no doubt that every one of the teams I just listed is a very good team, potentially Final Four quality. Its fairly easy to reverse engineer this part because the numbers are out there.
Winning percentage is easy. No issues here.
The adjusted winning percentage is a bit of an adventure. You get 1.4 "wins" for a road win, 1.0 "wins" for a neutral site win and 0.6 "wins" for a home win. For losses, the reverse is true. 0.6 for a road loss, 1.0 for a neutral loss and 1.4 for a home loss. This seems counter intuitive, as teams are punished for losing at home to very good teams. That makes very little sense.
Capping the scoring margin makes sense, but the cap is too low. There's a big difference between losing by 10 and actually getting blown out. Getting blown out isn't really reflected in this metric, which makes the margin useless.
The biggest flaw with NET is that it counts wins and losses at least twice, possibly three times. It counts it once in the winning percentage (WP) , then counts it again in adjusted winning percentage (AWP). It also counts it at the top in the Team Value Index (TVI). The AWP and WP do not account for opponent. It is entirely based on result (win or loss). TVI accounts for the opponent, but isn't clear on how that opponent is assigned a value.
The AWP will really skew once conference play starts. Imagine being a team like Texas Tech, knowing you have KU, K-State and WVU all coming to Lubbock. If you lose all three of those games, AWP counts that as 4.2 losses against you. If you win all three of those games, it counts as 1.8 wins. That seems... absurd. Those are three tough games. It rewards you for playing road games, but because it doesn't factor in opponent when doing AWP, scheduling weaker opponents on the road (you won't be punished for the loss as much) will actually help your AWP! That makes no sense.
It will be interesting to see how these rankings even out as more data is input, but I am not optimistic.
BigBad said:
Final Four decides everything and leaves no doubt who is #1 every year.
No way. Was George Mason a top 4 team in 2011? Countless examples like this.
I about said "be patient, more info may change it up for the better as the season moves on" -- until I say what Nate Silver said.
Nate Silver is jumping the gun. Look at what Ken Massey, who I have mentioned before and believe has the best rating system, had to respond.
!0_1543356275002_upload-eb071e7f-7830-498c-ad7c-6f61db4f0400 ↗
I will guess simulations of the system were run using information from previous years at different times of the season and the final results were found to be valid...at least a reasonable person would think they did.
It's a steaming pile. What Nate said.
After stepping back and reevaluating this, I am ok with it. This is a simple and transparent tool that places teams within the quadrants for the seeding later on. So, playing a weak schedule hurts you even more, imo.
If you play a weak schedule, and win, the NET puts you up higher. If you are higher in the NET than you should be, you will be gifting other teams higher quality wins and sacrificing them yourself. Once this is broken out into quadrants, that is when the evaluation starts for seeding. They didn't seed teams based on RPI alone in the past and they won't with this either. As far as lumping teams in to quadrants, this seems like it will serve it's purpose just fine by season's end.
It appears that all teams started as equals which makes sense for a computer based program. why throw in a human prediction after all. It also appears that as the date a game is played is not tracked the team order will be recalculated each and every time a game is entered into the system taking into account all of the games played to that point by every team no matter how remotely connected. That would result in a great deal of volatility at the beginning which would slowly stabilize as more data was entered. I am curious if the program is setup to recalculate at each entry or if it will be done daily or weekly. I would have set it up to show real time results however I likely would at most show a daily list with a weekly list being most probable just to avoid all the screaming as teams bounce about from games they are not involved with.
I haven’t done any programming in a long time so I may be totally off on some of this to be sure. I do cringe more than a bit about the amount of data that is going to be crunched each time the data is recalculated. Just keep in mind that to error is human and it takes a computer to really screw things up!!
The amount of data is not that big. Let’s assume that by the end of the season teams have played 30 games each times 350 teams divided by 2, since each game involves 2 teams, results in app. 5,250 games which in computing is a very small data set that would take seconds to recalculate and could be shown in real time but I assume it will be done initially weekly and later in the season hopefully more often.
@JayHawkFanToo information will be tallied for each team for each game so can’t divide by 2 I believe. However you are correct in that it will take just a few seconds to tally things up. With multiple data points per team and multiple calculations the data will add up however I am not worried about the data creating storage or runtime issues more of the possiblity of a small error compounding daily into a huge issue.
So....the NUMBER 1 SEED in the new ranking system gets their ass handed to them by a 3-2 Syracuse team on their home floor. What Nate said.
Michigan will beat UNC by 10.
Still not digging these NET ranking. Dook loses at home to a bubble team in Syracuse albeit missing some players but still only fall from 2 to 4 while we beat a Texas team that’s probably fairly even maybe a touch better on paper than Syracuse and drop from 12th to 13th. Yet a 4 loss Nebraska and UNC team sit 10th and 11th tho they ain’t missing players or playing a schedule that as tough as ours thus far. Just seems crazy that if they base a ton of those ranking and the tournament was tomorrow we’d be a 4 seed with 2 losses against Kenpoms #1 SOS.
@kjayhawks it is replacing the rpi. Not seeding teams. Just keep that in mind.
@kjayhawks What is unimpressive about it so far?
Side by side:
!0_1547668706280_Top 50 comparison.JPG ↗
Looks more accurate than the RPI through the top 50 to me. But, I'd love to hear why you disagree.
@Kcmatt7 NET still has some big flaws as well. Houston at 8, Nebraska at 10 immediately stand out to me.
Other teams that are nowhere near what their NET ranking is are LSU, Purdue, Louisville, Wisconsin, Nevada, San Francisco, Florida, Texas, Liberty, Lipscomb, and Seton Hall among the top 50 teams. Nevada is in there because of 23 is way too low compared to how good they actually are.
I'm not saying Houston is a bad team, but they're not 8th and Nebraska isn't the 10th best team.
I agree it's better than the RPI, but it still has some major flaws in its system that need to be worked out before the RPI should be kicked to the curb.
KU sure knew how to crush the RPI. LOL
@Texas-Hawk-10 Fully agree with you. Definitely has flaws. I'm just trying to show that it isn't any bit of a downgrade from what it replaced. Maybe it was a waste of time because it is pretty close to the same dang thing, but it isn't like SO atrocious that I can't believe they are using it.
@Kcmatt7 Well I pretty much stated my issue with it but while we are on topic, Nebraska and Houston are way too high along with UNC. Nevada isn't getting any love IMO. I personally think the NET must be putting too much on margin of victory because there isn't a rating system that I've seen KU lower than 9th in. I prefer Kenpom but don't always agree a 100% with it. I'm just saying a team with only 2 losses against the top SOS land should be higher. I don't have an issue with the top 4 in NET. I also never said that RPI is superior ether. I'll guarantee if you pick a bracket of off any of the ratings systems it will be junk after the first weekend.
I don't know... it seems to point to our late game drop offs... barely holding on to victories instead of keeping the score up.
I'm okay with that.
We need to close out games better and maybe this helps.
My point is simply that this is not meant to be the AP top 25. It actually falls somewhere in between the Massey and Kenp rankings if you compare it to that. Much closer than the RPI rankings. Both of which has Nebraska and Purdue high as well.
It is replacing only the RPI. Compare it to the RPI only and I think you see improvement. Instead of taking a bias against the NCAA just because, point out it’s flaws compared to the RPI. Is it better or worse? If better is the answer, than that’s an improvement and should be applauded instead of criticized. At least, that’s how I see it. We will see in 10 weeks though I guess.
@Kcmatt7 I'll believe when I see it when it comes to this replacing the RPI. The the RPI rankings still exist mean they're going to be used and the selection committee members aren't likely changing their ways any time soon. I still think the RPI will be used more heavily than the NET system in March.
