Are you ready to win a title? (Part II)

In Part I we discussed what it takes to be a National Champion and what they all have in common. If you missed the article then I highly suggest going back and reading it before proceeding with Part II. The next question that came to mind after figuring out the make-up of a National Champion was, “If I don’t have an ODS of 25 then where should I be finishing?” In Part II we will look at ODS in relation to what round the team lost in to get an idea of where you should expect to finish with a value of X ODS. This will help to gage a program on if they are close to winning a title (or advancing further into the NCAAT) as well as distinguishing if there is much difference between the teams that end in each round. For example, are teams that finish in the Sweet 16 that much different than teams that finish in the Elite 8?

Once again I’ll be using final numbers from each season from Kenpom.com to develop ODS values and then graphing them using a population bell curve. The teams used in each round are the losers of that round, with the lone exception being the champions. This means the Sweet Sixteen (S16) teams are the teams who made it to the S16 but lost their game, simple enough. First up are the bell curves for the rounds of Round of 32 (R32), S16, Elite Eight (E8), Final Four (FF), Runner Up (RU), and National Champions (NC).


BELLCURVE OF ALL ROUNDS

Well, that looks like a big mess. Lots of overlap of teams with similar values of ODS. So what do we make of it? First off this tells us that even the really good teams can lose early, thus having a really strong ODS is no guarantee. What it also tells us is the higher your ODS value the greater the chances are of making it further into the NCAAT, which hopefully by now comes as no surprise to anyone.

Using 6 Sigma as a tool we will determine the defects (teams that underperformed and over performed) in the round to help us set up a limit for analysis between rounds. This tool is normally used in production with sample sizes in the thousands to millions but we can still make it work here, it just won’t be as accurate the smaller the sample size becomes. We can use it to determine with the standard deviation if there is a statistical difference between the teams that lose in each round.

For those not familiar with 6 Sigma and standard deviations don’t worry, here’s it in a nutshell. It’s determining the variation within a set of data, for us it’s determining the variation between the teams ODS in each round. If the values of a population are close together you’re get a sharp spike, meaning there’s not much difference between the teams. If the values are spread out you’ll have a wide, short hill, meaning there’s a big difference in the teams. If you take +/-1 Sigma from the Average you will have 68% of the population. If you take +/-2 Sigma you now have 95% of the population. If you take +/-3 Sigma you have 99.7% of the population. Let’s look at the teams that made it to the Sweet 16 but lost that game since it has a nice bell curve.


3SIGMA BELLCURVE GRAPH

As you can see the middle of the bell curve has the greatest population where the middle is the mean (average) and then you go 1 sigma (standard deviation) to achieve 68% of the population. If you continue out another sigma then you will have 95% of the population.

SWEET 16


S16 6SIG GRAPH

If you set the lower limit at -2 Sigma (ODS=10.1) and the upper limit at +2 Sigma (27.3) you can see who clearly over achieved and under achieved, totaling 5% of the teams which is only three teams and much too small. If you take the population from -1Sigma (ODS = 14.4) to +1Sigma (23.0) we have 68% of the teams that reached the Sweet 16 but lost. That’s a good percentage of the population and helps us better determine who overachieved in reaching the S16 and who underachieved by failing to move further into the tournament. Setting the upper and lower limits of each round at +/1 sigma will produce the range we desire to determine if a team “should” make that round. Here are the results:

Can we truly be this accurate though in predicting if you’ll make it? I say maybe and here’s why: Overlap. What do I mean? Let’s take a look at just the teams who reached but lost in the R32, S16, and E8.

Each graph (population) effectively fits within the others thus stating the teams on the upper end of the E8 can just as easily end their season in the R32 or S16. The difference is the bottom end of each round, the population range shrinks and thus you have less variation in the round as you progress. So while you have no guarantee of reaching the E8 if you have an ODS value of say 25.0 what can see is if you are below 14.0 you stand almost zero chance. The question becomes is there a statistical difference between these 3 rounds?

If you look closely you can see the R32 Avg and E8 Avg lines have shifted almost a full standard deviation from the S16 Avg. This tells us yes there is a statistical difference between the majority of the population who loses in the R32 and the majority of the population that loses in the E8.

Here we are looking at the 68% population for the 3 rounds (R32, S16, E8) and see there is major overlap with the S16 in relation to the other 2 rounds. If I were to plot the E8 Avg (21.5) we’d see it is more than one standard deviation of R32 Avg. This tells us there is a clear and distinct statistical difference between R32 and E8. The majority of the population (68%) who lost in the S16 overlaps with both the R32/E8 with half being roughly in each. This tells us it may be more accurate to say you are a “First Weekend Team” or a “Second Weekend Team” putting the cutoff between weekends at the S16 Avg (18.7) than it is to say you will reach X round. For now we will still use the +/- 1 Sigma range to determine if you over or under achieved, showing the sigma lines helps us determine if you should take a chance on booking plan tickets and hotel rooms for the next weekend. Thus having an ODS value of at least 18.3 (-1 Sigma E8) should be a “safe” bet on booking those second weekend plans. Also, with roughly a full standard deviation shift between the means in each of those 3 rounds it’s clear there is a statistical difference in the population of each of those rounds and you can say yes, a higher ODS “should” get you further into the tournament.

What about the Final Four Weekend?

Now we’ll look at the third weekend, Final Four, and determine if there are clear differences between those teams within that weekend and from the second weekend.

That’s some overlap. The question is, is there a statistical difference between the rounds?

Again, another busy graph but let’s walk through it and it’ll become clearer. Looking at the Elite 8 Average (Red), E8 +/-1Sig (Black), Final Four Avg (Green), and Runner Up Avg (Purple) we can see there’s not much difference between the 3. The averages for E8 and RU are close to the FF Avg, while there is a statistical difference that difference is small, very small between FF & RU. This tells us not much separates the teams between the E8 and reaching Final Four and even less separates a team losing in the Final Four and one losing in the Championship Game. What we can also see is just how much better the National Champion is than the others who reach those other 3 rounds. In fact the average National Champion is in the top 2.5% of the population who lost in the Elite 8, that’s what I call a pretty clear and distinct difference between them. While not much separates a team who loses in the Elite 8 and those that lose in the next weekend, there’s still a big difference between a second weekend team and a National Champion.

 

What Are The Key Differences Between Rounds?

Well it changes, some rounds (on average) it’s the offenses that are different and some it’s the defense that separates them. Here is a chart of the averages of each round looking at the Offense, Defense, and Spread.

Looking at the table we see a few things:
1. The average difference between teams in the E8, FF, & RU is simply their defense.
2. The difference between the National Champion and the Runner Up is usually offense.
3. Better Offenses AND Defenses will help you go further into the tournament. DUH!!!

SUMMARY

So what can we take away from all of this? First we can see no one is immune from underachieving in the tournament but the higher your ODS the better your chances are for going further into the tournament. Second, the Elite 8 is clearly a step above the first weekend, very close to the Final Four but still has a long ways to go before winning a National Title. The Sweet 16 could easily be categorized as teams that overachieved from the first weekend or underachieved from making it to the Elite 8. It’s less about positioning yourself to reach the Sweet 16 and more about did you overachieve to reach it or did you underachieve in failing to reach the Elite 8 or higher. There are always exceptions to the rules for each round but over time the numbers prove it out.

I hope this helps to show what the differences are between the teams who reach each round and how close or far away a team is on average to making it to the next level and eventually a championship.

About 1.21 Jigawatts

Class of '98, Mechanical Engineer, State fan since arriving on campus and it's been a painful ride ever since. I live by the Law of NC State Fandom, "For every Elation there is an equal and opposite Frustration."

College Basketball

Home Forums Are you ready to win a title? (Part II)

  • This topic has 27 replies, 19 voices, and was last updated 9 years ago by 4in12.
Viewing 25 posts - 1 through 25 (of 28 total)
  • Author
    Posts
  • #77510
    1.21 Jigawatts
    Keymaster

    In Part I we discussed what it takes to be a National Champion and what they all have in common. If you missed the article then I highly suggest going
    [See the full post at: Are you ready to win a title? (Part II)]

    #77511
    BJD95
    Keymaster

    It always has seemed to me that there are frequent visitors to the Sweet Sixteen that qualify as true Cinderellas. Then, the clock strikes midnight, and the blade comes down mercilessly.

    I sit up and take notice when they make the Elite Eight. Sounds like the data back that gut feeling/observation up.

    #77518
    Tau837
    Participant

    State’s current ratings this season:

    Offensive rating: 113.84
    Defensive rating: 92.05
    Spread: 21.79

    If we are using data like this to judge our team’s performance and potential, looks like we are very capable of a S16/E8 run.

    #77527
    1.21 Jigawatts
    Keymaster

    Tau,

    The numbers used are adjusted efficiencies so these are the actual current numbers:

    Adj OE: 111.6
    Adj DE: 97.9
    ODS: 13.7

    So the numbers show we are a clear “First Weekend Team” unless they overachieve.

    #77535
    PittsburghPackFan
    Participant

    I love this SOOOO much!

    Great work Jigawatts. Just phenomenal. Is there still a 3rd entry coming up?

    #77536
    VaWolf82
    Keymaster

    Brilliant Work. Please forgive the edit, but I just absolutely had to highlight the overall conclusions for those whose eyes may have glazed over earlier. I turned the opening sentence in your concluding section to a Section Heading:

    What Are The Key Differences Between Rounds?

    I needed to see the last table as a chart. For anyone else that’s interested, here it is:

    http://www.statefansnation.com/wp-content/uploads/2015/03/Key-Stats.jpg

    #77547
    StateFans
    Keymaster

    Just absolutely unbelievable work!! How do we get this series in front of some national media and get it the kind of kudos and attention it deserves?

    MORE IMPORTANTLY, how can we leverage the work and put it to use ‘Vegas style’ this year!!

    #77548
    1.21 Jigawatts
    Keymaster

    PPF, yes Part 3 will be the last one and it will evaluate and compare coaches and depending on if I have enough time will incorporate the pre-tournament ODS to possibly help be a predictive tool. No guarantees on that.

    VaWolf,

    Please edit away. I’ve been working on this for so long, putting it away and coming back again and again with so many rewrites that it can become just a tad bit confusing with all the edits. A fresh pair of eyes is always welcome. The hardest part isn’t the data, that’s just time-consuming, it’s finding the best way to present the information that makes the most sense to those who aren’t as familiar with statistics. I appreciate the assistance.

    SFN – Parts 1/2 should help when evaluating teams prior to the tournament but it’ll take the pre-tourney data I have for Part 3 to bring that all together since 1/2 have the final numbers and most teams have to play 2-3 ODS points better through 6 games to win it all. Still it’s a useful tool to try and eliminate the obvious ones. Here’s something to help with this year’s picks…Don’t get caught up in the Gonzaga hype. Since 2002 they have reached the S16 a grand total of 2x. They have ended their season in the R32 the last 5 years and 8 out of the last 13. You’d probably be better off not having them past the S16 but I’ll likely have them out at R32.

    #77554
    Wufpacker
    Participant

    Fantastic work Jigsy. I’ve really enjoyed these.

    How do we get this series in front of some national media and get it the kind of kudos and attention it deserves?

    You’re presuming they have an attention span long enough to digest it?

    #77557
    VaWolf82
    Keymaster

    updated graph of last table with State/2014 added

    http://www.statefansnation.com/wp-content/uploads/2015/03/Key-Stats2.jpg

    Shocking Conclusion:
    Neither the offense nor the defense is good enough to go very far.

    #77559
    1.21 Jigawatts
    Keymaster

    Va, If I have the time I’ll get the R64 numbers done to have a complete historical look at the tournament but it’ll likely be awhile since Part 3 will take priority. Anyone who has watched this team knows putting together more than 1 good game is a stretch for this squad.

    #77573
    wufpup76
    Keymaster

    Impressive work. Very thorough. Fantastic analysis and fun to read – truly well done.

    StateFans – if you’re interested in media exposure, I’d suggest contacting the respective ‘Bubble Watch’ folks at various media sites. For example, Eammon Brennan at Espn – etc. They’re nerds like us when it comes to this kind of stuff.

    #77576
    wufpup76
    Keymaster

    Since last season’s UConn team was one of the exceptions to the rule, I wonder if there’s a way to differentiate their ODS from the regular season and AAC tournament and the delta in the NCAA tournament alone? I wonder if there’s a specific spike in the NCAA tourney games?

    Eh, no matter … they were a 7-seed and Napier went off for them.

    #77586
    ncsu12engr
    Participant

    I love what you are doing here Jig. Have you thought about analyzing all the underachieving teams and seeing if there is a common denominator? For instance, you mentioned Gonzaga as typically being an early out. I just looked at kenpom and their ODS is currently 26, which is 6th best in the country. However, their SOS is 85th which is significantly lower than all other teams around them. Is this something that we could consistently look at to say that their ODS might be higher because of a lack of quality opponents?

    #77587
    doug
    Participant

    I believe in the every 32 years formula.

    #77591
    ncsu1987
    Participant

    Absolutely amazing stuff, Jiggs. Fun to read, fun to digest – first read was a couple of hours ago, and I’m still digesting this. Kinda like listening to Robin Williams do standup – every time you repeat it, you learn something new…seriously, thanks.

    I believe in the every 32 years formula.

    ^I like the way you think…

    #77593
    charger17
    Participant

    Incredible stuff. Is there any point in cross-referencing the ODS info with a team’s tourney seeding? In other words, would there be a telling difference if two teams with similar ODS’s were ranked as a 5 seed vs. a 6 seed? Could this tell you if you’re a 5 seed with an ODS of 19, you have a much better chance of E8 appearance than a 6 seed with an ODS of 19?

    Just curious.

    #77594
    TheCOWDOG
    Moderator

    Exactly how fluid is the adjODS, compared to the “raw” ODS?

    #77598
    wufpup76
    Keymaster

    Is there any point in cross-referencing the ODS info with a team’s tourney seeding? In other words, would there be a telling difference if two teams with similar ODS’s were ranked as a 5 seed vs. a 6 seed? Could this tell you if you’re a 5 seed with an ODS of 19, you have a much better chance of E8 appearance than a 6 seed with an ODS of 19?

    ^I think this would be an interesting contrast and compare. Someone brought up Gonzaga’s SOS vs. ODS as a possible correlation to them flaming out in the first or second weekend with high seeds (or low seeds – your perspective). I wonder if there’s an objective measure to SOS that would correlate (directly) to a team’s overall ODS.

    #77603
    McCallum
    Participant

    You are all on drugs.

    McCallum

    1 AD (after dean)

    #77613
    1.21 Jigawatts
    Keymaster

    Since last season’s UConn team was one of the exceptions to the rule, I wonder if there’s a way to differentiate their ODS from the regular season and AAC tournament and the delta in the NCAA tournament alone? I wonder if there’s a specific spike in the NCAA tourney games?

    Eh, no matter … they were a 7-seed and Napier went off for them.

    KP now has pre-tournament data and I hope to get it included in part 3. If not I’ll work on it for next year when I revisit it before the NCAAT. I have looked at the champions and most increase by 2 points through the 6 wins but I think last year UConn increased by 4. Surprisingly that wasn’t the biggest, is was one of the Florida years that increased the most by around 7 points.

    I love what you are doing here Jig. Have you thought about analyzing all the underachieving teams and seeing if there is a common denominator? For instance, you mentioned Gonzaga as typically being an early out. I just looked at kenpom and their ODS is currently 26, which is 6th best in the country. However, their SOS is 85th which is significantly lower than all other teams around them. Is this something that we could consistently look at to say that their ODS might be higher because of a lack of quality opponents?

    Incredible stuff. Is there any point in cross-referencing the ODS info with a team’s tourney seeding? In other words, would there be a telling difference if two teams with similar ODS’s were ranked as a 5 seed vs. a 6 seed? Could this tell you if you’re a 5 seed with an ODS of 19, you have a much better chance of E8 appearance than a 6 seed with an ODS of 19?

    Just curious.

    I’ve thought about a lot of things over the last year of working this series, its just a matter of time to be able to invest. I needed to complete what I had or another year would have passed. Using KP’s adjusted efficiencies helps eliminate any SOS bias but that doesn’t help said team when they haven’t competed at a much higher level for most of the season. Remember VaWolf’s talk about mid-majors gaming the system to have higher SOS but never really playing those top teams? That works well to get you in the NCAAT but you can’t hide forever.

    Exactly how fluid is the adjODS, compared to the “raw” ODS?

    Like tempo, I average each team’s efficiency by game. The other way to do this would be to take a team’s total points on the season and divide it by total possesions. But this gives some games more weight than others depending on the number of possessions in a particular contest, and I don’t like that. Also, I only use games involving two D-I teams.

    The raw numbers are computed from the data contained in a box score. But then there’s the matter of adjusting for competition. The “adjusted” numbers (AdjO, AdjD as AdjT) are the results of these calculations.

    AdjO – Adjusted offensive efficiency – An estimate of the offensive efficiency (points scored per 100 possessions) a team would have against the average D-I defense.

    Basically he’s compensating for SOS by adjusting to the average team. Its his secret sauce formula so I don’t know his equations but it eliminates racking up great efficiencies against weak competition and not being penalized when playing top teams.

    #77623
    TheCOWDOG
    Moderator

    Thanks, Jigsie.
    That co-effeciency means alot.

    Waiting on the 3rd.

    #77626
    Primewolf
    Participant

    Great work. I assume a teams rating is based on its full season. What about trends and standard deviations. That is, a Jeckle and Hyde team could have a bimodal rating and perhaps play at the second mode.

    Pls send to Gott. It is the difference between O and D that determines success. Our D can be pitiful, mainly because it is a lot of hard work and G can’t seem to get it from our guys on a consistent basis.

    #77630
    choppack1
    Participant

    Jigs just good stuff.

    I think it reinforces what a lot of us believe and common sense. You can’t suck or even be mediocre on one end of the floor and win it all. But awesome stuff. Interested in seeing how coaches stack up – of course, you won’t have #s for v.

    #77631
    Pack85EE
    Participant

    I believed in the every 9 year formula in 92, I even felt maybe in 2001 it would pop up again and 92 was just a miss. I had abandoned all hop by 2010. Maybe in 2019. I’m hopeful.

    But great work Jigs. I think it will help me appreciate any success we have in the tourney and know that with each round beyond our predicted mean it’s just a matter of time – so enjoy. But I was on campus in 83 so I will always believe in the impossible.

Viewing 25 posts - 1 through 25 (of 28 total)
  • You must be logged in to reply to this topic.