Please Help Me

Okay, that was not a last ditch plea for help from a desperate man. Here is what I would like input on. A week or two ago someone linked an article here about KU being the most dominant program since Self has been at Kansas. It was obviously a nice article for us Jayhawks but the premises were not really sound in my opinion. It was really just an article about KU's home record and Allen Fieldhouse magic. It prompted me to want to put together a matrix of various metrics by which to measure a teams dominance. So the help I need is this; if you are measuring which team is dominant over a decade or in this case a 13 year period (since Self came to KU) what metrics should be used and how do you weight them. Here is what I have built so far. Before I publish it I wanted to get your input so I can refine it as necessary to be most valid and meaningful.

1. Wins/Losses -- I have, of course, all the records for all teams so I my scoring was 2 points for a win and -1 for a loss. A 30-6 season would net 54 points (30*2-6*1)
2. AP Poll -- I have all the AP polls in a database from 1949 through this week. My scoring is 2.5 points for a No. 1 ranking for each week in the poll, stepping down by 0.1 down to 0.01 points for a 25th place ranking. So if you were ranked 1st all 18 weeks of the AP rankings in a season you would get 18*2.5=45.
3. NCAA Tourney Wins -- My scoring goes 5 points for a round 1 or playin round loss, 8 points for a round 1 win, 12 points for a round 2 win, 17 points for a round 3 win, 23 points for a round 4 win, 30 points for a final four win and 40 points for a national championship. So a National championship would net 130 points (40+30+23+17+12+8 ), losing in the NC game would net 90 point, losing in the Final Four would net 60, etc.
4. Conference Title - 25 points for a conference title (outright or shared)
5. 1st team All-American selection is 25 points and 2nd team AA selection is 15 points

Other options
1. Academic All-Americans
2. NBA draft selections
3. Conference Tournament titles
4. Annual Statistical rankings (Scoring average, defensive scoring average, rebounding average, etc)

I can tell you that based on the system I have already built (above) two teams are way way head and shoulders above the rest. You can guess who those teams are. And neither is named Tar Heels or Wildcats. Not is either named Huskies, Spartans, Orangemen or Gators. Bear in mind that how we tweak these metrics can turn the tide for the Jayhawks to be No. 1 or No. 2. I would like to make it something any school would agree is reasonable. For instance if we made conference titles worth 100 points clearly that would be intentionally swaying toward KU. I really want an objective set of metrics.

Help please.

I do not like including the AP ranking as that is much more a popularity contest than a true indicator of a teams strength. I like to award teams for place in conference besides just winning the conference. I am not sure how you would adjust that for strength of conference other than maybe a combo of non con record and strength of schedule. I like using some of the deeper team metrics like, offensive efficiency - defensive efficiency, but I think they should be weighted by a strength of schedule weighting. I think the AP ranking is already in the formula in some of the other things like strength of schedule so I think it should be left out of the evaluation.

I think this is a really cool project.

I agree with your assessment of the previous article.
Of course any list of metrics is subjective, but my only change would be dropping your #5 (All-Americans). Glad to see your value for NCAA tourney.
I know your focus presently is the Self years. However, how would you measure NCAA games in earlier years when a team only played 4 or 5 games to win NC?
Surprised Huskies aren't at the top with 3 NC's. Look forward to seeing your final product.

I know this would be extremely difficult and time consuming, but I'd limit the win-loss component to wins and losses against teams ranked in the top 50 at the end of each season.

A victory over UMKC should not be the same as a victory over OU or Kentucky etc. Losses have the same issue. A true ranking would have to take this into account.
I also think consistency should have a multiplier impact. How many times did they make the NCAA tournament in that period. Same with 30 win seasons or conference championships. Consistency might boost scores by a factor of 1.1 -1.5.

Very interesting.

I tried to type this up earlier, but somehow it didn't post (probably my error).

It sounds to me like you're wrestling somewhat with the idea of what constitutes "most dominant". In that regard, I'd define it as follows: Take two teams from any era, get a time machine and get them both on the same neutral court. Throw the ball up and see who wins.

In that regard, it gets a little more simple. First, you lose all of the extraneous individual awards like academic all American. Just doesn't matter in that regard. You also lose whether a kid is drafted (and when). Laettner was a great college player, but not so much at the next level. Same with pretty much anything individual since teams play teams.

So really all that matters is W/L and SOS. With that in mind, you could pick whatever computer ranking you like best (or combination thereof) and you're most of the way there. The next issue is equalizing from year to year since a 95 rating THIS year isn't necessarily the same as a 95 rating LAST year. So the key is finding a consistent way to balance this out.

To do this, I'd look at offensive and defensive averages for the top 25 in each year's computer ranking. If the average score of the top 25 is 68 points and average defense gives up 62 points in Year A and it's 71 and 63 in Year B, you should (theoretically) be able to come up with standard calculation that a.) takes in all of the data over all time, and b.) can be applied to "equalize" each year. I'm thinking if you did this, you'd learn that a 95 rating in Year A is the equivalent of a 96 rating in Year B....therefore Year B was the tougher year and the ratings are adjusted accordingly (kind of like 3 points for home court advantage).

Now, if you want to go deeper, I'd also think about weighing each part of the schedule. I'd adjust the pre-conference season and conference tournament downward slightly. Conference regular season and NCAA tourney would be bumped slightly upward.

Again, this is all predicated on the idea that "most dominant" means two teams, mano-a-mano, who cares about individual and extraneous info...who's gonna win.

EDIT: Forgot the most important part! If you were able to do this (essentially come up with a computer rating for every team over a period of time) you could then either use a sum of all the ratings, or an average of them to rate the "programs".

I like that idea. The only question is whether it's more (or less) fair to penalize teams that, through no fault of their own) play in a crappy conference. Wasn't it St. Joe's a few years back that was running the table and being criticized because they played in a very weak league. I remember then, and other years, where this kind of debate bubbles up at seeding time.

I laughed out loud to point where my roommate asked me what the h*ll's so funny. I'm still smiling as type this (and see it on the bottom of your message). That's great stuff!

Glad you enjoyed it. I went to edit and it said "reason for edit" and wasn't sure if it was a required field. Not sure if it would appear, but I'm also not sure that anybody needs to justify editing since most times it's just correcting something. But yeah, it's a good catch-all excuse.