Rankings Questiom

CleverUsername1

11-man fan
I was looking at historical rankings, and I noticed a trend. It seems like slowly the ratings of the top teams are increasing. If you look at the mid 2000s the top teams were normally rated in the mid to high 200s, but now a 250 rated team would be somewhere in the middle of the pack in terms of rankings. Since the 2010 season there have been 18 teams rated above 400, and 17 of them came last season with the remaining one being the 2014 Crowell team. My question is, what's causing this? Changes in the algorithm? Increased quality of football? An increase in bottom dwellers that pump up offensive numbers and win totals for average teams that then make the top teams look better by comparison when they beat the previously mentioned average teams? I know the toy isn't the end all be all and isn't necessarily accurate, but in theory if it's the same system it should produce similar results every season, meaning that the top teams every year should have approximately the same power rating. Or maybe I'm just overthinking it and it's the drunken monkey's fault?
 
Grasshopper, this could require clarification from the man behind the curtain. If Dorothy can get the Great Oz's attention. I think your question is beyond dartification by even a sober monkey.
 
290 teams now versus how many 10 years ago? Plus the Toy is not a static creation. It is dynamic. Some say it might actually be sentient, which is why it consistently has Richland Springs ranked #1. I don't believe it myself, but there are rumors out there.
 
rainjacktx":3hxfvd60 said:
290 teams now versus how many 10 years ago? Plus the Toy is not a static creation. It is dynamic. Some say it might actually be sentient, which is why it consistently has Richland Springs ranked #1. I don't believe it myself, but there are rumors out there.
I understand that the increase in teams could inflate the numbers. Back in the mid-2000s there were fewer teams and most of them were at least somewhat competitive. Now there are a bunch of new teams that are struggling to get up off the ground and find success in six-man. But that doesn't explain the fact that from 2010-2015 there was only one team with a rating above 400, but last season there were 17 teams ranked above 400. To put it in perspective, the 2015 #1 ranked team (Richland Springs) would be ranked 19th last season based on their power rating. Something had to have caused the outlier. Did we witness a historic level of elite teams last season, a historic number of bad teams that inflated the numbers, or just a malfunction in the system?
 
the Ratings numbers mean NOTHING year-to-year.

In 1993, the system was created to have a theoretical perfect team be 100. If was a zero-sum game. If you want to understand this, I suggest Lester Thurow's book, The Zero Sum Solution or The Zero Sum Society. He was the Dean of the MIT Sloan School of Management. I created a system which used some of the ELO rating system, modified to meet this philosophy.

Early on, I realized that this made no sense. Especially because I was also trying to make the spreads easily interpretable from the ratings. This didn't work when Team A 45's Team B and Team B 45's Team C and Team C 45's Team D. Now you have Team A somewhere between 45-135 points better than team D.

One of main things I wanted to do was rank EVERY team and I certainly don't want teams to drop below 0, if at all possible.

So you can see the dilemma.

Also, as the system gets bigger and bigger with the number of schools, I have had to implement better ways of trying to keep everything in perspective. The system is much more robust than before and involves many more machine learning philosophies than before.
 
"In 1993, the system was created to have a theoretical perfect team be 100. If was a zero-sum game. If you want to understand this, I suggest Lester Thurow's book, The Zero Sum Solution or The Zero Sum Society. He was the Dean of the MIT Sloan School of Management. I created a system which used some of the ELO rating system, modified to meet this philosophy."
Or you could just ask Goob............
 
granger":14l6o0qe said:
the Ratings numbers mean NOTHING year-to-year.

In 1993, the system was created to have a theoretical perfect team be 100. If was a zero-sum game. If you want to understand this, I suggest Lester Thurow's book, The Zero Sum Solution or The Zero Sum Society. He was the Dean of the MIT Sloan School of Management. I created a system which used some of the ELO rating system, modified to meet this philosophy.

Early on, I realized that this made no sense. Especially because I was also trying to make the spreads easily interpretable from the ratings. This didn't work when Team A 45's Team B and Team B 45's Team C and Team C 45's Team D. Now you have Team A somewhere between 45-135 points better than team D.

One of main things I wanted to do was rank EVERY team and I certainly don't want teams to drop below 0, if at all possible.

So you can see the dilemma.

Also, as the system gets bigger and bigger with the number of schools, I have had to implement better ways of trying to keep everything in perspective. The system is much more robust than before and involves many more machine learning philosophies than before.
I didn't read the book, but I googled "Zero Sum Solution" and I think I understand. If you add the score differentials of the entire state, it should equal zero. At first there were less teams, so less points in the system. But as more teams began to play sixman, it created more games and thus more points to be added to the system. These points then began trickling up to the top of the rankings like inverse Reaganomics, inflating the top teams' ratings while plummeting the ratings of the bottom feeders which the aforementioned top teams were "taking" points from whenever they won. But like you said you didn't want teams with negative points, so the system had to be altered over time. Is that correct, or am I off worse than those darn drunken monkeys throwing darts?
 
51eleven":2h658kn0 said:
I'm not good with math. "inverse Reganomics"? So it's kind of a trickle down thing? Or something to do with logarithms?
Granger's the expert not me. But from what I understood, in a zero sum system the sum of all teams' point differential's should equal zero. There can be no give without take and vice versa. When a team wins and adds to their score differential, those same points are taken away from the opponent's score differential. Not to put words in Granger's mouth, but if I understand correctly the original rankings were guided by these principles, but have been tweaked over time to keep up with the amount of new teams and to prevent teams from having negative ratings. When I said "inverse Reaganomics", I simply meant that the points trickle upwards from the lowest teams to the teams at the top of the rankings, the opposite of true Reaganomics. I don't know about logarithms though. I barely escaped college algebra with an A and I still can't wrap my brain around those stupid things.
 
It was more like whatever goes into the system must come out. So if a team went up 5 points, then others lost a total of 5 points.

Now remember, this isn't the case now and hasn't been for a long time. Now the system expands both directions to derive spacing between all of the teams, so the spreads can be easily calculated without scaling.
 
Good thing about Granger's system is that it takes all discussion and controversy out of the mix...................
Right?.................
 
smokeyjoe53":3jeq3ljx said:
Good thing about Granger's system is that it takes all discussion and controversy out of the mix...................
Right?.................


absolutely. wink-wink-nudge-nudge. say no more.
 
oldfat&bald":1nqt50qc said:
Really, it’s been a while since I’ve seen a “You inhale” post directed at Granger. I guess the internet’s down in Aquilla.
I believe that particular aspect of Granger's aficionados has shifted ever so slightly 30 miles to the Northeast.................
 
Back
Top