Thanks for doing this again this season Satchmo! Love looking at this stuff.
clownassassin
offline
offline
Originally posted by Satchmo-n-Dizzy
Originally posted by clownassassin
What data or formula are you using?
The algorithm is actually very simple. It uses a comparative analysis of scores across all games to arrive at factors for each team (expressed as a %). The analysis will easily determine the SoS that a team has played to date vs what said team will play over the remainder of the season. It becomes more and more accurate as the season wears on. However, it has no ability to recognize or account for significant changes to rosters (gutting, recruiting) that may occur during the season.
There is far too little data to arrive at a consensus for week 16 games against the other conference, so it is not unusual to see the algorithm arrive at considerably different conclusions based on the perspective. But within a given conference, the results has been impressive thus far.
The concept...
A team 'A' that averages 56 PF and 6 PA is to play a team 'B' that averages 21 PF and 4 PA. How does one determine the better team going in? By looking at those numbers relative to the teams played to date. Team A's average opponent scores 36 ppg while allowing 28. Team B's average opponent scores 24 and allows 7. That means that Team A scores 200% of the norm allowed against its opponents while allowing 17%. Team B also scores 300% but allows only 14%.
Simplified, the algorithm would take Team A's 200% and multiply it by the 4 PA by Team B (8), then take the 17% and multiply it by 21 (3.5). For Team B it would take 300% and multipy it by 6 (18), then the 14% multipled by 56 (8). The algorithm would then resolve these numbers by taking the mean of each result = Team A will score... (8 + 8) / 2 = 8. Team B will score... (3.5 + 18) / 2 = 11.75. The projection picks Team B.
In practice, the algorithm will look at each game played by a team independently. So upward (or downward) trends will not be tempered by league averages. The determined percentage is, in essence, an average. And, of course, the projected score is and average as described above. However, the numbers that drive the analysis are much more volitile. This allows for greater accuracy based on trends and will typically gain in strength during the season.
Thanks for the detailed explanation. I wish there was a way to make a more realistic projection 5 games into the season. Duffman projected to give up 15 points on the season? They have given up 6 so far, but have yet to face Ktown, Mato, or Medellin. By these projections those 3 teams will score a FG each and Duffman shuts out everyone else. It's not impossible for that to happen, but it's very unlikely.
I'm not a math expert but I'll try to figure out a way that projects with more weighted towards strength of schedule and results vs common opponents. Now that I just typed that, I realized that I'm either way too busy or too lazy to do that.
Thanks for doing the projections.
Originally posted by clownassassin
What data or formula are you using?
The algorithm is actually very simple. It uses a comparative analysis of scores across all games to arrive at factors for each team (expressed as a %). The analysis will easily determine the SoS that a team has played to date vs what said team will play over the remainder of the season. It becomes more and more accurate as the season wears on. However, it has no ability to recognize or account for significant changes to rosters (gutting, recruiting) that may occur during the season.
There is far too little data to arrive at a consensus for week 16 games against the other conference, so it is not unusual to see the algorithm arrive at considerably different conclusions based on the perspective. But within a given conference, the results has been impressive thus far.
The concept...
A team 'A' that averages 56 PF and 6 PA is to play a team 'B' that averages 21 PF and 4 PA. How does one determine the better team going in? By looking at those numbers relative to the teams played to date. Team A's average opponent scores 36 ppg while allowing 28. Team B's average opponent scores 24 and allows 7. That means that Team A scores 200% of the norm allowed against its opponents while allowing 17%. Team B also scores 300% but allows only 14%.
Simplified, the algorithm would take Team A's 200% and multiply it by the 4 PA by Team B (8), then take the 17% and multiply it by 21 (3.5). For Team B it would take 300% and multipy it by 6 (18), then the 14% multipled by 56 (8). The algorithm would then resolve these numbers by taking the mean of each result = Team A will score... (8 + 8) / 2 = 8. Team B will score... (3.5 + 18) / 2 = 11.75. The projection picks Team B.
In practice, the algorithm will look at each game played by a team independently. So upward (or downward) trends will not be tempered by league averages. The determined percentage is, in essence, an average. And, of course, the projected score is and average as described above. However, the numbers that drive the analysis are much more volitile. This allows for greater accuracy based on trends and will typically gain in strength during the season.
Thanks for the detailed explanation. I wish there was a way to make a more realistic projection 5 games into the season. Duffman projected to give up 15 points on the season? They have given up 6 so far, but have yet to face Ktown, Mato, or Medellin. By these projections those 3 teams will score a FG each and Duffman shuts out everyone else. It's not impossible for that to happen, but it's very unlikely.
I'm not a math expert but I'll try to figure out a way that projects with more weighted towards strength of schedule and results vs common opponents. Now that I just typed that, I realized that I'm either way too busy or too lazy to do that.
Thanks for doing the projections.
Jcdawg23
offline
offline
Strength of schedule would only change from team to team if a team was gutted, added a new starter, or by the alpha vs zeta games at the end of the season. We all pretty much have the same schedule.
Originally posted by clownassassin
I'm not a math expert but I'll try to figure out a way that projects with more weighted towards strength of schedule and results vs common opponents. Now that I just typed that, I realized that I'm either way too busy or too lazy to do that.
Thanks for doing the projections.
There isn't a methodology that can account for strength of schedule before enough games are played. After game 1, this projection is the equivalent of throwing darts at a betting sheet. By week 5, it probably has about 40 to 50% of the data it needs to make analytical comparisons. Once we get to mid-season, it will really start cooking.
Feel free to search for a better way to deal with SOS but, I am a math expert, and your search will be in vain.
I'm not a math expert but I'll try to figure out a way that projects with more weighted towards strength of schedule and results vs common opponents. Now that I just typed that, I realized that I'm either way too busy or too lazy to do that.
Thanks for doing the projections.
There isn't a methodology that can account for strength of schedule before enough games are played. After game 1, this projection is the equivalent of throwing darts at a betting sheet. By week 5, it probably has about 40 to 50% of the data it needs to make analytical comparisons. Once we get to mid-season, it will really start cooking.
Feel free to search for a better way to deal with SOS but, I am a math expert, and your search will be in vain.
Last edited Jun 22, 2008 18:48:55
Originally posted by The Duff Man
I am thrusting my pelvis.....
we noticed.
I am thrusting my pelvis.....
we noticed.

Jcdawg23
offline
offline
Originally posted by The Duff Man
I am thrusting my pelvis.....
.....it...it...IT NEVER STOPS!!!! AHHH!!!!!
I am thrusting my pelvis.....
.....it...it...IT NEVER STOPS!!!! AHHH!!!!!
Provider
offline
offline
These projections don't factor in the most important stat: heart, of which the Duff Men have zero and the Donkies have a little more like Captain Planet.
Jcdawg23
offline
offline
Originally posted by Provider
These projections don't factor in the most important stat: heart, of which the Duff Men have zero and the Donkies have a little more like Captain Planet.
HAHA!!
These projections don't factor in the most important stat: heart, of which the Duff Men have zero and the Donkies have a little more like Captain Planet.
HAHA!!
Serge818
offline
offline
Originally posted by Provider
These projections don't factor in the most important stat: heart, of which the Duff Men have zero and the Donkies have a little more like Captain Planet.
Heart is for suckers
These projections don't factor in the most important stat: heart, of which the Duff Men have zero and the Donkies have a little more like Captain Planet.
Heart is for suckers
diego.mangueira
offline
offline
Originally posted by James128
Nottingham will not go unbeaten i promise you that. They should have 2 or 3 loses. Cheetahs will run the table... World slayers are not going 8 and 8 i promise you that. We shut Nottingham down but lost the TO battle 8 to 1 which is a credit to there D. We still lost by 8 points They can be beaten. They scored 3 points that weren't off turnovers. Not to mention there only TD was on a special team fumble. I know their our a couple of teams they have beaten so far that wouldn't mind seeing them again in the post season including us.
Come on...
The Outlaws have a kicker form some reason. Why run up the score with TDs? Just let the lowest level starter make all the points!
Nottingham will not go unbeaten i promise you that. They should have 2 or 3 loses. Cheetahs will run the table... World slayers are not going 8 and 8 i promise you that. We shut Nottingham down but lost the TO battle 8 to 1 which is a credit to there D. We still lost by 8 points They can be beaten. They scored 3 points that weren't off turnovers. Not to mention there only TD was on a special team fumble. I know their our a couple of teams they have beaten so far that wouldn't mind seeing them again in the post season including us.
Come on...
The Outlaws have a kicker form some reason. Why run up the score with TDs? Just let the lowest level starter make all the points!
Jcdawg23
offline
offline
Originally posted by Serge Perez
Originally posted by Provider
These projections don't factor in the most important stat: heart, of which the Duff Men have zero and the Donkies have a little more like Captain Planet.
Heart is for suckers
Kicking is for suckers!
Originally posted by Provider
These projections don't factor in the most important stat: heart, of which the Duff Men have zero and the Donkies have a little more like Captain Planet.
Heart is for suckers
Kicking is for suckers!
rabideasterbunny
offline
offline
I miss having these to look at over in AAA#1. They are pretty accurate as they did predict a Monkeys loss last season.
Don't let Satchmo fool you guys with his explanation of how he figures these things up. I don't know the exact method, but it involves chicken blood and voodoo magic!
Don't let Satchmo fool you guys with his explanation of how he figures these things up. I don't know the exact method, but it involves chicken blood and voodoo magic!
rabideasterbunny
offline
offline
Originally posted by The Duff Man
He was going to use virgins....but the Duff Men "used" them first.
OH YEAH!
ALL the virgins?
He was going to use virgins....but the Duff Men "used" them first.
OH YEAH!
ALL the virgins?
You are not logged in. Please log in if you want to post a reply.





























