User Pass
Home Sign Up Contact Log In
Forum > USA AAA Leagues > USA AAA #2 > WEEK 8 POWER RANKINGS
Page:
 
BobJoeJim
offline
Link
 
Originally posted by Dravz
I know you have to play math politics so I understand you putting a team we've beaten by 16 points in the #1 spot -- I enjoy the subtle and understated manner in which you put the Yell at the true #1 spot. Good job, BJJ! The Rebel Yell's dream of a bias-free ranking may never actually come to pass, but things like this give us great hope for the future.


wat
 
Mornacale
offline
Link
 
Originally posted by BobJoeJim
Originally posted by Dravz

I know you have to play math politics so I understand you putting a team we've beaten by 16 points in the #1 spot -- I enjoy the subtle and understated manner in which you put the Yell at the true #1 spot. Good job, BJJ! The Rebel Yell's dream of a bias-free ranking may never actually come to pass, but things like this give us great hope for the future.


wat

He's trying to say that we are the best team in the league. (Because we are.)
 
Mofo
offline
Link
 
Originally posted by dfitzo53
Originally posted by lunchbox6186

WOW Cincinnati Owls #1 after a big game against J'Ville. Tough competetion at both ends. Big game against Fairfox coming up Bring the heat boys. We got something some plans for ya. But on a serious note, Looking forward to rest of the season and the Playoffs


I think last week showed that you can't count us out. I'm looking forward to a good game.


No one on our team is counting you out, we know this is going to be a tough game.
 
whodey08
offline
Link
 
Definitely a different team from the last Fairskins that I scouted! You guys look good! Beating Jacksonville and the Montanas is huge! Congrats guys!

Like us...you guys have worked hard to put together a successful francise

WhoDey
Cincinnati Owls
 
PP
offline
Link
 
Originally posted by BobJoeJim


Yeah, I did a double take on that as well. Their two losses came while unboosted though, and after boosting they (for the moment) became the highest level team in the league by a good margin. They then proceeded to back it up with a very impressive win today. Despite the two losses, I think it's quite reasonable to say that at this exact moment, Green Bay is the best team in the East, and that's what the rankings are trying to measure. Now, when other teams start boosting too that'll be a different story...


Don't get me wrong. I'm extremely relieved that we won these last 2 games and are starting to almost play the type of ball I'm used to again. Still, after the way we got our asses handed to us the 2 previous games, boosted or not, I don't think we deserve the top slot. Boosting doesn't make up for getting blown out, IMO. Sure, it may take a 3-14 game and make it 17-10 instead, but it doesn't mean as much as those two thrashing do, IMO.

I'm on a family vacation and just haven't been on much. So, belated thanks Palmer and I'll be REAL interested to see how things shake out against you guys too nathbear

 
BobJoeJim
offline
Link
 
Originally posted by Mornacale
Originally posted by BobJoeJim

Originally posted by Dravz


I know you have to play math politics so I understand you putting a team we've beaten by 16 points in the #1 spot -- I enjoy the subtle and understated manner in which you put the Yell at the true #1 spot. Good job, BJJ! The Rebel Yell's dream of a bias-free ranking may never actually come to pass, but things like this give us great hope for the future.


wat

He's trying to say that we are the best team in the league. (Because we are.)


I'm going to stick with "wat".

Originally posted by PP

Don't get me wrong. I'm extremely relieved that we won these last 2 games and are starting to almost play the type of ball I'm used to again. Still, after the way we got our asses handed to us the 2 previous games, boosted or not, I don't think we deserve the top slot. Boosting doesn't make up for getting blown out, IMO. Sure, it may take a 3-14 game and make it 17-10 instead, but it doesn't mean as much as those two thrashing do, IMO.


Yeah, I'm not sure if you should really be #1 right now or not. Certainly in a "performance ratings" sense you don't deserve to be there based on what you've done so far this year. As for power rankings, which are just trying to measure how good the teams are right now, you might currently deserve the spot thanks to your recent boosts, but it's probably pretty close. I will say that one game is always a small sample size, with a lot of variance. I highly doubt you were "actually" 32 and 34 points worse than Cincinnati and Las Vegas when you played them. Probably you were more like 10-15 points worse, and variance went against you as well.

People love to say that levels mean nothing, and that this game is all about builds. While it's certainly true that good builds are incredibly important, and a team with good builds can certainly be better than a higher level team with bad builds, the fact remains that levels do matter. If two teams have identical, or even just equally good, builds then it's usually levels that will decide things. In a league like ours, where the top level players are level 30 (my apologies to our one level 31 guy), a three level boost for an entire team I think generally makes that team 20-25 points better. I'm not sure if your boost covered every player across your whole roster or not, though it looks like it didn't quite. I'm guessing by boosting you improved your team by about 15 points per game. If you were to replay Cincinnati and Las Vegas again right now, I think it would be reasonable to consider you small favorites (with a single digit spread), which is right where the rankings have you. I think #1 is reasonable for your team in a power ranking right now. Of course other teams in the league have the option to boost just as you did, and when they do they'll negate the temporary edge you have, and you'll certainly lose the #1 spot. I can't feel bad about you sitting there right now though.

I know I'm rambling a little, this post is as much a way for me to think about things because I just genuinely like ratings, and thinking about how ratings systems work (which is why I'm doing this in the first place), as it is an attempt to convince anyone else of anything. I have other football ranking formulas that I've developed besides the one here, that I use for college football. The distinction between "power rankings" and "performance rankings" is one that I've spent a lot of time thinking about, because of the BCS. A power ranking is designed to measure exactly how good teams are at the moment of the ranking. It doesn't matter how they've done, or what their record is. You can have a 4 loss team ranked ahead of an undefeated team, if the data suggests that the 4 loss team would win a game between the two. A performance rating (which is what I think the BCS should be... but let's not get started on that tangent), meanwhile, will (or should) have the undefeated team rated higher.

Personally, I find power rankings more interesting. Mostly I suppose it's because I enjoy sports betting, and obviously power rankings are much more effective as predictive tools, and so are more useful in assisting bet calculations. I have both a power and a performance rating system for college football that I keep every season, and I won't deny that the power rating formula is the one I've put a lot more work into, and is much more complex. Of course since those formulas are for real life, where there are no "levels" to look at, they are based on actual results (the performance rating completely ignores scores, and looks entirely at wins and losses, while the power ranking completely ignores record and looks only at scores). When I first started doing power rankings for GLB (Season 2, in Canada AA), I thought about using one of my existing formulas, but decided not to. I think that a ranking formula based on results just isn't good enough for GLB, where teams can change SO drastically mid-season. In real football a team is generally about the same all year long, and ranking them is just a matter of waiting for them to play enough games that you can get the statistical noise of variance worked out, and figure out what the results are actually telling you. In GLB though, teams can improve by 20 points from one day to the next by boosting. Playoff caliber teams can be sold mid-season, get gutted, and lose the rest of their games by 200. A traditional results-based ranking system simply has no way to account for things like this.

The reason I designed this rating system the way I did isn't because I think levels are the be-all end-all measure of team caliber. Obviously they aren't. I designed this system to be level-based because it made it dynamic. Every time I calculate the ratings I start from scratch and recalculate the base rating off of the current rosters, adding the accumulated adjustments on top afterward. This way, when a team boosts they immediately jump, and when a team guts they immediately drop, allowing the ratings to be much more accurate and effective as a predictive measure, at all points throughout the season. Additionally, the formula can apply in week 1. Obviously since levels don't tell you everything, the adjustment factors need time to accumulate before things really start to get accurate, but there's no question in my mind that, while not perfect, my rankings predict week 1 results better than any performance based rating system It's true that there are downsides to the route I chose. People frequently object to my rankings due to an over-reliance on levels, and while I generally disagree with the way in which these objections are phrased (I really don't think my rankings "suck" or are "worthless", lol), I certainly do agree with the sentiment. I do rely more on levels than I'd like, but so far haven't figured out a way to reduce that reliance while still maintaining the basic framework of the system. And I like the framework, because of the dynamic factors I just mentioned.

Huh, I guess this post turned into something of a manifesto. I didn't mean for that to happen, lol. I've had a lot of these thoughts bouncing around though, and I'm responding not just to things that people on this board have said, but also things people have said in the other leagues where I've posted rankings, in both this and previous seasons. It felt good to get some of it down in words. Hopefully if anyone actually takes the time to read the whole thing, they won't be too bored
 
nottom
offline
Link
 
tl:ra
 
Dravz
offline
Link
 
Will these be updated before tomorrow's games?
 
Reggie Smash
offline
Link
 
Originally posted by Dravz
Will these be updated before tomorrow's games?


He's trying to say "Will these be updated with our team at the top of the Rankings by tomorrow's game?"
 
Link
 
Originally posted by BobJoeJim
Originally posted by Mornacale

Originally posted by BobJoeJim


Originally posted by Dravz



I know you have to play math politics so I understand you putting a team we've beaten by 16 points in the #1 spot -- I enjoy the subtle and understated manner in which you put the Yell at the true #1 spot. Good job, BJJ! The Rebel Yell's dream of a bias-free ranking may never actually come to pass, but things like this give us great hope for the future.


wat

He's trying to say that we are the best team in the league. (Because we are.)


I'm going to stick with "wat".

Originally posted by PP


Don't get me wrong. I'm extremely relieved that we won these last 2 games and are starting to almost play the type of ball I'm used to again. Still, after the way we got our asses handed to us the 2 previous games, boosted or not, I don't think we deserve the top slot. Boosting doesn't make up for getting blown out, IMO. Sure, it may take a 3-14 game and make it 17-10 instead, but it doesn't mean as much as those two thrashing do, IMO.


Yeah, I'm not sure if you should really be #1 right now or not. Certainly in a "performance ratings" sense you don't deserve to be there based on what you've done so far this year. As for power rankings, which are just trying to measure how good the teams are right now, you might currently deserve the spot thanks to your recent boosts, but it's probably pretty close. I will say that one game is always a small sample size, with a lot of variance. I highly doubt you were "actually" 32 and 34 points worse than Cincinnati and Las Vegas when you played them. Probably you were more like 10-15 points worse, and variance went against you as well.

People love to say that levels mean nothing, and that this game is all about builds. While it's certainly true that good builds are incredibly important, and a team with good builds can certainly be better than a higher level team with bad builds, the fact remains that levels do matter. If two teams have identical, or even just equally good, builds then it's usually levels that will decide things. In a league like ours, where the top level players are level 30 (my apologies to our one level 31 guy), a three level boost for an entire team I think generally makes that team 20-25 points better. I'm not sure if your boost covered every player across your whole roster or not, though it looks like it didn't quite. I'm guessing by boosting you improved your team by about 15 points per game. If you were to replay Cincinnati and Las Vegas again right now, I think it would be reasonable to consider you small favorites (with a single digit spread), which is right where the rankings have you. I think #1 is reasonable for your team in a power ranking right now. Of course other teams in the league have the option to boost just as you did, and when they do they'll negate the temporary edge you have, and you'll certainly lose the #1 spot. I can't feel bad about you sitting there right now though.

I know I'm rambling a little, this post is as much a way for me to think about things because I just genuinely like ratings, and thinking about how ratings systems work (which is why I'm doing this in the first place), as it is an attempt to convince anyone else of anything. I have other football ranking formulas that I've developed besides the one here, that I use for college football. The distinction between "power rankings" and "performance rankings" is one that I've spent a lot of time thinking about, because of the BCS. A power ranking is designed to measure exactly how good teams are at the moment of the ranking. It doesn't matter how they've done, or what their record is. You can have a 4 loss team ranked ahead of an undefeated team, if the data suggests that the 4 loss team would win a game between the two. A performance rating (which is what I think the BCS should be... but let's not get started on that tangent), meanwhile, will (or should) have the undefeated team rated higher.

Personally, I find power rankings more interesting. Mostly I suppose it's because I enjoy sports betting, and obviously power rankings are much more effective as predictive tools, and so are more useful in assisting bet calculations. I have both a power and a performance rating system for college football that I keep every season, and I won't deny that the power rating formula is the one I've put a lot more work into, and is much more complex. Of course since those formulas are for real life, where there are no "levels" to look at, they are based on actual results (the performance rating completely ignores scores, and looks entirely at wins and losses, while the power ranking completely ignores record and looks only at scores). When I first started doing power rankings for GLB (Season 2, in Canada AA), I thought about using one of my existing formulas, but decided not to. I think that a ranking formula based on results just isn't good enough for GLB, where teams can change SO drastically mid-season. In real football a team is generally about the same all year long, and ranking them is just a matter of waiting for them to play enough games that you can get the statistical noise of variance worked out, and figure out what the results are actually telling you. In GLB though, teams can improve by 20 points from one day to the next by boosting. Playoff caliber teams can be sold mid-season, get gutted, and lose the rest of their games by 200. A traditional results-based ranking system simply has no way to account for things like this.

The reason I designed this rating system the way I did isn't because I think levels are the be-all end-all measure of team caliber. Obviously they aren't. I designed this system to be level-based because it made it dynamic. Every time I calculate the ratings I start from scratch and recalculate the base rating off of the current rosters, adding the accumulated adjustments on top afterward. This way, when a team boosts they immediately jump, and when a team guts they immediately drop, allowing the ratings to be much more accurate and effective as a predictive measure, at all points throughout the season. Additionally, the formula can apply in week 1. Obviously since levels don't tell you everything, the adjustment factors need time to accumulate before things really start to get accurate, but there's no question in my mind that, while not perfect, my rankings predict week 1 results better than any performance based rating system It's true that there are downsides to the route I chose. People frequently object to my rankings due to an over-reliance on levels, and while I generally disagree with the way in which these objections are phrased (I really don't think my rankings "suck" or are "worthless", lol), I certainly do agree with the sentiment. I do rely more on levels than I'd like, but so far haven't figured out a way to reduce that reliance while still maintaining the basic framework of the system. And I like the framework, because of the dynamic factors I just mentioned.

Huh, I guess this post turned into something of a manifesto. I didn't mean for that to happen, lol. I've had a lot of these thoughts bouncing around though, and I'm responding not just to things that people on this board have said, but also things people have said in the other leagues where I've posted rankings, in both this and previous seasons. It felt good to get some of it down in words. Hopefully if anyone actually takes the time to read the whole thing, they won't be too bored


Can you mix in a fuckiing cartoon or something for people like me?
 
Linker2596
offline
Link
 
Mass quoting. Ban hammer!
 
Page:
 


You are not logged in. Please log in if you want to post a reply.