DrPoods
Joined: Nov 14, 2013
|
 Posted:
Mar 19, 2017 - 03:40 |
 
|
"Could end up anywhere".
Yes. Anywhere from #30-#32 |
_________________
"Gallifrey falls no more"
Do your part! Join the Adoption Agency NOW! |
|
Wreckage
Joined: Aug 15, 2004
|
| Posted:
Mar 19, 2017 - 04:28 |
|
| licker wrote: | But the point of power rankings (generally) is to rank teams in terms of what they have done and where they are at, not what they might be.
|
I wasn't aware of that. I thought I was supposed to make predicitions for how teams will place in the season. If the season was advanced those predictions could be done with a high confidence based on the numbers of this season. But the league isn't advanced.
Therefore I used data from last season. Your team didn't have any data from last season, therefore instead of 17-18 games, the confidence for your predicitions was just 4 games.
If there is low confidence in the data it means odds are you are closer to average than the data suggests.
If you are at the bottom end you are closer to the middle but if you are at the upper end you are also closer to the middle.
| Quote: | | There is in fact no way to tell how good someone really is who didn't mess up. Just a way to tell how terrible they are not. |
To make a more extreme example if someone played only a single game and won, he would have the same win percentage as you and correctly so, but the confidence in the accuracy of that statement would be lower.
Simple statistics would predict anyone who has never lost is predicted to win infinitely, but the confidence in that statement goes always towards 0 aswell. It merely implies the roof hasn't been found yet. So one way around that problem is to make a projection towards the next game and to say, just to have any data that doesn't require division through 0, lets presume the next game is a loss. Now I can say with some confidence your win percentage is at least 80. But really it might be any value between 80-100%. In difference to someone who has an actual win percentage of 80%. I don't actually know because I have 0 data about where your roof is and the lack of games you have played is the the direct source of that problem. Like: There would be no way from the data of four games to ever get to a win percentage of say 90%. Because 4 games simply can't provide that accuracy of data. But with the extreme values confidence is more important than say with the question whether someone is in the area of 60-80%. Because that the actual percentage projected over infinite games is the same as somewhere between 60-80% lies in the realm of reality whereas any projection towards 100% to be accurate lies in the realm of fables.
| Quote: | | Isn't the fact that they didn't mess up evidence that they are 'good'? I mean grues team is beat to shit and he has 2 losses, but he is ranked 4th? Slann are not exactly a powerhouse race either. |
About the assessment of a teams strength. It isn't solely based on race, but also about what material options the team has. I have plenty of experience with Slann and I know what they are capable of and if managed well they are an extremely powerful race. I also have a lot of respect for Vampires and believe they are in fact very well suited to beat Slann.
But your team is increasingly freshly created and quite frankly your good performance with them needs to be called surprising by any standards.
But regardless of all those informations I had still marked you in third place at this point of evaluation. On the other hand I was badly in need for any information that would increase my confidence in the position I placed you in.
That was when I noticed that virtually all the teams around you had a higher strength of schedule and a higher strength of victory as explained in the notes. By such a significant margin that I just couldn't ignore it.
So what does that mean?
It means we know your team does well against weak and midfield teams but we have actually no data whatsoever on how it would perform against a good team.
Add to that that your performance seems surprisingly good to me, it actually makes a lot more sense when the matchups get evaluated.
But what I needed to justify keeping you in third place was data from your team how you would perform against an elite team. That data was simply missing completely. That doesn't mean you wouldn't perform well against them. I simply have no idea if you would.
So I figure there is a very high chance you will reach playoffs but from there on out how far you make it is not clear at all.
I guess the top 8 in my mind have all a good chance to win the superbowl.
Sutherland I had previously labeled in first place but that was before the whole Piling On drama and then he didn't play too well either.
For the spots further above I favored coaches that gave me very reliable data and still placed them incredibly high. I figured I can't go wrong with that. |
|
|
Wreckage
Joined: Aug 15, 2004
|
| Posted:
Mar 19, 2017 - 04:51 |
|
As for the three parts I used to evaluate at teams strength that was:
Standing, Team, Coach Potential.
Team was probably the most subjective out of these three factors and I could be way off.
I did slightly favor high TV teams, which could be wrong. But the lack of Wizard access helped me in that decision.
Nobody can actually tell at this point how that will play out.
About standing I explained a lot in the post to Licker.
Coach Potential was based on Fumbbl data:
Winpercentage, CR and number of Games.
Winpercentage gives the most general idea. CR gives information about whether a coaches performance can peak and has peaked at the moment in another environment. And number of games serves as a control aspect of the other data.
If number of games is low the other data is not very telling, if the number of games is high, the other data should be expected to also be high because of massive experience. So if there is conflicting data there is also a lower confidence if the data is correct.
For instance to place someone in the top 8 based on his coaching abilities it should be in the area of at least 61% win percentage, 162 CR and more than 1000 games.
None of these stats would be outstanding, a really good coach should be able to hold those stats by only playing fun teams.
But any of those 3 data points massively exceeding the other data would be an indication that the coach has in fact a lot of potential.
Other meta data would be join date, which if very far back could give an indication that a coach uses primarily different platforms to play and won tournaments of course in the profile.
If you got a lot of those, particularly majors, your other stats won't really matter.
Anyways that's how I went on about it.
Obviously I have little experience with the league and really just have an outside view.
So it should be clear that there is a lot of metadata you guys probably have access to, I don't.
I figured this was more of a fun thing and was just doing it because I was asked to do so. |
|
|
JackassRampant
Joined: Feb 26, 2011
|
| Posted:
Mar 19, 2017 - 05:02 |
|
Thanks for doing it, Wreckage! It's a very fun thing.
(I assure you, Licker's having a good time with it.) |
_________________
Lude enixe, obliviscatur timor. |
|
Wreckage
Joined: Aug 15, 2004
|
| Posted:
Mar 19, 2017 - 05:03 |
|
|
licker
Joined: Jul 10, 2009
|
| Posted:
Mar 19, 2017 - 05:26 |
|
Mmm, like JR said, we are just having fun with it.
But...
I think you're working too hard with it 
My opinion (not a critique, just my opinion), past performance means nothing, all that matters is the season and teams as they are now.
I'm glad to see you're interested enough to have this much fun with it though. |
|
|
Wreckage
Joined: Aug 15, 2004
|
| Posted:
Mar 19, 2017 - 05:40 |
|
| licker wrote: | | But the point of power rankings (generally) is to rank teams in terms of what they have done and where they are at, not what they might be. |
| licker wrote: | | My opinion (not a critique, just my opinion), past performance means nothing, all that matters is the season and teams as they are now |
I think what you are loooking for are the league standings.
Well you can find a rough approximation (not really handling ties the same way) in one of the notes I posted. |
|
|
licker
Joined: Jul 10, 2009
|
| Posted:
Mar 19, 2017 - 05:46 |
|
Well yes, power rankings for other sports generally follow the standings fairly closely.
The interesting thing is comparisons across conferences, and to a lesser extent divisions, because not all teams are going to get a chance to face each other.
What you are doing, and doing a great job, is projections. Which I'm fine with, my only quibble, is that those are not really the same thing as power rankings. At least in as much as I am used to seeing them.
Carry on though, don't mind me |
|
|
Sutherlands
Joined: Aug 01, 2009
|
| Posted:
Mar 19, 2017 - 15:44 |
|
| Wreckage wrote: |
Sutherland I had previously labeled in first place but that was before the whole Piling On drama and then he didn't play too well either. |
Excuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuse me? |
|
|
JackassRampant
Joined: Feb 26, 2011
|
| Posted:
Mar 19, 2017 - 17:49 |
|
I think a power ranking has to be a cocktail of a few things. I could just release some stats cocktail, and I wouldn't have needed Wreckage. But we all know a few facts about this league that haven't been borne out by the stats as they've accumulated so far, or even by the stats and a good look at the teams.
Grue is in 4th place at 2-0-2, 'cuz that bodes for like a 3-loss season for him or something, and he'll make the playoffs and compete for the division title. He might or might not even be in bye-round position by the time the music stops. I think #4 is perfectly right for a stumbling champion with a team that we all know is better than it looks on paper and still a .500 record. |
_________________
Lude enixe, obliviscatur timor. |
|
licker
Joined: Jul 10, 2009
|
| Posted:
Mar 19, 2017 - 22:56 |
|
| JackassRampant wrote: |
Grue is in 4th place at 2-0-2, 'cuz that bodes for like a 3-loss season for him or something, and he'll make the playoffs and compete for the division title. He might or might not even be in bye-round position by the time the music stops. I think #4 is perfectly right for a stumbling champion with a team that we all know is better than it looks on paper and still a .500 record. |
Then just do them based on CR.
*shrug*
Not really that interesting then is it?
What we get is basically a coaching power ranking, not a team one.
Like I said before, I think there's a bit too much thinking going on for power rankings. Calculating strength of schedules, looking at past performances, picking out CR, ...
So it's fine though, it's interesting, it's just... kind of a weird way to do it in my opinion.
It's almost as though you're taking it seriously  |
|
|
happygrue
Joined: Oct 15, 2010
|
| Posted:
Mar 19, 2017 - 23:33 |
|
I've been hit pretty hard already this season. As I see it, it will be all the more humiliating when "#4" crashes and burns. Brad Jumps has two niggles and the other sacker is av 6. We're one bad game away from the edge, but if they stay alive then anything is possible...
Power rankings are made to be complained about. Welcome to the NBFL! |
_________________
Come join us in #metabox, the Discord channel for HLP, ARR, and E.L.F. in the box!
 |
|
Mr_Foulscumm
Joined: Mar 05, 2005
|
| Posted:
Mar 22, 2017 - 17:04 |
|
I just wish I could read the power rankings, as other have said, the formatting on this one doesn't help. Anyway, glad to see I'm not dead last! I'm making progress! 
Managed to read through it. I think the power ranking is fun and entertaining. And I do like how it's trying to predict how the season will go as well as looking how it's actually gone so far.
Great work Wreckage! |
_________________
Everybody's favorite coach on FUMBBL |
|
Wreckage
Joined: Aug 15, 2004
|
| Posted:
Mar 22, 2017 - 23:37 |
|
Well, I can make an image that is larger but I can't compress the same amount of informations. So here you go, Mr Foulscumm.
 |
|
|
Stimme
Joined: Jun 30, 2013
|
| Posted:
Mar 22, 2017 - 23:53 |
|
AFC West vs. NFC North: the highest power density in all the NBFL. We water our pitches with Gatorade. |
|
|
|
|
| |
Blood Bowl 2024 Edit...
Search
