CircularLogic
Joined: Aug 22, 2003
|
  Posted:
Nov 19, 2008 - 19:03 |
|
RealMadns wrote: | oryx wrote: | ... I think we can all agree that no amount of refinement on the part of the br/bbr system is going to create a system in which a coaches mettle, skill, and wit are accurately measured. Only a system based on TS, TR, or some other neutral factor can fulfill that role... |
I think that this is a common misunderstanding of how the sheduler works, and a reason for some of the hostility aginst it.
A system that uses BR to strive towards 50/50 games is the most refined way of 'testing coaches mettle' because when this system is stabilized the BR will be the exact unpolluted measure of your skill...
A sytem that uses only TS to pair games, and win% to measure skill, is polluted by the fact that you might get lucky and get your high win% by drawing mainly weaker coaches.
"Look, i got a 95 win%.. im the greatest bb-coach in the world!!!"
"yeah, but 25 of your wins are against RealMadns... it hardly counts" |
I bolded the part where you go wrong. You would still have a BR, just not as a factor in matchmaking. And that BR would be adjusted taking the BR of the opponents into account. So when all has settled, you get beaten 24 out of 25 times by coach X, but this 1 time nuffle hands you the win will weigh up the 24 wins of coach X, because for him winning over a low BR coach yields only a small gain while losing to you will make his BR drop significantly. |
|
|
RealMadns
Joined: Jan 16, 2007
|
  Posted:
Nov 19, 2008 - 19:18 |
|
@CircularLogic
True, you could still look at the BR...
But this is not what people are arguing, I hear the "I want my win% to mean something... 50% is boring" argument a lot.
If you need the BR anyway, what is wrong with making it pure?
If the win% is misleading anyway why is it importent?
Im not saying that a 100% TS matchup system wouldnt work fine, and I would probably enjoy it as much as this one, but I fail to see why the existing system is considered almost worthless in comparison, and why coaches are even 'threadning' to leave because of it... |
|
|
westerner
Joined: Jul 02, 2008
|
  Posted:
Nov 19, 2008 - 19:21 |
|
CircularLogic wrote: | westerner wrote: | But, take 2 coaches with BR/BBR of 180/150 and 150/150. Their best match will be a TS handicap of 24 with a score of ~896. (assuming TS of 150 for the 1st coach; score doesn't change significantly for different TS). |
Yes, but you can still pair a 180/180 coach with a 150/150 coach and get more than 998. |
Well, you could say the same about bashiness then - that BBR-incompatibility can be cancelled out by the addition of further BR-incompatibility. Either way, I find that an odd result of the scheduler.
I am thinking that what matters is not the theoretical TS handicap of best-matchup, but what matchups are likely to occur in practice. The box doesn't "spot" you a handicap; it simply looks for your best fit among the teams it has to work with. |
_________________ \x/es |
|
CircularLogic
Joined: Aug 22, 2003
|
  Posted:
Nov 19, 2008 - 19:51 |
|
@RealMadns:
Imagine there is a client, that uses CR - flawed as it may be - to determine how lucky you are during a match. Would you want to play your matches - fair as they may be - with such a client? How many matches would you watch and grind your teeth in frustration, because your opponent is just so bad, he -2d killed your best player and your apo fails. Sure, every other match you sweat out a narrow win, but the other half you are just getting mad because your opponent is doing things that are outright stupid and gets away with it to beat you. How many matches, before you say screw it, I want to play him on even grounds? Would you never shout at him, that if his 6+5+6+ dodge would have failed once out of the 3 times he did it, that you would have won?
@Westerner:
Yes, it`s true, the scheduler doesn`t determine a handicap. Yet, it is reasonable to believe that teams of each coach are equally distributed over the TS scores. That means that you are just more likely to get paired against coaches that can reach a higher max-score against you.
Example:
Coach A vs Coach B: Max score of 925 at TS 156
Coach A vs Coach C: Max score of 987at TS 163
If coach C has a team between TS 157 and 171 he will still have a higher score than coach B could ever reach. That means Coach A is more likely to be paired with Coach C, even though Coach B might be closer BR and TS wise.
So it is simple. If your 'BR minus BR is the same as my BBR minus BR, then we are most likely to get matched. Others that have other values for BBR minus BR need a more perfect match to my team than you do in order to get paired with me. |
|
|
Koigokoro
Joined: Sep 29, 2005
|
  Posted:
Nov 19, 2008 - 20:20 |
|
|
Purplegoo
Joined: Mar 23, 2006
|
  Posted:
Nov 19, 2008 - 20:22 |
|
Bear in mind they are Woodies, they're designed to beat everyone (well as he's doing)!
Keep that up with bashers.... No chance. |
|
|
Mr_Foulscumm
Joined: Mar 05, 2005
|
  Posted:
Nov 19, 2008 - 20:23 |
|
Purplegoo wrote: | Bear in mind they are Woodies, they're designed to beat everyone (well as he's doing)!
Keep that up with bashers.... No chance. |
What he said... |
_________________ Everybody's favorite coach on FUMBBL |
|
oryx
Joined: Jun 08, 2005
|
  Posted:
Nov 19, 2008 - 22:57 |
|
RealMadns wrote: | oryx wrote: | ... I think we can all agree that no amount of refinement on the part of the br/bbr system is going to create a system in which a coaches mettle, skill, and wit are accurately measured. Only a system based on TS, TR, or some other neutral factor can fulfill that role... |
I think that this is a common misunderstanding of how the sheduler works, and a reason for some of the hostility aginst it.
A system that uses BR to strive towards 50/50 games is the most refined way of 'testing coaches mettle' because when this system is stabilized the BR will be the exact unpolluted measure of your skill...
A sytem that uses only TS to pair games, and win% to measure skill, is polluted by the fact that you might get lucky and get your high win% by drawing mainly weaker coaches.
"Look, i got a 95 win%.. im the greatest bb-coach in the world!!!"
"yeah, but 25 of your wins are against RealMadns... it hardly counts" |
Luck isn't something that we can reasonably control (or, as circularlogic pointed out, would want to in this case), and so we must hope that it will eventually even out. If it does not, then either our number generator is broken, or reality is not what we think it to be.
Coaching skill *is* something that you can control. And, as I have seen in countless online games, if a system can be manipulated for personal gain or satisfaction, it will be. Introducing elements that are not random into the matching algorithm allows it to be manipulated by the players.
Personally, I prefer that bad luck to that any day - which is why I don't play in [Pick] |
|
|
RealMadns
Joined: Jan 16, 2007
|
  Posted:
Nov 19, 2008 - 23:49 |
|
@oryx
That is not my point...
My point is, that the BR adjusting system will indeed be competetive, and the BR in such a division will be a very acurate measurement of the coaches skill. |
|
|
vanGorn
Joined: Feb 24, 2004
|
  Posted:
Nov 20, 2008 - 20:08 |
|
RealMadns wrote: | @oryx
That is not my point...
My point is, that the BR adjusting system will indeed be competetive, and the BR in such a division will be a very acurate measurement of the coaches skill. |
Yes, indeed. The feedback from BR in the formula enhances the accuracy of the BR calculation. The convergence to the adequate BR should quickly be reached. |
_________________ Gimme a pint of fungus beer!
Then we will climb the ladder.
|
|
|
| |