DukeTyrion
Joined: Feb 18, 2004
|
  Posted:
Jan 22, 2015 - 11:43 |
|
PurpleChest wrote: | It's almost like the dark musing of 'clawpombers are the biggest pixel huggers' are true. |
I think clawpomb coaches have their share of pixel huggers, along with a decent mix of players that are not overly fussed. I know BillBrasky and I are not overly fussed about losing players.
That's not to say I don't get frustrated at the current scheduler, but that's not through fear of losing players, as much as frustration of trying to play a team with multiple loners against a much higher TV side. |
|
|
Harad
Joined: May 11, 2014
|
  Posted:
Jan 22, 2015 - 11:58 |
|
That's why I think people like Bill (and maybe yourself Duke, I haven't had the pleasure) are generally well accepted. As when one's getting smashed you still feel the other person would be happy if the boot were on the other foot (and there is at least a degree of irony in their enjoyment from the carnage).
It seems to be those who dish it out but can't take it and those who are completely unsympathetic to their opponent's plight who annoy. I don't mean they need to feel sorry or stop piling on or anything like that, but there is a difference between 'splat I killed your man heheheh (bad luck pal)' and 'splat I killed your man I am the greatest you are a loser'. |
|
|
Christer
Joined: Aug 02, 2003
|
Ok, so people are saying these large TV gap matches are unfair. In my world, unfair means "no chance of winning".
In order to get an idea of actual results, I figured I'd take all blackbox matches played on the site and look at win percentage over different TV differences (normalized, meaning 1000 vs 1100 is equivalent to 2000 vs 2200). After that query was done on the site DB, I plotted a graph showing the percetage of wins, ties and losses for every 10 TV difference up to 1100.
The resulting graph is pretty telling in how TV difference affects win probability of a match. As you can probably guess, the volatility at the high-end on this graph is caused by lack of any significant number of games (going below 10 games of data beyond -720 difference).
I did a similar query a while ago but chose to not publish it because it simply feels so counter-intuitive, but with this new chart I did showing the same correlation, I figured I am probably not too far off in terms of actual results.
With all that said, I'll leave you with the graph (which you undoubtedly saw before reading any of this anyway )
|
|
|
Harad
Joined: May 11, 2014
|
  Posted:
Jan 22, 2015 - 12:21 |
|
Hurrah! A triumph for science. Thanks Christer. |
|
|
Roland
Joined: May 12, 2004
|
  Posted:
Jan 22, 2015 - 12:23 |
|
so, after 450 TV difference, things become funky...?
(I don't really know how to read the graph) |
|
|
m0gw41
Joined: Jun 12, 2012
|
  Posted:
Jan 22, 2015 - 12:24 |
|
Are the results of the graph from the point of view of the team playing vs higher TV or lower TV team? |
_________________
|
|
Harad
Joined: May 11, 2014
|
  Posted:
Jan 22, 2015 - 12:28 |
|
1) The graph is noisey at high TV differences because of the lack of data (as Christer points out).
2) It appears the probability of a tie gradually decreases as the TV difference increases (downward sloping line).
3) The probability of winning and losing remain fairly constant until about 600 TV difference.
4) If anything there might even be a suggestion that the win ratio starts to go UP! (blue line might curve up a little).
Overall we should be cautious but what it does seem to show is that losing at high TV differences does not become hugely and conclusively more likely. |
|
|
bghandras
Joined: Feb 06, 2011
|
  Posted:
Jan 22, 2015 - 12:33 |
|
Such a graph with conclusive background data would be a PERFECT indication whether inducements are too good or bad in general. That would be a great analysis. Unfortunately the volatility (and possibly the sample size) is not good enough to draw such conclusion. |
_________________
|
|
koadah
Joined: Mar 30, 2005
|
  Posted:
Jan 22, 2015 - 12:35 |
|
|
Aino
Joined: Jul 15, 2005
|
  Posted:
Jan 22, 2015 - 12:37 |
|
All well, and nice graph! thx Christer
win or lose % is however not the main "issue" that was addressed in the opening post; "Cost me the death of my 2 fave legend players"
I don't see an issue really , but hey, it takes all sorts... |
|
|
Cloggy
Joined: Sep 23, 2004
|
  Posted:
Jan 22, 2015 - 12:42 |
|
Nice graph, but it doesn't address the huge issue brought to our attention of the OP, namely that he stands a chance of getting one of his precious clawpomb Cdorfs killed! |
_________________ Proud owner of three completed Ranked grids, sadly lacking in having a life. |
|
Lorebass
Joined: Jun 25, 2010
|
Roland wrote: | so, after 450 TV difference, things become funky...?
(I don't really know how to read the graph) |
Seems it becomes funky once Morg can get into the mix.
Morg = Unfair! |
|
|
DukeTyrion
Joined: Feb 18, 2004
|
  Posted:
Jan 22, 2015 - 12:57 |
|
Christer wrote: | Ok, so people are saying these large TV gap matches are unfair. In my world, unfair means "no chance of winning".
|
I would argue that unfair means "no chance of winning unless you sacrifice your team to do so". Sadly, I have a tendency to give up players to Nuffle in order to try and get a win
The irony is, the teams most likely to survive the huge TV differences are the bashers, who have built up a fund of gold, so in a way the current scheduler goes against promoting the other types of teams. |
|
|
Garion
Joined: Aug 19, 2009
|
  Posted:
Jan 22, 2015 - 13:01 |
|
The problem with that graph is it takes in to account all races. E.g. khemri are fine playing against same tv teams, but when they start playing against far higher tv teams they stand pretty much zero chance of winning assuming coaches are of a similar standard. However if you are using wood elves or skaven the chances of beating a higher TV team are much higher because of the wizard.
I think we all agree for the most part matching by tv should produce mostly fair match ups. What skews this is min maxing like chaos pact teams with no rrs
1 leader player and 4 block tackle cpomb players orzon teams with all blodge linemen and 1 pomb tackle blitzer and both these races could play against teams with a tv of 1100 which was mental.
So the scheduler change has dealt with problem 2 very well but there is still a problem as some races have been hit hard by the change and if anything it only benefits high tv cpomb the most as rebuilding is very perilous now as you could be matched up 750 tv. other teams like khemri take an eternity to rebuild TGs so are almost certain to run in to some high tv killers en route making a rebuild really tough with little chance of winning in the process.
Not sure what the answer is mind.
I really enjoyed playing in black box recently but it's not really suited for teams playing more than 30 or 40 games imo. |
_________________
|
|
bigGuy
Joined: Sep 21, 2009
|
  Posted:
Jan 22, 2015 - 13:09 |
|
Harad wrote: | ...
3) The probability of winning and losing remain fairly constant until about 600 TV difference.
... |
If I understand right, TV diff is normalized. So, that point in data represents 1200 TV diff for 2000 TV team.
Can we see graph where TV diff is not normalized? |
|
|
|