JimmyFantastic
Joined: Feb 06, 2007
|
  Posted:
Feb 20, 2012 - 15:26 |
|
I made this from Hitonagashi's 960-1210TV blackbox stats to see how the best teams(and Orcs because they are so popular) fared against each other at low TV.
I was thinking about choosing a team for NAF champs or just keeping a team at low TV here.
Chaos Dwarfs look like a good choice to me due to a lack of bad matchups to other good teams, any thoughts welcome. |
_________________ Pull down the veil - actively bad for the hobby!
Last edited by JimmyFantastic on %b %20, %2012 - %15:%Feb; edited 1 time in total |
|
hizard
Joined: Sep 11, 2009
|
  Posted:
Feb 20, 2012 - 15:31 |
|
I think this chart is worthless but that might be only me... |
|
|
Garion
Joined: Aug 19, 2009
|
  Posted:
Feb 20, 2012 - 15:34 |
|
I agree with hizard, but yeah Cds Dwarves and Undead are all typically good starter teams, as are Wood elves and pro elves. But its also down to your style really. |
|
|
JimmyFantastic
Joined: Feb 06, 2007
|
  Posted:
Feb 20, 2012 - 15:34 |
|
Why do you think it is worthless? I think it is servicable as a rough guide. |
_________________ Pull down the veil - actively bad for the hobby! |
|
the.tok
Joined: Jan 25, 2006
|
  Posted:
Feb 20, 2012 - 15:35 |
|
Looking at these numbers, Dwarves and CDwarves seem like a good choice.
Amazones too, but it depends on the metagame cause the only bad match-up is dwarves and to a lesser extent CDwarves. If you're facing dwarves rarely, you can cope with the odd game against them, for the sake of having an advantage against everyone else.
Having informations about the % of each race in the environment would be valuable to weight the importance of each MU. But it can vary among each tournament |
_________________ Unnerf Mummies 2013! |
|
hizard
Joined: Sep 11, 2009
|
  Posted:
Feb 20, 2012 - 15:38 |
|
Well first of all its box ratings. Second its only few teams. Third what do you mean by low TV? |
|
|
JimmyFantastic
Joined: Feb 06, 2007
|
  Posted:
Feb 20, 2012 - 15:41 |
|
Box ratings is bad why?
I consider winning% vs goblins etc largely irrelevant so condensed the table to the best teams only.
It is 960-1210TV |
_________________ Pull down the veil - actively bad for the hobby! |
|
hizard
Joined: Sep 11, 2009
|
  Posted:
Feb 20, 2012 - 15:48 |
|
Its bad cos teams are built for box there... Goblins might be irrelevant but there is chaos, nurgle, norse and etc that isn't. Also box does not ensure that matches are from equal coach level, or even more does not ensure that the level of coaching will be similar at NAF. Or anywhere else.. Anyway you shouldn't pick race by winning percentage. Also you don't need percentage to know which MU is good and which is bad. Also there is no team without weakness against some teams. |
|
|
gpope
Joined: Jun 04, 2010
|
  Posted:
Feb 20, 2012 - 15:50 |
|
hizard wrote: | Also box does not ensure that matches are from equal coach level, or even more does not ensure that the level of coaching will be similar at NAF. |
Yeah, looking at the numbers for e.g. orcs I think the lesson is not "orcs aren't a competitive team at low TV" but "orcs are really popular with easily trounced newbies." |
|
|
Sp00keh
Joined: Dec 06, 2011
|
  Posted:
Feb 20, 2012 - 15:51 |
|
problem with your chart is you've neglected the quantity of games. VoodooMike did the same mistake when he looked at hito's data end of last year
basically, if a race1 wins:
2 out of 3 games against raceA,
2 out of 3 against raceB,
40 out of 100 against raceC,
then your method would show it having an average win rate of 57%
however, its actual win rate would be 41%
this problem is real, in the data, certain matchups are a lot more frequent than others, even among the 'main' teams.
it is significant enough to move team ranking around by several positions. |
|
|
Sp00keh
Joined: Dec 06, 2011
|
  Posted:
Feb 20, 2012 - 15:58 |
|
here's my analysis of the same data, same 960-1210 bracket.
the number is: (% chance to win) + (% chance to draw halved)
opposition race is ignored, all games played counted, sample size 36,414
Code: | Amazon 60.23
Undead 56.81
Lizardman 56.03
Skaven 55.80
Chaos Dwarf 53.99
Norse 53.52
Dwarf 53.29
Wood Elf 51.84
Orc 51.10
Human 51.08
Necromantic 49.65
Dark Elf 49.26
Khemri 49.11
Chaos 48.32
Elf 47.63
Chaos Pact 47.60
High Elf 46.97
Nurgle 46.24
Slann 44.72
Underworld 39.55
Vampire 35.08
Halfling 28.65
Goblin 28.35
Ogre 26.42 |
|
|
|
JimmyFantastic
Joined: Feb 06, 2007
|
  Posted:
Feb 20, 2012 - 16:00 |
|
Sure it's not perfect, but I thought worthless was too harsh a word.
I think it is food for thought.
You really do need stats to determine whether matchups are favourable or not.
Chaos and Nurgle are terribad starting teams so not worth consideration for a tournament style team in my opinion.
Sp00keh, I don't really see your point as the point of my stats are purely the vs race win%.
This is why I didn't include Norse because as you can see their overall win% is good but it is inflated by outstanding records vs suboptimal teams like stunties. |
_________________ Pull down the veil - actively bad for the hobby! |
|
Sp00keh
Joined: Dec 06, 2011
|
  Posted:
Feb 20, 2012 - 16:16 |
|
the point is, the data is lacking.
36,414 sounds like a lot of games, but it doesn't have statisically significant numbers of games for every matchup, eg there's only 14 games of elf vs high elf, there's only 28 khemri vs undead, and a lot of others in low double figures
whereas chaos vs chaos dwarf has 210 results
this means outliers can skew the results too much, like you noticed with the norse.
so your numbers give the indication that chaos dwarves have the best chance against any opponent.
but, my numbers show that amazons can expect to win or draw more games overall*
*win or draw more games overall, against the average blackbox selection of teams
the data is from blackbox, and i don't have any data as to what teams you'll face at a tournament. maybe people don't take chaos to tournaments |
|
|
JimmyFantastic
Joined: Feb 06, 2007
|
  Posted:
Feb 20, 2012 - 16:20 |
|
Yep exactly, the data is lacking. We can't really do anything about that.
I would assume everyone knew that amazons can expect to win or draw more games overall than CD's*.
My idea was to take the top performing teams and see how they compared to each other to maybe get more relevant stats for tournaments.
*win or draw more games overall, against the average blackbox selection of teams
the data is from blackbox, and i don't have any data as to what teams you'll face at a tournament. people generally don't take chaos to tournaments cos they suck. |
_________________ Pull down the veil - actively bad for the hobby! |
|
Sp00keh
Joined: Dec 06, 2011
|
  Posted:
Feb 20, 2012 - 16:29 |
|
the norse-case-scenario is worth focusing on actually, as it explains what i mean with an example
they played 2318 games in total, of which: goblin + halfling + underworld + ogre + vampires = 192 or 8.2% of their games, which is a fairly typical percentage.
these 5 teams lose about 75% of the time overall.
if norse had a very strong record against those 5 teams, say 99%, that'd ruin your numbers and make them look like the best team
however my numbers suppress this effect because its scaled by number of games played, and 8% of games played isn't a big deal
Quote: | My idea was to take the top performing teams and see how they compared to each other to maybe get more relevant stats for tournaments. |
my data is skewed for the average box team. i see that in tournaments it isn't going to be like that. your method probably makes more sense when picking team for tournament, but you also had plans to use the info here ... |
|
|
|