Welcome

Please register for Total War Access to use the forums. If you're an existing user, your forum details will be merged with Total War Access if you register with the same email or username. For more information please read our FAQ’s here.

Categories

Do Tomb Kings Dominate Dark Elves?

tzurugbytzurugby Posts: 262Registered Users
I am looking for some outliers in the tournament data win rates and this one sticks out.... Tomb Kings have faced Dark Elves in competitive play 7 times so far and won all seven games for a 100% win rate over DE.

I'm wondering if this passes the sniff test.... do Dark Elves really struggle against Tomb Kings, and if so, why? What units or parts of the match up, make it easy for the Tomb Kings or hard for the Dark Elves?
«1

Comments

  • Lotus_MoonLotus_Moon Posts: 7,729Registered Users
    tzurugby said:

    I am looking for some outliers in the tournament data win rates and this one sticks out.... Tomb Kings have faced Dark Elves in competitive play 7 times so far and won all seven games for a 100% win rate over DE.

    I'm wondering if this passes the sniff test.... do Dark Elves really struggle against Tomb Kings, and if so, why? What units or parts of the match up, make it easy for the Tomb Kings or hard for the Dark Elves?

    I wouldn't say dominate but yes TK are good vs DE, actually TK are quite good vs all 3 elves, its not bretonia level of good but i would personally put TK at mild advantage vs all 3 elves.
  • ReymReym Posts: 477Registered Users
    I don't think they are close to dominate them. It's more due to the fact that the TK build to counter most goodies of DE is well known.
    Lot of nehekara cav and cav archers which counter crossbow cav and arkhan being a good lore of death caster he can ruin CoK with leechs and still use one bjuna for an eventual doomfire warlock unit.
    Having to counter that DE will have to invest a lot in mobility making his low tier infantry being bullied by like one chariot unit.

    Story short it's probably one of the MU where TK aren't forced to over invest in order to win important fight while DE seems to have to do it if they don't want to get swarm once they catch the mounted archer kite.
    But is talking about what is appropriate to talk about in this thread appropriate to be talked about in this thread ?
  • PippingtonPippington Posts: 2,021Registered Users
    tzurugby said:

    I am looking for some outliers in the tournament data win rates and this one sticks out....

    It sticks out more than WE having a 78% win rate vs Brets?

    Again, I think this is probably saying more about your data than about the factions.


    Get on, Kroq-Gar, we're going shopping

  • SarmatiansSarmatians Posts: 3,229Registered Users

    tzurugby said:

    I am looking for some outliers in the tournament data win rates and this one sticks out....

    It sticks out more than WE having a 78% win rate vs Brets?

    Again, I think this is probably saying more about your data than about the factions.
    Wood Elves master race!!! Woo-hoo!

    #nocounter #op #peasantssmellbad
  • DeludeDelude Posts: 299Registered Users
    I've been saying it's a bad MU since pre-patch and no one believed me. I'm beginning to feel blessed by your data, as it reflects many things I've been saying for a long time.

    Dominate? Eh, I wouldn't go that far but TK certainly have advantage in army select and on the battle map. That TK are underused makes it harder to prepare to counter them as well, compared to someone like Emp who you see every other match.
  • PippingtonPippington Posts: 2,021Registered Users
    edited November 7
    Delude said:

    I've been saying it's a bad MU since pre-patch and no one believed me. I'm beginning to feel blessed by your data, as it reflects many things I've been saying for a long time.

    Dominate? Eh, I wouldn't go that far but TK certainly have advantage in army select and on the battle map. That TK are underused makes it harder to prepare to counter them as well, compared to someone like Emp who you see every other match.

    The data say that DE vs TK is harder than than BM or Chaos vs VC, harder than Brets vs Lizards, harder than WE vs Brets or Dwarfs. Meanwhile the data also say that VC have a favoured matchup vs lizards and that Chaos has a 100% win rate vs coast.

    I'm not making a statement one way or the other about how hard the DE vs TK matchup really is, but if you take numbers like these as vindication I think that's confirmation bias at work.


    Get on, Kroq-Gar, we're going shopping

  • Green0Green0 Posts: 5,160Registered Users
    edited November 7
    Delude said:

    I've been saying it's a bad MU since pre-patch and no one believed me. I'm beginning to feel blessed by your data, as it reflects many things I've been saying for a long time.

    Dominate? Eh, I wouldn't go that far but TK certainly have advantage in army select and on the battle map. That TK are underused makes it harder to prepare to counter them as well, compared to someone like Emp who you see every other match.

    idk I feel DE have nice tools vs TK. Malekith in the air, Black Guard, Crossbow Riders all shine. COK, Darkshards etc playable also. Manticore summons seems strong in the MU. Likewise Lore of Fire or Death if you wanna go that way.

    TK for Elves is never a free win but honestly I find it one of the more fun and balanced matchups where there's a lot of little moments where each side has a chance to outplay the other.
  • The_real_FAUSTThe_real_FAUST Posts: 218Registered Users
    I think it is most likely the data. From my experiene on both sides its a pretty even and almost always fun match up
  • DandalusXVIIDandalusXVII Posts: 4,167Registered Users
    No they don't. It's just a few games outcome. I had 20 streaks in games, doesn't mean what I play dominates rest. Also I don't use OP factions, I play what I like.
  • Loupi_Loupi_ Posts: 1,290Registered Users
    I don't think it's a bad matchup for Dark elves, it can be tricky, but I think its tricky for both sides. I recommend cold one chariots. Harpies can be useful too.
  • DeludeDelude Posts: 299Registered Users

    Delude said:

    I've been saying it's a bad MU since pre-patch and no one believed me. I'm beginning to feel blessed by your data, as it reflects many things I've been saying for a long time.

    Dominate? Eh, I wouldn't go that far but TK certainly have advantage in army select and on the battle map. That TK are underused makes it harder to prepare to counter them as well, compared to someone like Emp who you see every other match.

    The data say that DE vs TK is harder than than BM or Chaos vs VC, harder than Brets vs Lizards, harder than WE vs Brets or Dwarfs. Meanwhile the data also say that VC have a favoured matchup vs lizards and that Chaos has a 100% win rate vs coast.

    I'm not making a statement one way or the other about how hard the DE vs TK matchup really is, but if you take numbers like these as vindication I think that's confirmation bias at work.
    Obviously the sample size is small, and even in my post I disagree that TK "dominate" DE. What I am saying is that the data reflects many things I have been saying for months, and I find that particular detail vindicating, as I tend to catch a fair bit of crap the first time I bring up something that might not be a popular opinion.

    7 games does not prove anything, but it does imply that if you were to gamble on a 100 games it might be safer to bet on TK.
  • rymeintrinsecarymeintrinseca Posts: 535Registered Users
    edited November 7
    The most lopsided MUs according to tzurugby's tournament-only table:

    WE 10-1 LM
    Empire 9-0 VP
    VP 8-2 LM
    Skaven 8-2 DE
    TK 7-0 DE
    Skaven 7-1 GS
    WE 7-2 Bretonnia

    (qualification: >5 wins for one side, win:loss ratio >3)



    The question is which of these are actually bad MUs and which are just random variation.
  • PippingtonPippington Posts: 2,021Registered Users
    edited November 7
    Delude said:

    Obviously the sample size is small, and even in my post I disagree that TK "dominate" DE. What I am saying is that the data reflects many things I have been saying for months, and I find that particular detail vindicating, as I tend to catch a fair bit of crap the first time I bring up something that might not be a popular opinion.

    If the data are flawed (or more accurately, if the win rates are a flawed way of looking into faction balance - the data are the data, they're not intrinsically 'flawed' as such), then they're flawed for all matchups. You can't dismiss the matchups that fail basic sanity checks as being due to a flawed methodology but then hold up the ones that happen to coincide with your opinions as evidence that you are right. There are 84 unique bins in tzu's table - just by coincidence, chances are that some of them are going to agree with any given position.

    The WE vs Brets figure is based off 9 games, 2 more than TK vs DE - this isn't just about sample size.

    The question is which of these are actually bad MUs and which are just random variation.

    The point I'd like to stress again is that it's not just random variation or sample size - there are systematic effects in the way these data are collected that strongly affect the results.


    Get on, Kroq-Gar, we're going shopping

  • tzurugbytzurugby Posts: 262Registered Users
    .

    The most lopsided MUs according to tzurugby's tournament-only table:

    WE 10-1 LM
    Empire 9-0 VP
    VP 8-2 LM
    Skaven 8-2 DE
    TK 7-0 DE
    Skaven 7-1 GS
    WE 7-2 Bretonnia

    (qualification: >5 wins for one side, win:loss ratio >3)


    The question is which of these are actually bad MUs and which are just random variation.

    I was not going to go into the analysis side of things, but that is almost exactly the data set that I looked at last night.... I ran a Standard Deviation of the entire data set (sigma is about 0.21) then just looked at every match up with more than 7 games and greater than a 72% win ratio. Some I can make sense of, some are obvious outliers and some strike me as odd as I have never seen them described by anyone as difficult.....

    To keep it simple I planned to tackle them one at a time, and continue looking as the data comes in.

    @Pippington I left that Bretonnia data alone as I knew someone would enjoy pointing it out.... I know some people love data and some people don't and some people have very precise internal definitions of what makes sense. I like giving everyone a chance to participate, which is why I have tried to make any data I'm gathering as transparent and easy to independently verify as possible, and very open to everyone so it can be easily criticized with concrete examples.
  • PippingtonPippington Posts: 2,021Registered Users
    edited November 7
    Yeah and I appreciate the time and effort you're putting into this @tzurugby , but I think this thread already shows people running away with these numbers as propaganda tools rather than the sort of circumscribed information they really are.

    Here are the main 2 effects I would guess (emphasis on guess) that are driving the results that strike you as odd.
    1. You are not preserving data about the players involved. A 7-2 win record for faction A vs faction B does not accurately reflect the faction balance if the majority of the 7 wins were repeat tournament finalists vs newbies, and it also isn't accurately represented as having a sample size of 9 if a disproportionate number of the wins (or losses) were from the same individual player.
    2. Players have their own mental picture of what this matrix looks like and pick factions accordingly. So in any free-pick format, you are not sampling a uniform distribution of games in each matchup - you're sampling an enhanced no. of games where the player who allowed the matchup to happen thought they could win, for any number of different reasons (but probably highly correlated to difference in skill to their opponent).
    You can kill point 2 if you limit your analysis only to faction-locked round-robin/swiss tourneys, but obviously at the cost of greatly reduced sample size. Point 1 I think requires a really really serious data-gathering structure to address, and I would guess is easily strong enough to be a dominant factor behind results like WE 7 - 2 Brets. In the current dataset I think you can see a very strong indication of both of these effects in the fact that >50% of the 7 'worst' matchups quoted above are bad for either DE or LM - i.e. simply the most frequently picked factions in tournament games.

    Even for results that do seem to conform to common sense, it's impossible currently to tell how large an impact these issues are having, and interpreting the results as any kind of reliable statement about faction balance is IMO wrong.


    Get on, Kroq-Gar, we're going shopping

  • SeldkamSeldkam Senior Member Posts: 4,105Registered Users
    Don't think TK have much of an advantage tbh. It's more likely that TK are just a good faction overall. It's easy to build yourself into a win or lose in the matchup though imo. You need darkshards, but only if the TK bring constructs, and usually ones that get into melee. Therefore if you bring them vs a wide Skelly and bowshabti army you're wasting a lot of funds already.
    The inferior races of this world will be crushed one by one, as our armies move from shore to shore, and hill to hill, and city to city-- and each of their cries will be as music to our ears, for we are the Druchii.
  • DeludeDelude Posts: 299Registered Users

    Delude said:

    Obviously the sample size is small, and even in my post I disagree that TK "dominate" DE. What I am saying is that the data reflects many things I have been saying for months, and I find that particular detail vindicating, as I tend to catch a fair bit of crap the first time I bring up something that might not be a popular opinion.

    If the data are flawed (or more accurately, if the win rates are a flawed way of looking into faction balance - the data are the data, they're not intrinsically 'flawed' as such), then they're flawed for all matchups. You can't dismiss the matchups that fail basic sanity checks as being due to a flawed methodology but then hold up the ones that happen to coincide with your opinions as evidence that you are right. There are 84 unique bins in tzu's table - just by coincidence, chances are that some of them are going to agree with any given position.

    The WE vs Brets figure is based off 9 games, 2 more than TK vs DE - this isn't just about sample size.
    I'm not sure that I'm the one you should be directing the bulk of this post toward. I don't think I've ever criticized Tzu's methodology, I'm quite grateful for his effort. Nor have I held winrates, or any particular point of the data, as holy gospel. Nor have I dismissed points that I disagree with while upholding others.

    I do think it's interesting that many points I have made in the past are being reproduced in the statistics. This doesn't necessarily make me right, it just means that what I have observed is being repeated on a small scale within a small community. And, while I'm not going to bust out the "TOLDYA SO'S" just yet, I do take some reassurance from the existing data that I am not just imagining things. My theories on balance might just be close to true, in that aspects of them can, not necessarily will, but demonstrably can, happen. (On a small scale.)

    And I think I'm allowed to do that. Fite me, Lizard boy!


  • PippingtonPippington Posts: 2,021Registered Users
    edited November 7
    Delude said:

    I do think it's interesting that many points I have made in the past are being reproduced in the statistics.

    I'll try to sum it up in less technical language: if you had made a number of yes-no predictions in the past, and someone then came along and flipped a coin to determine an answer to each of those questions, the number of times the coin-flip agreed with you would not be any form of validation of the arguments you had made.

    I didn't mean to imply you were criticising tzu's methodology: I am criticising it, or at least criticising the use of his numbers to make arguments like these. If from 9 games the data can predict that WE vs Brets is a 78% WE favoured matchup, then it's clear that they are wholly unreliable. Interpreting what they predict from 7 games about any given matchup as being a meaningful statement about balance is like interpreting a coin-flip.


    Get on, Kroq-Gar, we're going shopping

  • Green0Green0 Posts: 5,160Registered Users
    edited November 7

    Delude said:

    I do think it's interesting that many points I have made in the past are being reproduced in the statistics.

    I'll try to sum it up in less technical language: if you had made a number of yes-no predictions in the past, and someone then came along and flipped a coin to determine an answer to each of those questions, the number of times the coin-flip agreed with you would not be any form of validation of the arguments you had made.

    I didn't mean to imply you were criticising tzu's methodology: I am criticising it, or at least criticising the use of his numbers to make arguments like these. If from 9 games the data can predict that WE vs Brets is a 78% WE favoured matchup, then it's clear that they are wholly unreliable. Interpreting what they predict from 7 games about any given matchup as being a meaningful statement about balance is like interpreting a coin-flip.
    it was clear from the start that there are too many variables in the model (such as player X loving faction z and doing well with them even in difficult matchups, or the map being unfavorable in spite of a favorable matchup) and the sample size is so small that any conclusions drawn are meaningless. Not sure why people even bother.
  • PippingtonPippington Posts: 2,021Registered Users
    Green0 said:

    it was clear from the start that there are too many variables in the model (such as player X loving faction z and doing well with them even in difficult matchups, or the map being unfavorable in spite of a favorable matchup) and the sample size is so small that any conclusions drawn are meaningless. Not sure why people even bother.

    It's still a very nice factual record of what's going on in tournament games, and I don't want to pooh-pooh tzu's efforts at all - I just want to urge people to be careful about over-interpreting the numbers.



    Get on, Kroq-Gar, we're going shopping

  • DeludeDelude Posts: 299Registered Users

    Delude said:

    I do think it's interesting that many points I have made in the past are being reproduced in the statistics.

    I'll try to sum it up in less technical language: if you had made a number of yes-no predictions in the past, and someone then came along and flipped a coin to determine an answer to each of those questions, the number of times the coin-flip agreed with you would not be any form of validation of the arguments you had made.

    I didn't mean to imply you were criticising tzu's methodology: I am criticising it, or at least criticising the use of his numbers to make arguments like these.

    I'm not making any argument from the data, though. I already believed TK vs DE was advantageous to TK from my own experience, and games I've watched between the factions in the hands of other (better) players, and from roster analysis. I have posted that before. The only argument I made is "army select and battle map both favour TK, but it's not domination".

    The reasons I believe that are separate from the admittedly pleasing to behold numbers produced in this thread. But, it does make me feel warm and fuzzy. As would someone flipping coins to match a set of yes/no predictions I'd made. It's fun to feel like the designer of reality, master of causality, and lord of probability that I was always meant to be.

    I wouldn't use them to support my pre-existing opinion, they're simply too small and variable. Darn are they pretty, though.
  • Green0Green0 Posts: 5,160Registered Users

    Green0 said:

    it was clear from the start that there are too many variables in the model (such as player X loving faction z and doing well with them even in difficult matchups, or the map being unfavorable in spite of a favorable matchup) and the sample size is so small that any conclusions drawn are meaningless. Not sure why people even bother.

    It's still a very nice factual record of what's going on in tournament games, and I don't want to pooh-pooh tzu's efforts at all - I just want to urge people to be careful about over-interpreting the numbers.

    other than admiring the beauty of organized work, I see little possibility to draw conclusions from this data. The biggest noise variable seems to be skill level, which spans a vast interval of values and isn't easy to measure either.
  • PippingtonPippington Posts: 2,021Registered Users
    edited November 7
    Green0 said:

    other than admiring the beauty of organized work, I see little possibility to draw conclusions from this data.

    I think you can draw conclusions about trends in pick rates at least - and the way that tzu is preserving 1st pick vs 2nd pick information helps with that. Pick probability is (probably) less sensitive to skill differences than win probability.


    Get on, Kroq-Gar, we're going shopping

  • Disposable HeroDisposable Hero Posts: 3,697Registered Users
    I think that so far people have been very sober about this dataset I must say. There are some interesting results there, not because they prove anything, but because that's how results have been lately. I mean it's hilarious to see WE in a strong lead vs Bret, and we all know why that is so no sweat. :smile:
  • The_real_FAUSTThe_real_FAUST Posts: 218Registered Users
    edited November 8
    I agree from a purely scientific stance this record won't offer empirical data with which to balance with.

    What it is however is a fascinating record and a clear demonstration which hopefully acts as an ongoing catalyst for community involvement. Further it aids record keeping and for that reason beyond anything else it is to be much praised and applauded.

    As a tool for balancing it is inadequate alone without further tinkering.

    It's a shame CA don't offer player records we can access in game so we can see our win/loss rate, and with each faction against each faction.
    Post edited by The_real_FAUST on
  • rymeintrinsecarymeintrinseca Posts: 535Registered Users


    The question is which of these are actually bad MUs and which are just random variation.

    The point I'd like to stress again is that it's not just random variation or sample size - there are systematic effects in the way these data are collected that strongly affect the results.
    Yes, to some extent we're in a Literary Digest situation (https://www.math.upenn.edu/~deturck/m170/wk4/lecture/case1.html)

    But randomness should not be underplayed here. George Gallup's 'small' sample was still 50,000 people! For individual MUs we have preposterously small sample sizes and randomness is likely to be a huge factor - probably more so than the sampling issue, as individual MU samples are not so strongly biased (in respect of player quality) in the way that the Literary Digest sample was (in respect of social class etc). The randomness should diminish as the dataset increases.

    If higher standard players systematically play certain factions/MUs, and always will do so, we would have a case of selection bias rather than randomness, and it won't be removed by a bigger sample. To some extent that's likely to be the case, but is it such a strong effect that it undermines the value of the data? That seems at this point to be speculation.

    And even if it was true, it's not easy to deal with. I suppose in theory you could either design the sample so players were of an equal standard for each faction, or build a statistical model with player quality as one of the independent variables, e.g. by using players' tournament performance. But who's going to do that in their free time?

    I think it's more realistic to accept that this is the data we have. As long as you keep in mind the limitations it's a lot better than nothing.
  • rymeintrinsecarymeintrinseca Posts: 535Registered Users


    It's a shame CA don't offer player records we can access in game so we cam see our win/loss rate, and with each faction against each faction.

    CA can't stop you from keeping a spreadsheet :)
  • AsamuAsamu Posts: 581Registered Users

    Yeah and I appreciate the time and effort you're putting into this @tzurugby , but I think this thread already shows people running away with these numbers as propaganda tools rather than the sort of circumscribed information they really are.

    Here are the main 2 effects I would guess (emphasis on guess) that are driving the results that strike you as odd.

    1. You are not preserving data about the players involved. A 7-2 win record for faction A vs faction B does not accurately reflect the faction balance if the majority of the 7 wins were repeat tournament finalists vs newbies, and it also isn't accurately represented as having a sample size of 9 if a disproportionate number of the wins (or losses) were from the same individual player.
    2. Players have their own mental picture of what this matrix looks like and pick factions accordingly. So in any free-pick format, you are not sampling a uniform distribution of games in each matchup - you're sampling an enhanced no. of games where the player who allowed the matchup to happen thought they could win, for any number of different reasons (but probably highly correlated to difference in skill to their opponent).
    You can kill point 2 if you limit your analysis only to faction-locked round-robin/swiss tourneys, but obviously at the cost of greatly reduced sample size. Point 1 I think requires a really really serious data-gathering structure to address, and I would guess is easily strong enough to be a dominant factor behind results like WE 7 - 2 Brets. In the current dataset I think you can see a very strong indication of both of these effects in the fact that >50% of the 7 'worst' matchups quoted above are bad for either DE or LM - i.e. simply the most frequently picked factions in tournament games
    Your point 1 is actually too difficult to quantify to ever apply to any data analysis. Sure, it's probably true to some extent, but, there's no way to assess the skill of a "new" player on the scene, or even necessarily to do so for existing players. Sometimes people show up and take a top 2 or 3 spot at a tournament with no prior tournament experience; it's not common, but it happens occasionally. In this case though, top players play a very diverse set of factions, and it's highly unlikely that all cases of any match up are a better player vs a worse player.

    Like in any game, if the best player is picking a particular thing more often than others and winning consistently with it, it's an indicator that that thing is perhaps too good at a high level of play. The more competitive players tend to pick what they feel gives them the best opportunity of winning. - If the top players are letting their wood elves get "counterpicked" by Bretonnia and winning anyway, and not picking Bretonnia into their opponent's wood elves, the common opinion regarding the match up that it heavily favors Bretonnia is, at the very least, questionable.


    Point 2 is probably a bit more applicable.
    People have the perception that Dark Elves and Lizardmen don't have any really bad match ups; they pick them, then lose in a match up they might not be as familiar with, or that is actually poor, if not as poor as the stats indicate.


    The stats at least lend credit to the idea that perhaps, for example, the WE vs Bret match up is not as bret favored as people like to claim. Even the WE player being better in all of those games, which I'm fairly certain it is not, if the match up was as bad as people frequently claim for the wood elves, the better player wouldn't be winning 7/9 games, even in this relatively small sample size; that would require a more substantial gap in skill, than is likely among tournament players. It's more likely that the match up is close to even, and people just have the perception that it's bad for wood elves for one reason or another, and then the skill difference drives the wood elf win rate up some.

    Of course, with such a small sample size, nothing is definitive, but it does indicate some things that are worth exploring. Vampirates being 0-9 vs empire for example is a decent indicator that the match up is empire favored, even if it's not necessarily the worst match up in the game. With the more extreme stats like this at such a small sample size, it should be pretty obvious that the actual percentages are extremely suspect; even 1 game with a different result would change that percent by 10% or more.
  • dge1dge1 Moderator Arkansas, USAPosts: 18,633Registered Users, Moderators, Knights
    Let's all get back to discussing the thread topic and question. Thanks.
    "The two most common things in the universe are Hydrogen and Stupidity." - Harlan Ellison
    "The right to be heard does not automatically include the right to be taken seriously." - Hubert H. Humphrey
    "Never argue with an idiot. They will only bring you down to their level and beat you with experience.” - George Carlin
  • sonofabhorashsonofabhorash Posts: 78Registered Users
    edited November 8
    Agree with the mod and thx,wasting my time trying to get throught tons of would like to be math-philos crap which i have neither time nor mood for
Sign In or Register to comment.