Pinned HE General and News - Discussion

  • With apologies to @20phoenix...

    So, I've been tinkering a little bit with the actual data. First result: I did a chi-squared test on army-army pairs in the same team. Bottom line: nowhere near significant. (Now, I'm ballparking this, because determining degrees of freedom is non-trivial here. And the straightforward way to do it is wrong (double counting issues). So I ballparked a minimum and maximum degrees of freedom, and it's not even close to significant in either. Can't even get p < 0.5, which when you're looking for p < 0.05, that's a huge difference.)

    So, when people claim that better armies end up in better teams, I'm reasonably sure this is wrong, because there is no tendency to put even the same two armies on the same team. I might upload a table later, but it's not going to be very informative, because the fact of the matter is that the army sample sizes are way too small for the degree to which there is any correlation. You'd need about 1000 teams worth of data if this data is typical to detect any possible signal. That caps the effect size.

    I haven't done a strict analysis, but you could also look at the top 5 teams. No one army is shared between all of them. The top two teams share only a single army. 12 of 16 armies appear in the top 3 teams, which is, if anything, more diverse than expected given the frequency of army entries in the tournament (although not so much that it would be significant). That number expands to 13 different armies with the 4th place team, and 15 different armies with the 5th place team. In short, the top teams aren't all selecting the same armies or army groupings. They're super diverse in terms of army representation.

    Now, there likely is some signal on team composition, but it's not so simple as AB-AB pairings or superior ABs. It's about functional roles. For example, of 9 DE players in the WTC, 4 of them are on teams with HbE, and 4 of them are on teams with SE. That's 8/9 DE players are on a team with another elf army. But no team includes all three. We'd need a good model of functional roles for team play to even start analyzing that kind of choice, and it would be complicated by books that could switch roles depending on list. But that's clearly moving away from AB strength toward functional team dynamics with team pairing mechanics.

    ---------------------------

    So, how does HbE performance stack up. I wanted to separate out the effect of team performance from this analysis, because obviously the whole point of a tournament is to figure out which teams have better players. So your team doing worse isn't so much a function of one army book strength, but the team's strength. And if the team is doing better or worse, well, you'd expect the individual player to be doing similarly, because he's on that team. If AB strength is driving things, we have to wonder what we mean by a team's result being dictated by one army book.





    So this is how I tried to get at it. First I looked at the correlation between team place and raw points for the HbE player. Team place predicted HbE raw score fairly well. But what happens if we remove the effect of team score from the data? This gives us the percentage of the team's points the HbE player earned - ie, how well did the HbE player do independent of how well his team did. Once you do that, most of the explanatory power goes away. Especially if you remove the Slovakia HbE player (as he was nowhere near the skill level of anyone else at the tournament, there is good reason for doing this a priori, and the fact that he personally accounts for 14% of the variance explained is a good a posteriori reason - that's almost as much as all the other HbE players together).

    Note, those latter relations - not significant at all. 20% of the variance is nothing with 10 data points. (Even leaving Slovakia in, 34% of the variance is nothing with 11 data points). (I haven't tested significance on the left graphs - not relevant, there's so much auto-correlation going on, so the significance test assumptions are wrong). But the key take away is that most of the variance is explained by how good the team is, not the HbE player or army book. And in fact, 20% of the variance is exactly what you'd expect, as the HbE player is 20% of his team (Woh! Your mind can be blown).

    In conclusion: HbE's showing wasn't due to book strength, but differences in player strength. And while it's possible stronger players deliberately avoided HbE, they didn't deliberately choose any particular other book or pairing. With the small amount of data, it's impossible to tell if HbE's under-selection at the top is significantly different from expectation. (Also, i found another top 20 team's HbE player I'd missed before - Spain 1, and some places got adjusted, so it's teams 10,11,16,19 with HbE players in the top 20).

    (Edit: Bottom 2 graphs are the same as the top 2 graphs, but with Slovakia's HbE player removed).
    Just because I'm on the Legal Team doesn't mean I know anything about rules or background in development, and if/when I do, I won't be posting about it. All opinions and speculation are my own - treat them as such.

    Legal

    Playtester

    Chariot Command HQ

    The post was edited 1 time, last by Squirrelloid ().

  • oh for crying out enough.
    ENOUGH ALL OF YOU !!!
    This has become just a s*** fest of whose math and data analysis is better than the other guy's and, considering this is a COMMUNITY driven project, it is frankly sickening.

    Numerous play communities within Europe are actually smaller, than the size of the 9th age realization team. Too many opinions, unflexible leadership. Those are the problems.

    And here we are. Somebody said we are creating balanced single player game. Yet, our mosr prestigious tournament is team event. The f*** ?

    Directly regarding HBE standing...almost average number of armies. Top teams ? No

    Top teams had time to playtest and, as someone here very accurately described...HbE can not create strong scewed lists, are hard pressed to conserve points, can hardly go for overwhelming wins and are unreliable.
  • PadForce wrote:

    Borjnfer Wraith wrote:

    goatman wrote:

    Squirrelloid wrote:

    And you know what I still haven't seen? Any statistical analysis of WTC data that proves HbE actually underperformed. At least show me a p-value. Otherwise I have no reason to believe, even if I accept the data is meaningful, that it actually shows anything at all. If 720 games is sufficient data, then you have statistical proof HbE underperformed and it's not just random chance, right?
    Agree, I read some negativity on HBE's performance on WTC, so as former data analysis member I fully support this statement.For those worried about the mean of 67.9pts/player out of 80 (8.5 points per game), look at individual results. It could be argued that someone scoring less than 5 points per game is a bit of an outlayer in the data (either had very, very bad luck, or was thrown to the lions by his teammates, or maybe was not experienced enough to go to such a big and competitive event).
    Out of 180 players, 9 of them are below 40pts out of 160 possibles and, guess what: TWO of them are HBE. If you'd remove them from the average, you'd get a new mean of 78.1 points (9.76 points per game), which is not statistically underperforming.

    So, please don't let the MEAN value of a small sample of a TEAM event with potential outlayers guide your opinion towars a (possibly) wrong statement.
    If the HBE community wants accurate results then I suggest that the HBE players at WTC submit their battle reports and then a more responsible critique can be calculated.
    imo doing things to that level just doesnt sound feasible surely? I cant imagine the project has time to perform such detailed analysis across the number of games it would need reviewed?
    If each AB community gathered their own data and processed it using a common guide of course, then this would go a long way in helping the decision members get the detailed data they need to do their job.
    Failure is not an option.
  • @Squirrelloid Good resume, thank you very much. 9th age is lucky to have some people understanding statistics to guice team.

    Evaenarion wrote:

    oh for crying out enough.
    ENOUGH ALL OF YOU !!!
    This has become just a s*** fest of whose math and data analysis is better than the other guy's and, considering this is a COMMUNITY driven project, it is frankly sickening.

    Numerous play communities within Europe are actually smaller, than the size of the 9th age realization team. Too many opinions, unflexible leadership. Those are the problems.

    And here we are. Somebody said we are creating balanced single player game. Yet, our mosr prestigious tournament is team event. The f*** ?

    Directly regarding HBE standing...almost average number of armies. Top teams ? No

    Top teams had time to playtest and, as someone here very accurately described...HbE can not create strong scewed lists, are hard pressed to conserve points, can hardly go for overwhelming wins and are unreliable.
    Also Evaenarion: why do you consider statistics frankly sickening? This is beyond me, because the only other way to proceed is to simply buff or nerf through the balance team own analysis. What would you say if they think HBE are too good? After all it is purely an educated guess on what we should do.

    Stats allow to be the further from one's personal opinion and thus allow less bias for or against a given army.

    And also regarding the state of this board: I agree that HBE may (MAY) be under-performing but frankly the game is much more balanced than it ever was under GW there is no doubt. The people complaining that it is identical have simply not played a game against V8 deamon uncomped.

    The only difference is that maybe HBE are on the low end while with GW they tended to be on the powerful side. However this is not a sign of good balance even if pleasurable for HBE players ;) .
    (\_/)
    (='.'=)
    ('')-('')
  • Good morning,
    I am just looking for some opinions on my personnel thoughts for the HBE book. I have been comparing the units we have to the DE book for comparison considering they should be similer, DE more frenzy style and HBE more tactical and reserved. ( personal option )

    When comparing dragon princes to dread knights I'm not sure why Dragon princes cost so much more.
    1 less resilience but better move yes, prices fear in combat where as knights don't, prices one attack unless charge where as nights 3 (given 2 are only S4 )
    Yes granted prices gain the 6+ ward save ( and Fireborn which I think is useless and should be an option) but knights are reroll ing 1s !!! Combined with lighting reflexes that's deadly

    my opinion would be to keep prices the same cost but 2 attacks base.
    Also to make dragon forged amour an option ( +2 points per model to take dragon fired amour ) and have them at 52pts not 54
    This will make the cost difference more fair and make the extra cost of the prices make sence

    I have some other ideas such as sea guard as special and swordmasters with bodyguards to MoCT but first I will see responce to the above.

    Kind regards,
  • Borjnfer Wraith wrote:

    If each AB community gathered their own data and processed it using a common guide of course, then this would go a long way in helping the decision members get the detailed data they need to do their job.
    That is called Tournament Analysis. It is a Team of its own.

    Yes it would greatly help if I had one member per Army. (The Collection and entering of the results Form the fluxx card app, Form each armyboard and the lists Form the armylist Board would go much better than.

    So feel free to apply.

    SA-ACS

    Coordinator Translation DE

    United Nations DE

    Playtester

    Info over the work of TouS and where to report your lists.

    The post was edited 1 time, last by Just_Flo ().

  • Wardy wrote:

    Good morning,
    I am just looking for some opinions on my personnel thoughts for the HBE book. I have been comparing the units we have to the DE book for comparison considering they should be similer, DE more frenzy style and HBE more tactical and reserved. ( personal option )

    When comparing dragon princes to dread knights I'm not sure why Dragon princes cost so much more.
    1 less resilience but better move yes, prices fear in combat where as knights don't, prices one attack unless charge where as nights 3 (given 2 are only S4 )
    Yes granted prices gain the 6+ ward save ( and Fireborn which I think is useless and should be an option) but knights are reroll ing 1s !!! Combined with lighting reflexes that's deadly

    my opinion would be to keep prices the same cost but 2 attacks base.
    Also to make dragon forged amour an option ( +2 points per model to take dragon fired amour ) and have them at 52pts not 54
    This will make the cost difference more fair and make the extra cost of the prices make sence

    I have some other ideas such as sea guard as special and swordmasters with bodyguards to MoCT but first I will see responce to the above.

    Kind regards,
    hmm I'm not sure I agree with the above as I actually think knights of Ryma are pretty good. Also, the dragonforged armour is one of their best assets! Play against alchemy once and you will agree. I would suggest they are one of the best cav in the game (hmm maybe a bit optimistic but they ain't Bad for sure) Going first in combat, str4 basic, hitting on 2s, wounding on 2s on the charge (especially if you have Ryma banner) with a shed load of AP. Not much stands up to a large unit charging let me tell you.

    Also from a fluff point of view, isn't dragonforged armour kind of their thing? That's like saying give Phoenix guard their aegis save as an option?? Especially at 2pts extra per model I don't know many people that wouldn't take that upgrade as standard anyway when taking them.

    I think the main problem with them is the same as all cavalry in the game - every wound is precious. But wouldn't say that's a particularly HBE specific problem...
    Visit our Youtube channel, Blog and latest project for book reviews, video battle reports and much much more!

    youtube.com/channel/UC6RqtSiZ_YeZP_Sif7Cf65Q
    proxytablegaming.blogspot.co.uk/
    The Wooden Spoon Cup!!!
  • New

    fair enough.

    Just thought the fact that the dread nights are 80 points cheaper for the first 5 was weird as I think there better. Considering they also go 2s and 2s. And resonance 4
    Hey I might turn to the dark side !!!

    Ryma banner on them is good but I prefer the stalker standard so I can mess with my opponant and charge through ruins

    I've played infernal dwarfs once and Dragon forged amour was a beast tbf
  • New

    The problem of the Rhyma knights:
    They are a little too expensive, just watch Cold Blood Knight (DE) or Knight of Chaos even see those of the Old Saurians.
    Give their attack 2 all the time please.

    The dragon armor is good but its capacity only serves one part out of 6, their mobility is computed by their R3 I find, especially that they are mostly locked in contact. And they have the same problem as all the other cavades so why are they so expensive to buy?

    I know, we have to put the units back in the context of the army, but here there is a problem all the same. Are we so little to see him?

    Hbe 340 (68) / 54 points, R3, Svg 2 (dragon armor), 1 / 2A f4 / 6, 1A f3 mount, M18, 1 ruler to touch more easily.
    DE 255 (51) / 50 points, R4, svg2, 1A f4 / 6, 2A f4 mount, M 14, 2 special rules to touch more easily.
    SA 280 (56) / 55 points, R4 svg 2, 2A f4 / 6, 2A f4 mount, M14, blood cold,
    Chaos 285 (57) / 50 points, R4, Svg1, 2A f4, 1A f4 mount, M14 special shield, yes it has mandatory options but I look at the basic characteristics.


    The elves are effective in attack so a cost in consequence (which is normal) but a goblin resistance so the figures are removed by whole handles.
    This is the main problem of balancing elven armies.
  • New

    PTG-Lucky-Sixes wrote:

    Also from a fluff point of view, isn't dragonforged armour kind of their thing? That's like saying give Phoenix guard their aegis save as an option?? Especially at 2pts extra per model I don't know many people that wouldn't take that upgrade as standard anyway when taking them.
    True, but where does it say dragonforged armour = +2ppm?

    Contrary to 4+ Aegis from the FW, of course dragonforged is only useful once in .. 10 matches? 1 in 20?
    +17 ppm/+85 points per unit of 5 compared to DE cavalry (which has way better grinding power) is quite a lot. And Yema + Witchcraft can make them M9 as well (even M10) and have Strider.
    Playing/painting: SA, DE & HbE ..
  • New

    Teowulff wrote:

    PTG-Lucky-Sixes wrote:

    Also from a fluff point of view, isn't dragonforged armour kind of their thing? That's like saying give Phoenix guard their aegis save as an option?? Especially at 2pts extra per model I don't know many people that wouldn't take that upgrade as standard anyway when taking them.
    True, but where does it say dragonforged armour = +2ppm?
    sorry was just going on the suggestions from wardy. Not an actual thing.

    And the Phoenix guard comparison was just to highlight something that was associated with the unit rather than how useful it is. I do conceed the PG have a better more useful aegis save...
    Visit our Youtube channel, Blog and latest project for book reviews, video battle reports and much much more!

    youtube.com/channel/UC6RqtSiZ_YeZP_Sif7Cf65Q
    proxytablegaming.blogspot.co.uk/
    The Wooden Spoon Cup!!!
  • New

    Alzam wrote:

    Teowulff wrote:

    +17 ppm/+85 points per unit of 5 compared to DE cavalry (which has way better grinding power) is quite a lot. And Yema + Witchcraft can make them M9 as well (even M10) and have Strider.
    FYI, I don't see how Dread Knights can get Yema. If someone played them like that against you, I think it was a mistake :/
    I think the mistake was mine - with Yema General, it was Dark Raiders in 1.34 that could be Yema and +2 from Witchcraft att (for M12).
    So nope, not Dark Knights .. :oops:
    Playing/painting: SA, DE & HbE ..
  • New

    actually during list writing i would love to have dread knights! :)

    hehe nah funny things aside, actually i think rymaknights are a tick overpriced or should just give them 2 attacks back, i mean we have 2 good magic picks with moct and cosmo that can buff strengh....then they would have at least some potiential to grind if first round goes to bad dice...
    I really love the Unit and tested it at least 10+ games since Beta but they just dont seem worth their points on the battle field too often ;(
  • New

    Wardy wrote:

    Just thought the fact that the dread nights are 80 points cheaper for the first 5 was weird as I think there better. Considering they also go 2s and 2s. And resonance 4
    Just small question about that Resistance 4. Did you consider Aegis against Flaming Attacks & Aegis (6+) in this comparison? IMO this is better if opponent has Alchemy, Flaming Attacks or S6 & is comparable if not.
    Current armies:
    WDG
    SA

    -------------------------------------------------------------------------------------------
    Armywide Signature Spells - Check! Maybe you could add something more? Success! We got Hereditary Spells!
  • New

    yes moment does balance the two units a bit
    In regards to the aegis 6+ and for both thing

    Hands down Ryma knights are tanks against flaming
    But how many times do you honestly face flaming attacks ?
    And it might just be me but I have never made a wards save of 6 on my princes

    Think that R4 is better as stops strength 4 men wounding on 3s


    But hey hoo
    I have resolved my thoughts and my opinion is made that I believe that for the points the princes should have two attacks base
    Or the price for 5 should be reduced by like 40 pts or something

    I still take a unit of ten with stalker standard in every list for giggles charging through forests