安裝 Steam
登入
|
語言
簡體中文
日本語（日文）
한국어（韓文）
ไทย（泰文）
Български（保加利亞文）
Čeština（捷克文）
Dansk（丹麥文）
Deutsch（德文）
English（英文）
Español - España（西班牙文 - 西班牙）
Español - Latinoamérica（西班牙文 - 拉丁美洲）
Ελληνικά（希臘文）
Français（法文）
Italiano（義大利文）
Bahasa Indonesia（印尼語）
Magyar（匈牙利文）
Nederlands（荷蘭文）
Norsk（挪威文）
Polski（波蘭文）
Português（葡萄牙文 - 葡萄牙）
Português - Brasil（葡萄牙文 - 巴西）
Română（羅馬尼亞文）
Русский（俄文）
Suomi（芬蘭文）
Svenska（瑞典文）
Türkçe（土耳其文）
tiếng Việt（越南文）
Українська（烏克蘭文）
回報翻譯問題
討論區規矩與守則
檢舉此篇文章
Of course we do NOT see that. We see increased win% on Elf Grass. I'd suggest your suggestion is the warm feces of a much-unburdened equine.
Nope, Mike knows nothing of the sort. It means, very clearly, that across all the games played Skaven have enjoyed the greatest success on Elf Grass as compared to all other types of stadiums, including those without any upgrades.
Highest win% across all games played by all relevant teams in BB2. If it was only "scrub teams" that were benefitting then the non-"scrub" teams should have a higher overall win% than those teams.. unless you're now going to claim that you win less, as an "expert" coach than scrub teams and scrub coaches, with the same roster.
That we're seeing the highest win% for Skaven among all possible stadium upgrades (or lack of) with a specific stadium upgrade does not fit with your claim that "experts" like you don't benefit from it... it can suggest that your non-standard form of play doesn't benefit from it, but that would fall under "things nobody really asked about".
Your being awful confrontational for a guy who is citing stats he can't show. I don't know who you are enough to say you are a liar or you are the greatest statistician ever. I'd just like to see the data before bowing to it. We are talking about empirical evidence so I don't think that's much to ask.
There has been no investigation into any relationship between estimated skill levels and Elf Grass utility... which no doubt looks like an open door to people who want to make ♥♥♥♥ up about that relationship. Given that the data includes games involving literally everybody, that includes all the Bobby Fishers and Elmer Fudds of blood bowl, and it still puts Elf Grass at the front.
If we divide matches played by Skaven teams into groups based on the stadium upgrade in effect during that match, we see significant differences between most of the upgrades and the win%s. If we decide we want to treat all the data as though it were a sample (which, in this case is NOT a necessity, but an argument can be made for it) then we can't say with 95% confidence that Elf Grass is the de facto best, but there's still a good reason to choose it.
Consider these win rates based on stadium upgrades, in COL, including all matches (that means concessions as well), reported as though they were samples, and covering the 95% CI:
Those are the upgrades that overlap the win rate of Elf Grass such that it fails to be ahead with 95% certainty, but as you can see, each of those (with the exception of Elf Grass) also fails to show, with 95% certainty that it is superior to no upgrade at all, and each has a higher degree of variation in outcomes. All the upgrades not shown don't even come close and, in fact, suggest themselves to be worse than having no upgrade at all.
So, if for some reason you play Skaven and avoid using gfis at all... you're probably not playing them to best effect, but to suit your style of play you'd consider the wizard upgrade instead, or possibly the sandwich shop or altar... or even nothing.
In more than 6 years of online Blood Bowl there have been exactly TWO people who have ever used any data for actual statistical analysis. Two. Plenty of people have demanded access to data and yet of those plenty, only two did a damned thing with any of it other than holding it in one hand like it was a talisman and continuing to make ♥♥♥♥ up as though it provided them with legitimacy.
You do not have to believe the data, but choosing not to will simply make you wrong.. just as you can choose to believe that the earth is flat or that vaccinations cause autism. It's totally up to you.
I don't suppose you could investigate the correlation between average gfi count and win %?
From there, it wouldn't be too much of a stretch to speak to the utility of elven grass or astrogranite.
Unfortunately we cannot get any more specific than that. The relationship may not be strictly linear, but the numbers we do have suggest it is primarily positive.
I'm almost afraid to ask, because unlike the other guy who uses statistical analysis , it's pretty clear your intention is to belittle and insult the people who don't have access to the stats instead of teaching and and making a point.
However, I'd still need to know what the ranges are for and the sample sizes with the different enhancements before declaring the Elf Grass the unquestioned only best answer.
If you actually knew about stats instead of just having them on you, you'd know that. You should also know the difference between lacking understanding and lacking information.
You are being a total doosh about this sir. First off you are preaching to the choir on elf grass for Skaven, it's the choice I was going with even if I think Royal Box has a good case. Second, just a month ago I made the statement that orcs were a "top 5" team and Dode politely told me I was wrong and showed me the stats. I immediately gave him the reason.
You OTOH, are insulting my intelligence without providing all the stats and eating me alive for not immediately believing your methodology and interpretation of stats you are not completely showing me.
I'm a reasonable man who posts here to discuss, teach and learn. You aren't coming across as anything more than a message board bully.
It's population data not a sample you ♥♥♥♥♥♥♥ ignoramus. I'm not sure how many times I'm going to have to tell you that before it gets through your thick skull. I'm humouring the idea of treating the population data as a sample by reporting the 95% CI as though the data for all games isn't the data for all games, and you don't need to know the size of each group to know if its valid because the size of the group is taken into account in the generation of the 95% CI ranges!
Declare whatever you want. You don't have a firm enough grasp of the topic for your declarations to be worth anything.
Actually, if YOU knew about stats you'd know that what you said was silly, but clearly did not and still don't understand why it is. You lack both understanding AND information, though I've given you all the information needed for the topic.
Good grief, your momma should wash your dirty mouth!
Alright, my statistics aren't great, but I do believe that the bigger ranges in CI for non elf turf are due to less sample sizes, correct? A sizeable difference probably. So there's one contention. I mean the Magician's shop is within striking range isn't it? How many skaven games are there with magician shops is actually a valid question.
Another point, this collection of data will not tell you how the elf turf stacks against the Royal Box. The Royal Box doesn't affect the game pers se, it's effect is post game and in the team building aspect.
So despite your rude, self absorbed rhetoric, there is indeed a conversation to be had at least about the Magician shop and the Royal Box, which were two of the suspects raised before your rant.
I'm having a rough time trying to simultaneously have a valid conversation about this and trying pull the stats from voddomike to a barrage of insults just so I can learn more.
Stadium Enhancement combos fascinate me and it sucks to have my first big convo on them go down like this, but so is the interwebzz.
I'd say the above statement sounds right. Again I'd like to see the actaul sample sizes not deluded in CI, but Voodoomike has clearly done some homework that pretty flatly states that 1.) Skaven do indeed like Elf turf and I'll venture we can infer that 2.) It is the intuitive choice for good skaven players.
I say good skaven players because the fact that any stadium enhancement outperforms no enhancement suggests that mostly good Skaven coaches build their teams to the treasury level fo building stadiums.
One thing I'd consider suspicious is that any stadium enhancement could actually increase a win% by 4-5%. That's a lot, way more than you would think one or two extra passed av rolls a game would gain. So there appears to be other factors at work, maybe it's the increased games against non bash teams that also favor the elf turf for their home stadiums and the Skavens are visiting. I don't know.
Wider intervals are due to larger intragroup variance and thus are "due" to any and all sources of variance. What you need to understand (among the many things) is that the range represents our best estimation of where the population value lies, with the greatest likelihood being at the midpoint of that distribution.
"Striking Distance"? No, how many games is not a valid question - the number of games played is factored into the width of the intervals, so what you're really saying is you want to eyeball the number and make random feelzies about it based on... nothing.
IF, and its a big ♥♥♥♥♥♥♥ IF, we want to treat this data as a sample of "all games that will ever be played" rather than the population of all games that have ever been played, which is what it is, then what we can say is that we cannot be 95% certain that Elf Grass's superior performance is statistically significant when compared with the Magician's Shop, Squig Sandwiches, or the Nuffle Altar. However, what we can say is that Elf Grass's superior performance IS statistically significant as compared with no stadium upgrade at all, which we cannot say about the other three.
So, even if we do decide to treat all data to date as a sample I'm still completely comfortable saying that Elf Grass is the way to go for Skaven, based on the data.
I'm afraid you're wrong on this point, too. We can say that Royal Box doesn't have a mechanical effect on match-level play, but we certainly cannot say it doesn't have a measurable effect on the match outcome... and it seems to. Royal Box win% is 42.0% - 48.9% which just barely scrapes the bottom of no upgrade's 95% CI.
One of the biggest mistake you, and people like you make is believing that things are wrong simply because you do not understand them. You don't need to know why there is an effect, but not knowing why does not negate the reality of the effect. Maybe it's psychological, causing opponents that would have otherwise conceded to the Skaven team to play out the match, reducing the certainty of a positive outcome for the Skaven team. We simply do not know the "why".
You can discuss anything you want. You can discuss the merits of astrogranite as a choice too, even if the data says its a crap choice. Nobody is forcing you to heed the data, and certainly you don't have a firm enough grasp of the data to use it in the first place so make up whatever argument you want and play it out with other people who want to do likewise if that's what makes you happy. Just understand that you're spitting into the wind in that case.
Look, I have at no point said I simply don't believe, I have only repeatedly asked for more information about the statistics compiled and the methodology you used to make the blanket statement that Elf Turf is superior for Skavens. I was imitially dubious because you mentioned "1069 coaches disagree" and that seemed like a small sample size. However, little by little, insulting rant by insulting rant, I am getting a better understanding for why you are so confident in the statement. I still don't see the need to resort to the name calling and such. Yes, I also took some statistics courses in college, I passed, it was a long time ago I don't use that nowadays. I'm actually interested in the work you've put in even if this is pulling teeth from an absolute troll.
Now, I'm still not clear on the Royal Box and why the straight up w-l statistics on it are valid. To measure the effect of the Royal Box you would have to take the win loss record of Skaven teams that own the Royal Box, not play on the Royal Box randomly, because its effects are long term for a team that plays in it often.
Also going back to the Elf Turf, more than interested in GFI stats, I'm interested in the road games against non bash teams that also favor Elf Turf and if those matchups are warping the results. Not that that would invalidate the findings, but it would be interesting to know.
So, 1069 was me poking fun at his "intuition" on the topic.
That said, 1069 would be a pretty good sample size in anything that dealt with sample sizes. The relevance of sample sizes is that they affect the effect size of test data and without sufficient sample sizes we are unlikely to find any statistically significant effects. You can't eyeball the numbers and have a hope in hell of knowing if they're too small for that... and you can't ever say that a statistically significant difference between distributions is invalid because of a small sample size - if a small sample size was having any effect, it'd be to make it impossible to find a difference.
People on forums are very fond of saying "your sample size is too small" but the vast majority of the people who say that don't have the slightest clue as to what they're talking about. That has been YOU the whole time in this thread.
Now, you can complain that I'm a big meany head, but here's a simple fact: you've been fighting me tooth an nail on this stuff based on nothing more than your pride. You're dead ♥♥♥♥♥♥♥ wrong, and you take offense at being told so, but you are... and you put up a fight about it, despite being wrong, but are only really fighting over having taken offense at my demeanor. That means one of us is focused on facts, and the other on butthurt. Which one of us is really being "self-absorbed" and rhetoric-focused?
Nope, it has nothing to do with long-term effects, it has to do with match level effects or it wouldn't show an effect at the match level... which it does. Because it's match-level effect we do not want to limit ourselves to Skaven who own the stadium upgrade.
There's no GFI data to be had so... nothing to be done in that area. The data covers all games against all sorts of teams, so I don't see that there'd be any "warping" of results. Similarly, the question was about COL play, meaning we don't need to or even want to control for composition, which is something we do want to do when we're talking about things outside the context of COL.