I also noticed he was a Merfolk and tested two for awhile. He's pretty good, actually. The number of games he saved me from random Dreadnought/Tarmogoyf/Tombstalker/Marit Lage Token/Other Random Threats was actually what prompted me to start testing Echoing Truth.Originally Posted by chokin
Dude. Seriously. Your data is retarded and has zero credibility. You would need a -ton- more results to even remotely begin to compensate for the margin of error. You'd probably need at minimum a documentation of every Merfolk list to ever make top 8 in any tournament of reasonable size ever to even begin, and even then the margin of error would crush any viability a strictly numerical analysis would have. Your top 8 lists don't factor in the matchups that each of the decks faced, which makes a much bigger difference than the inclusion or exclusion of 3-6 cards. And they -can't- factor in the matchups, because then you'd need numerical data to substantiate how favorable or unfavorable an actual matchup is, and this would vary beyond statistical probability on the Merfolk player's card selection, the opponent's card selection, and who won the die roll.
In summary, while data is neat and not to be dismissed, any attempt whatsoever to defend any point related to deck construction strictly on numbers is going to fail horribly without at least 100 times the data that you've collected.
Ever watch those car insurance commercials on TV, where Geico says "People who switched saved an average of $156", or where Allstate says "People who switched saved an average of $212," or wtfever? This is a perfect example of how meaningless statistics can be when they don't tell you the whole story. The simple reason these statistics are dumb? People rarely if ever switch car insurance companies if it isn't going to save them a significant amount of money. It's not worth the hassle generally. On top of that, we have no basis of knowing how the statistics were gathered in the first place.
So "Kira is better than Jitte because my data says so" is the most ridiculous argument ever. If you think Kira's better than Jitte, argue why or get a ton more data. Don't bombard us with error-filled statistics.
Your argument here, while partially correct, actually contradicts what you're saying. To whit:
This is correct, it's not intrinsically good in and of itself. However, by only running 14 Merfolk, you basically are ensuring the fact that you would be better off running 14 other creatures.Running merfolks isn't intrinsically good
There's a reason that every successful Merfolk decklist you will find plays at least 18 creatures who have "Merfolk" in their creature type. This is because without a decent amount of support, the +1/+1 you're getting for each "Lord" creature in play is not a worthwhile effect.
I know you and Forbiddian can't stand anyone saying that either one of you is wrong, but I really don't have to playtest that list he came up with to tell you there are a few things wrong with it.The only relevant question is if the deck's good. If it's not possible to build an optimal deck and have it be merfolk-filled at the same time, then so be it.
Also, I'm not disregarding the statistical data that he's using to base these claims off of. What I am doing is saying that it doesn't seem to me like that decklist is the best extrapolation of the things that data set is telling us. And I also agree that the data set he used is small enough that people shouldn't be deluded into thinking it's somehow the "final, authoritative word about Merfolk" or anything.
I think experience has shown most folks that Dreadnought is not really the best way for this deck to address its problematic matchups. That doesn't mean that Dreadnought isn't a fine card on its own, and it doesn't mean that the meta and the card-pool might develop in such a way that Dreadnought would be an optimal card for us to include. But I really don't think that's where the meta is at right now. I think the white splash is way better than Dreadnought, for the problems facing Merfolk players in the current metagame, as I understand it.
Argue with any part of this that you want to.
Bless your heart, we must consider Blue/White Tempo's strategy and win percentages in an entirely different deck thread. -4eak
Oh my god DDK... I... I... Completely agree with everything you just said. Well said, I will try to remember this in the future
And yeah, forbiddian's decklist is pretty bad. 3 Kira, 4 dreadnaught, 1 trickbind and <16 merfolk are definitely poor decisions. And yes, those are all comments based out of tournament experience with Dreadfolk.
First, there are *MANY* problems with your analogy. Your analogy reads: "In the past, statistics have been used to trick people who don't know anything about statistics. [Implied: Therefore, you're trying to trick us.]" You can actually identify a critical flaw with your example (bravo), but you failed to point out one with my methods. You leave it to the reader to assume that there's something wrong with my analysis. That's an extremely dishonest debating technique, and it's logically unsound to boot.
There are obviously minor flaws in my methods, but I'm reasonably certain that, given the data that I have and rational time constraints on combing the data for outliers, I'm reporting the most accurate data possible.
The second is: I'm not trying to make a sale, so I'm not selectively reporting data that support my cause. It seems like you and SJUD have the delusion that I, in some way, benefit from this (other than the benefit of trying to help out Merfolk players). SJUD even accused me of tampering with the data, or selectively reporting data to "prove [my] point." What the fuck does that mean? My point that I think you can use the concepts of mathematics and statistics to analyze the performance of different cards over time? That point? Yeah, I'm really going to prove that by tampering with data.
The results confirmed some of my guesses and went against other of my guesses. For example: I thought the white splash would dominate, but it's only slightly better than average. I thought Kira was a bad include, but the data show it's not with a fairly high certainty. I thought that Jitte was good maindeck and bad sideboard, but the opposite seems to be true. I also thought Dreadnought wasn't good, but statistically it's the best performing variant. It also confirmed some of my other analyses (not using statistical analysis). Like my justification that Divert was much, much better than Misdirection. I wouldn't bore you with the other examples.
Due to the large sample sizes and/or large deviations from expected, I can be confident in those conclusions.
I explained exactly how the statistics were gathered, in my post. Did you not read it? Obviously I'm not going to bore you with the technical details (it's not a lab report), but from what I wrote, you can easily be repeat the experiment yourself if you'd like to confirm any of the data points. You could even just pick a few at random, confirm that they're accurate (or not), and then use statistical analysis based on that to calculate the probability that I'm lying! WOW! Statistics is amazing (although actually, you'll get a div#0 error...).
It seems like an awfully dishonest debating technique to accuse me of not being transparent with my data collection. It's not even pay-per-view premium or any bullshit. It's free. Right on deckcheck. Right like I said. Try reading.
I have no idea what you're talking about. The standard deviation on a proportion is a well-known and easily calculable statistic, even by you, if you'd like to check. Again, some of the deviation is explained by random distribution, and some of the deviation is explained by the cards in the deck. There are numerous statistical tests that you could use. F-test, Chi-square are two that you could easily be familiar with (although both are approximations). Or you could go through and calculate r for all cards vs. not cards.
I'm curious. You said that I would need 100x as much data in order to get accurate results. Where did you get that number?
You certainly didn't get it by having even a rudimentary understanding of statistics and doing a calculation. I'm over 99% confident, for instance, that Divert in the sideboard does better than ~Divert, with the sample size that I have right now. I'm over 99.9% confident that Divert in the sideboard does better than Relic of Progenitus in the maindeck. Again, that's just with the sample sizes that I have now. Is 99.9% not enough for you?
If it's not, then holy shit, you should NEVER "waste" your money on prescription drugs or medical care.
I dunno, what really bothers me about your debating technique is that you don't seem to be bothered by facts or even having a coherent argument. It just jumps from half-baked lie to quarter-baked conclusion.
Sorry I posted the decklist. It was just that some trolls felt left out of the topic because it involves 10th grade Economics or Statistics. So some trolls wanted some decklist to test, and I tossed a bone under the bridge.
I even said, "This is all the best performing cards mashed together" (or something to that effect, I think I even used the word "mashed" which can hardly be construed as me widely espousing the deck as anything more than what it is: a curiosity), and I didn't make any comments about its overall performance, since I hadn't tested it or anything.
It was probably a mistake (to post a decklist, even one posted so obviously for novelty purposes), because as many trolls as there are who freely admit that they don't know what's going on, there are twice as many trolls who don't know what's going on, but will use anything they can possibly get their grubby paws on to claim that mathematics is useless. Or something. I don't know what they're trying to prove.
Anyway, depending on the level of confidence in the given result, we can use the data as anywhere from explained by the null hypothesis and the gospel according to Feynman.
But I have no idea why people somehow think that some playtest result would be more accurate than 70% or 80% confidence, and that's quite on the low end of anything I'd consider to be valid data to use in an argument. Playtests results are like witness testimony. It's by far the least accurate data, and far below the threshold of confidence that any scientist would back, but it's still the most-impactful data used by juries across the country.
Well then we at least agree upon that Dreadnought isn't worth it and i took away valuable slots. And that's exactly what I am saying altough he came as a surprise (at least month ago it was) and it kills quickly cause even Zoo doesn't always have a solution at hand. Anyway the Mono U version still has a bad matchup against rg and Kira doesn't help enough including all your sideboard stuff. Don't forget that Zoo players have relevant SB cards too. Our creatures are generally outclassed and burned and it is often just a matter of time when the Zoo player gets board presence. I still render this matchup around 35/65 pre and 40/60 postboard for Mono U. With white there are better options to fight the rg matchup and I have won a lot of matches because players just folded to Absolute Law.
This argument is at its base flawed. You are assuming that every good player is also a good deckbuilder/innovator. That's simply not true. If you look at people who consistently top 8 (LSV, Nassif, etc) and listen to interviews about the decks they play, many of them are playing something someone else designed. They are great players and can learn how to play a deck. That doesnt mean they know how to build it.
Back to the Dreadfolk stats thing: its an interesting idea. I really like Dreadnought in merfolk but I dropped it when Sovereign came out. It does help against zoo and goblins though. Maindeck goblins doesn't have much to fight it besides racing and maybe Stingscourger. While zoo has more to fight it, it also makes them keep Pridemage in the deck which is by far the worst creature they have. And after SB we can always side out Nought for other things and use the fact that they SBed hate against them.
But maybe a creature base like this would be better since we have so few folk:
4 Cursecatcher
10 Lord
4 Dreadnought
2 Kira
That way we still abuse the Lord synergy but also have the Dreadnought plan.
I'm completely in love with this decklist varient of merfolk by Patrick Dierbach:
http://www.deckcheck.net/deck.php?id=30285
It comes as close to running goblins as i've seen a merfolk deck do.
The banneret is just an additional explosive measure a la Goblin Warchief and combines well with the Reejerey, while the Tidal Courier serves as an additional draw a la Goblin Ringleader. Now where's my merfolk lackey so I can dump standstill entirely!?
I think 3 Force of Will and 3 Daze isn't really good... It's more random than a good build of Merfolk.
Zoo<3
The big differences between goblins and merfolk is that:
Goblins has a tutor
Goblins has haste
Goblins has spot creature removal
Those three are pretty much why merfolk can never do what goblins does as a strict aggro deck. Courier is nowhere near as good as Ringleader with no haste and one less power. And without warren weirding/gempalm incinerator. Like it or not, Merfolk just can't run as many merfolk as goblins can goblins. However, we get other types of fun, such as Force of Will and Standstill. Those replace goblins spot removal and their ringleaders. That being said, I think there *are* good ways to do this deck with a higher merfolk count. Courier *could* replace adept, but at 2 mana less and 1 power more, I still feel like adept is better. Courier is going to net you one additonal card on average at best.
Why would you ever want to drop Standstill from Merfolk? It seriously a huge part of what makes the deck stupid good.
Merfolk and Goblins play nothing alike. Goblins has, on the whole, stronger creatures, and creates a lot of "Oops, I win" scenarios. Goblins, on the other hand, has no answers to certain things, where Merfolk always has Force, Daze, and if you run it, Echoing Truth. Merfolk plays the Lord strategy better than Goblins because all of Merfolk's Lords not only grant the +1/+1, they also all grant a form of evasion. Merfolk has an ungodly midgame, but suffers a bit on the early and late game. Goblins has a terrible midgame, but a potentially lethal early game and a fantastic late game.
What version of Merfolk would be the best for a tempo thresh/burn/random meta? I'm starting to think that submerge might be pretty good in the side and the waterfront bouncer/echoing truth ideas sound pretty good as well.
The version I'm debating playing is the mono blue version with the full tempo package... 4 stifle, 4 waste, ~3 back to basics. I've thought of running 2 Kira main, but they seem so bad against sweepers that I don't think they would be good enough in most matches...
I also personally feel that Jitte isn't a good card for a midrange deck. When I've played the card in those types of decks, I almost always wanted it to be something else when I drew it. It takes so much mana to play and attach that it just felt like it was slowing me down.
All those superfluous Merfolk and he's not even running the good superfluous ones! Where's Tidal Warrior, Sygg River Cutthroat, and Waterfront Bouncer!? Fuck, for that matter, where the hell is Merfolk Sovereign? I'd seriously run any of those four I just mentioned over Tidal Courier and Stonybrook Banneret, at least they actually provide some utility. And Sygg has a fatter ass than Courier, costs half as much mana, and will probably end up drawing you as many cards as you would have off that Courier (which, in that list, will draw its owner two cards less than 50% of the time).
In all seriousness though, I think that list is running a few too many fish and not enough utility spells. And the extra fish he's using aren't the ones I would use if I wanted a really creature-heavy build (which I don't think is close to being the best version of this deck, but I can respect someone for playing the style of deck they want to play). Also, Force of Will is completely awkward as a three-of. And I'm not a huge fan of the three Standstill approach either.
Bless your heart, we must consider Blue/White Tempo's strategy and win percentages in an entirely different deck thread. -4eak
If Tempo Thresh is really big in your meta, you might be better off staying Mono Blue, since you have a much more stable mana-base and you don't have to fear getting your fetches Stifled.
On the other hand, the Mono Blue version has a somewhat hard time with Burn and "Random Aggro" decks, which it sounds like comprises the other part of your meta. So you definitely need a plan there.
I think you have to think long and hard about whether or not the full-on mana disruption route is going to be your best choice. While it can be a blowout against Tempo Thresh, mana disruption does jack shit against Burn. The Stifles are just going to be blank cards during your game one against Burn if you run them. I can't speak to what it's going to do against the "random aggro" without knowing more, but I suspect Stifle is not going to be great in that matchup either.
We sort of need to know how many Tempo Thresh players vs. how many Burn players there are to make a spot-on call. Because neither of them are blowout matches in your favor, you probably need some decent sideboard hate against both of them.
On to specific sideboard cards...
Blue Elemental Blast and Hydroblast are probably good (assuming you're talking about Canadian Thresh when you say Tempo Thresh), since you're going to be facing a lot of red spells, and they come in for many matchups.
Submerge is definitely good against Thresh, and it might be useful against the "random aggro."
Phyrexian Dreadnought out of the sideboard is a decent option if you want a "surprise factor" play, since it can outrace Burn (assuming you're running four Stifles). On the other hand, Thresh might bring in Krosan Grip against you, which makes Dreadnought less good.
Threads of Disloyalty is decent versus Thresh, and probably pretty good against the "randoms."
Back to Basics is a real judgement call. It's gonna be good versus Thresh, but it sucks versus Burn and other mono-color decks. And it can occasionally screw you off your own Mutavaults.
Relic of Progenitus is decent versus Thresh, but if you don't need graveyard hate for other matchups I'd hesitate to recommend including it. It's a judgement call.
Hell, if you want to run something really unconventional against Burn, look into Choke, or even Suffocation. (But probably not. Veeery narrow cards, and not even that impressive really. Although I kinda like Suffocation. Cantripping blue direct damage. Lol.)
...Anyways, this is just me spitballing. I hope it helps.
Bless your heart, we must consider Blue/White Tempo's strategy and win percentages in an entirely different deck thread. -4eak
Thanks for the insight. It's much appreciated. I am getting the feeling that my metta is just too random. The decks I am likely to face on any given Friday would be the following;
Pox
TT
Burn
stax
ANT
cb/top
merfolk
random aggro
Divert seems like it may be pretty good against pox, TT, and burn, so I will likely be running that in the side (mabey even main?). Another key matchup that divert shines in is Zoo. If you resolve one, it is probably game. It 2 for 1's them whenever you are able to use it and has single-handedly won me some games against zoo when I was playing Ugwr cb/top.
Welcome to Legacy!
The Meta is just unpredictable. Best ist to scout directly before the tourney which I personally dislike but know that many ppl do it. I had tourneys where I was 100% sure that I could anticipate what was played and failed hard. I also had tourneys where I was 100% correct with my foretellings but just played anybody else outside the Meta. I had tourneys where luck was with me and I played just good matchups without preparing for the Meta at all. Except for very few decks that have a critical mass and really small tournemants I am of the opinion that your personal Metagame cannot be predicted at all. What does it mean. Don't care to much about everything that is not DTB for preparation. Check your local metagame for DTB or decks with a critical mass and prepare for that.
Sage, I personally loved the idea of Divert until I tried it. Once I did, it was immediately apparent to me that even its supposed good uses were not effective. I switched to Misdirection, which has continued to serve me, but these days Spell Pierce is probably the better choice. In particular, Divert and Misdirection suck against some really important spells like:
Engineered Explosives
Aether Vial
Duress
Counterbalance
Chalice of the Void
Red Elemental Blast
"Anyone who is capable of getting themselves made President should on no account be allowed to do the job."
"Politicians are like diapers. They should be changed often and for the same reason."
"Governing is too important to be left to people as silly as politicians."
"Politicians were mostly people who'd had too little morals and ethics to stay lawyers."
I agree and disagree with this. Divert did very well for me in testing against the decks that I wanted it to, but I agree that Spell Pierce is a better card now. Divert can stop removal and even some funny situations with discard spells, but Spell Pierce is your cure-all against all of those things he mentioned. But more importantly, he forgot to mention it stops Volcanic Fallout or Pyroclasms, as well as the entire decklist of Stax, Enchantress, Landstill (if you keep them off of their lands early on), Combo of any sort, and any important spells against ichorid.
I run Divert alongside Spell Pierce, which is just a vicious way to stop any sort of red damage for 1, along with helping out against heavy control decks.
Volcanic Fallout cannot be countered.
Also, Red Elemental Blast can be effectively countered using the Divert by redirecting the REB to the Divert (assuming they don't have the mana; if that's the problem, you're probably in trouble)
@Finn: How would you slot in Spell Pierce? Would you simply do a 1-for-1 replacement for Misdirection? When would you side it in?
Most people blindly suggest new cards for decks. True contributors also suggest what to remove. It's not about what's good, but rather what's better than the current selections.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Reply With Quote