I think the answer is to just do it, Nike style.
You don’t try to predict the future. That’s impossible. Instead you want to manipulate the present, to mold the future towards your desired outcome. You see something that could work, you slap it in there, and you analyze the results. If it’s a hit, great, if it’s a miss, also great. Progress isn’t measured in one single direction.
Actually, I could use your help, now that we’re back talking about stats again.
If a card in my deck is 2nd or 3rd highest played winrate card, but I play it 10% of games, is that worth having in a deck?
How does one even evaluate this to answer it?
Should I calculate all played winrates, multiply by them by the frequency of playing them and then re-order to see for myself?
I made my first homebrew deck (not exactly, but it’s 13 cards difference from the original archetype), and I wanna make it perfect.
It’s already…suspiciously strong.
Played winrate is a garbage stat. Use drawn or mulligan winrate.
Same thing in this case - I only keep the card if I intend to play it.
Actually, I was referring to mulligan (Kept) winrate
The thing is, I refuse to keep it despite the stats. The card is rarely playable, and most of the times it’s a dead card in hand. I only keep it against specific matchups because I know from experience they are weak to that card.
But in this case, intuition isn’t enough. I’m rocking a 60% winrate deck in top 100-s and I still can’t break that cursed barrier and stabilized sub-100.
I need drastic changes. I constantly experiment with piloting and mulligans and not a damn thing changes. I’m still stuck in same ranks, 50 or 60% winrate it’s all the same.
I need changes in my rules of thumb/heuristics/decisions I make in a split second and never think about. I need to re-program everything I knwo about this game.
If this is what you want, then:
Have you considered drastically changing that behavior?
I mean, I get it, HSreplay tells me Innervate is “18th best out of 18” to keep while Eonar is “1st out of 18” and I’m like, you’re kidding right?
But, figures don’t lie and liars don’t figure. If I start with Inner and draw the 10-cost, unless it’s prior to me having 9 mana (which admittedly happens stupidly quickly, stupidly consistently) then the 10-cost is still dead. But if I start with the 10-cost and draw the Inner, it’s still dead. But if I start with neither and draw either, it’s still dead.
So what this tells me, is that it’s less about mystical odds that shift from complex math, and rather more about perception and tilt. If I keep Innervate I’m thinking, “man, I get to cheat something out” but then I draw crap that I still can’t cheat, or don’t WANT to cheat, etc. Instead if I keep the 10-cost I’m thinking “well, that’s 1 less dead card I could draw when I really need something playable.”
The odds are reflecting binary outcomes. Win, or Lose. To get the full story you have to dig deeper. “I won this game because I played this INSTEAD of that” is a far more interesting and important observation.
Yeah, no, I’ve progressed beyond that point of wondering why the stats are as they are.
The thing is, this is my first homebrew deck, which means I’m in the uncharted territory. The stats? All me. Nobody else is playing my deck. It took 50 games for it to start displaying stats at all, and the stats it displays are all my creation.
The sample is too low and it’s biased by own knowledge about the game. For them to show anything meaningful to me which I didn’t know already, I have to misclick mulligan 50 times or so.
The card in question isn’t played in any other deck anywhere on the ladder. It’s like it doesn’t exist. Which is why I need help figuring out what to make of it.
EDIT: Also, I made the deck yesterday when I was encountering too many druids, got tilted and made an anti-druid deck. 7-0 against druids means the project was a success xD
Surprisingly, I got 57.5% winrate on the deck (was even more, but it doesn’t matter, the sample is low), so I can’t just give it up. But today, no druids on the ladder, and I’m losing, and it’s time to make changes again.
I’m finally beginning to realize why all streamers keep making changes to their decks and have low samples.
If you give me a decklist, I can tell you what HSGuru mulligan winrate data says is the optimal mulligan strategy for said decklist. Ain’t nothing but a spreadsheet math problem.
I mean, I can do it myself, if you tell me how you intend to do it to make it more viable
Give the man a fish, feed him for a day; teach a man to fish, feed him for a lifetime.
Do you want me to bother you to help me every day when I have to update my deck?
P.S. I don’t have a clue how you intend to do this, when we both agreed the card stats depend on the deck it’s in and nobody plays my deck.
P.P.S. And the mulligan strategies depend on the matchup, that’s a lot of spreadsheets xDD
Yeah, that’s a good point. I hadn’t thought about that.
Honestly, I’m just trying to make sense of the little data I have access to, because apparently, I have a 24-hour window with any iteration of a deck before it becomes a 50% winrate deck, which doesn’t leave enough time for racking up data.
Adapting to all changes fast and unpredictable is the only strategy I haven’t yet tried to break out of my MMR. I’m trying it now.
I blanked on how this was about you trying a homebrew. My bad. In that context, I have no valuable insights to share.
It’s a literal paradox. Play something long enough til you have enough data and experience to make calculated changes, but lose the surprise factor and adaptability, or make quick, blind changes and hope something sticks.
It looks like it’s back to playing meta decks on auto-pilot for me. I don’t think I can be as lucky as I was yesterday with whatever I decide to change today, because today I’m not seeing one matchup all the time, it’s pretty evenly spread.
Not to mention my class (Warlock) is already favored against Druid in general, so when you attempt to make an anti-druid warlock deck it’s much easier than to do the same for Mage and Shaman. I have historically bad stats against those classes.
They are incentivesed by their audience to entertain them. I think there is little benefit to swapping from a game theoretic point of view.
If every one is swapping decks constantly the meta will never arrive at a Nash equilibrium, that is, where there is no apparent benefit to change game play strategy. Note my wording, I said apparent benefit, because in this case short of an omniscient being, one can not know for certain there is no benefit.
If there is no equilibrium, and the decks in play are highly polarized, then you have the classic rps meta, sans equilibrium, so closer in principle to a game of musical chairs.
It’s good to come up with rigid rules for when you break, and try a change to the deck you are playing, because this allows you to later update your rule formally. This will, over time, let you develop a reasonably efficient scheme to update the homebrew you are playing with respect to what you are encountering.
Most importantly, however, in my opinion there should be two broad modes in running a homebrew.
The first mode is experimentation. This allows you to run a deck with suboptimal choices for cards provided you want to test the win condition concept. In this mode your primary goal is to gather information rather than win games. The question you want to answer is, can this deck win 50 percent of games or more against the most popular meta choices. To answer this with suboptimal cards potentially running in your deck you have to review each game and consider, is there a card not in my deck such that if it were it would drastically aid in this situation? A lot of these end up being defensive cards such as aoe or zillax. Similarly, you want to track dead cards or cards that are win more that don’t contribute.
The second mode is refinement. Provided you think you have something from the experimentation stage, it’s time to apply all that information you gather during that stage. You now need to figure out which cards need in (generally harder) versus which cards need out (generally easier.)
This process can be repeated ad infinitum, honing in on really good homebrews that rival some of the best t1 decks in any meta.
That said, it’s not for the faint of heart, especially the experimentation stage, be prepared to lose hundreds or thousands in rank. But in my opinion that’s the funnest part of the game, drop the rank while you experiment and refine then hone in on that optimal mix and sky rocket back up.
Most of them are garbage under various contexts or at least most of them are extremely misleading if you don’t put a lot of extra thought about what manipulates them and makes them misleading.
I’ve talked before about the blatant example of the Shopper DH; the weapon tutoring made Shopper the “worst stats” card when it was the best card; but what that did can happen more subtly.
E.g. are you sure card A is still good if you just removed card B that draws; maybe card A is bad if the deck draws less because card A is too high cost; and that’s just 1 simple example.
I don’t think they’re misleading, they just mean what they say. Shopper had a terrible mulligan winrate, and that makes sense — you don’t want it in your opening hand. Mulligan winrate is an important stat for card power, but it has diminishing returns because, well, mulligans exist. You can remove low mulligan winrate cards from your opening hand and replace them with random draws, which are probably higher mulligan winrate. So as long as about half of your deck has good mulligan winrate, it’s okay if the other half are low — you’re not losing much power, and you can nearly ignore the lower mulligan winrate cards.
Regarding drawn winrate, the draw from Umpire’s Grasp did count towards that statistic. Draws NOT from Umpire’s Grasp also counted towards that statistic, but opening hand doesn’t count as drawn, so pre-nerf Umpire’s Grasp would usually be 4 draws with a 2/27 to 2/22 chance of each drawing Shopper; on average, about one third of a Shopper would be drawn during those four turns. And I don’t think that Shopper draws on turn 5 or later are in any way unfair to include.
Shopper had a pretty decent drawn winrate. Was it high compared to the best drawn winrate cards in the deck? Not at all. But it wasn’t the worst card in the deck either.
Umpire’s Grasp was a flat out more powerful card than Window Shopper pre-nerf. The stats there weren’t misleading, they were accurate. I understand that Grasp was forced by deckbuilding to draw a reduced cost Shopper, but just because a Shopper is stapled onto a Grasp doesn’t mean that Grasp isn’t the stronger card. The entire point of including Shopper was to make Grasp stronger, which it did.
The only thing you are kinda right about is that when you force a tutor, the stats don’t tell you what is being tutored. If you ran Grasp with non-Shopper Demons, then your Grasp would probably do worse than the stats indicate. However, this means that if you don’t nerf Grasp, but you do nerf Shopper, then as soon as a playable Demon is in the meta, Grasp will just be overpowered with the new Demon instead. Nerfing Grasp is the only solution, because the alternative is to nerf or “prenerf” (never even print in a powerful form) EVERY card with the Demon tag.
A bit side question but it is related to the “learn” part of the topic.
How to identify that meta have changed when meta reports won’t show it immediately? In other words how many games would mean you see a meta shift and not just unusual decks showed up for few games?
What I mean here is with BSM we’ve seen many aggro decks to counter them so Insanity warlock went to field centered option adding another domino and throwing out Fizzle. On the other hand yesterday I’ve seen lots of druids and warriors which have been mainly targeting BSMs as it seems just to outlive their burst turns and slower Fizzle setup was better again. So at this stage of establishing meta - what’s your number of games or loses to make a decision to adapt your deck or switch to direct counter?
You seem to have deleted the part of the reply that I said Shopper was just a blatant example; there’s a more subtle cases that the individual card stats are misleading; e.g. if a random card in the deck draws more then it can make the draw stat of expensive cards that you don’t want to be drawn early less bad but you don’t know exactly how much bad before you remove the draw cards but you can’t remove the draw cards and check the stats again because you only have the current stats and you didn’t make a deck that does that and you may never make it and even if you make it who will play it to collect stats etc.
Sidenote on Shopper: you probably checked Shopper stats from this meta; they are extremely contaminated compared to the original Shopper DH stats from Whizbangs because more Demons are now in the pool to be discovered (I remember all of the card’s stats being bad); but it’s not very important anyway (it was only an example that makes it clear).
I have a more monstrous way that contaminates those stats; a discover card that may produce very rarely a card that affects the stats of other cards already in the deck in a big way because it may be rare but also too powerful; good luck figuring that out intuitively or at least good luck expecting people to not make mistakes at that.
I literally skipped this part xD Straight to the 58-60% winrate deck from scratch xD
But as I said, it wasn’t that hard, since I had to counter too many druids, and my class/archetype already beats Druid by default
The problem is when the classes I have to target have an edge over my archetype, and I think it’s time I accept there are other classes in the game xD
Yeah, it’s not about losing ranks, it’s about…Idk, uncertainty intolerance, ig?
I just can’t stand not knowing if the deck can go the extra mile…like, if I go 5-20, is it the deck’s fault? Is it my piloting? Is it lowroll streak?
I’m sorry to have to say this, but too many people fall into a trap thinking their decks are good or bad based on 10-20 matches which could all be a highroll/lowroll, and then they go changing it if it’s negative winrate, or stick to it if it’s positive
That delays their progress, because they won’t even question their own piloting skills and won’t change anything to improve them.
That’s mainly why I never experiment(ed) with homebrews so far, and why I wouldn’t recommend it to anyone who’s lower ranked than they’d like to be.
If y’all just playing for fun, then sure, it’s a great idea.
I wish I could answer this simply, but I can’t. I main Warlock’s 3 decks, and this is my first Warlock homebrew. In other words, I don’t really change decks.
I do update them each change of the meta, true. That happens when I see many new decks and my winrate consistently falling, with games being tougher than usual. Then I stop and ask myself: “what caused this? Which cards are abused in many decks and warping the meta?”
I find the answer to that question and adapt to counter those.
For example, before the mini-set patch, my sludgelock was running 9 mana Zilliax for heals and a Yogg-Saron, mainly to steal other Yogg’s or Unkilliax’s. It worked!
But then it got out of date, and then I had to replace those cards with better, faster early game - so I added Party Fiend and 2x Thornveil tentacle, borrowing them from Insanity (it’s always a good idea to copy what best decks do, if you can). I believe that’s the version you tried.
And now, all those versions are on hold again. It’s time for another big update. This time, I suspect I’m swapping classes entirely.
I might be stubborn and resistant to change, but hey, I can’t stand 6 months of playing without any improvement. I’m literally stuck.
Also, you’re around 5k legend, if I recall correctly? That’s around 1-2 days behind top 200, and probably 3-4 days behind top 50 meta, so if you want to get ahead of your opponents, read what top 200 players write on the forum or better yet, turn on a stream or two to check what streamers are playing.
Yeah, correct. I’m going to play tuesday/thursday for 4 hours straight since next month so going to start fighting for better ranks.
Taking some break playing BGs until then to relax a bit. 
Thanks for the tips.
If a deck variant, relative to it’s parent deck, has significantly different win rates against, lets say, the top 10 most prevalent meta decks (something you can test easily enough in theory), then listing them as separate decks makes sense.
That’d be my general rule of thumb.