Data, stats and consistency - deck building questions

Multiplayer Discussion
1

  1. How much of the cards have to overlap between two or more versions of a deck before you stop calling it “Deck X”?

  2. Do stats which reliably answer question 1 exist and if so, which ones are those?

I know from experience that swapping only 2 cards can make or break a deck completely, so things that VS report does confuse me.

They stone-cold just called a deck with 2-4 different cards the same way and ranked it number 1 in tier 1 for 2 reports in a row, while the addition of 2 new cards to the deck enabled 3 combos which weren’t possible before, and the fact that you now have more combo pieces to take into account, it changes your playstyle of the deck significantly.

How do we approach that if we want to:

a) learn,
b) talk about statistics and data without too much biases (any biases, which includes blindly trusting data when it’s obvious to some people that those data can’t be trusted blindly).

Discuss only if you have experience with these things. If needed, let the thread die - I don’t need speculation and theorycrafting here, I need something measurable.

TLDR - Sometimes changing only 1 card in a deck of 30 cards can completely change your win cons, playstyles and deck winrates. Do we have reliable measures which can be used to predict those changes?

2

I mean changing one card in any deck can have massive impact. For example swapping out sif in a sif deck would probably drop winrate from like close to 50 to maybe 5 percent.

Obviously an extreme example.

I’m not sure which deck you are referring to in your post, but one interesting deck that had many variants recently was Sonya rogue. That deck went through a slow evolution. It was a deck that focused on the weapon finish initially, then it started running griftah and people realized that could be used as the finisher, then tidepool pupil was slotted in and you had the monster deck that it became. That was an evolution over a period of time. And happened naturally.

So when we talk about a small number of cards changing, it can be extremely impactful.

For example I like to run painlock with hellfire sometimes, and have gotten into top hundred doing so, I sub trolley problem for it. It makes a huge impact because there’s tones of close games where your opponent thinks they are safe cause you can only mostly do board damage.

Teching cards in decks will be forever, I hope, a core part of Hearthstone.

1 Like
3

Yeah, I’ve also found that teching in something helps to refresh the deck and probably catch people by surprise, which is kinda why I’m asking this - I’ve had a lot of experience with doing so.

The example from the OP was about Rainbow Shaman, which, with greedy partner in, got access to 3 new combos, much more powerful (albeit, late in the game, kinda too late for this meta) than it had a month before, even though, somehow, it was top of tier 1 both months.

They even have 2 versions of the same deck, with 3-4 cards diff, and they put those 2 versions as having the same winrate xD

I don’t know what they’re doing, but anyway, doesn’t matter - I meant in general, how would we go about predicting the expected drop or rise in winrates after swapping 2 cards for another 2?

I think I would try these couple of steps (if I had reliable source of data, naturally):

  1. evaluate the cards you were dropping ( drawn/played winrates ),
  2. evaluate the cards you were replacing them with (drawn/played winrates from decks it saw play in, if any exist),
  3. Evaluate the synergies between 1) and the deck and 2) and the deck (if any exist)
  4. Add (or subtract) all of those and reach some sort of a score which could predict the change in winrates.

I’m not sure how one would go about 3) until I see the stats, so seeing some would be great :smiley:

But I don’t know where to find reliable stats for those. That’s an issue. Sample sizes aren’t big enough to check those stats on HSguru for high ranks, and if you check lower ranks instead, you’re risking misinformation because lower ranked players might not play them properly.

4

I feel targeted.

Whatever. You can have a thread (almost) without me.

5

I have been having this issue too playing off meta rogue decks like reno rogue.

I played lamplighter rogue from d3 to legend yesterday and today. It performed pretty decent against the decks targeting mage, but overall it can be a draw dependent kind of deck. The combo is relatively slow in this meta by about a turn or so, but the real problems come from the games where you miss your draw.
The list is pretty tight so its hard to see ways to add draw to the list without compromising your elemental chain. Frequently i find shadowstepping shale spider to be a good play even if it could compromise your early combo potential.

This is one of those instances where it feels like 1 or 2 cards could make a difference in the consistency of drawing combo pieces

elemental

Class: Rogue

Format: Standard

Year of the Pegasus

2x (0) Preparation

2x (0) Shadowstep

2x (1) Breakdance

2x (1) Fire Fly

2x (1) Glacial Shard

2x (1) Tar Slime

2x (2) Flame Revenant

2x (2) Quick Pick

2x (2) Rolling Stone

2x (2) Shale Spider

1x (3) Bounce Around (ft. Garona)

2x (3) Sweetened Snowflurry

2x (4) Dubious Purchase

2x (4) Lamplighter

1x (4) Sonya Waterdancer

2x (5) Sandbox Scoundrel

AAECAYO6AgLuwwWKqAYO9p8E958ElrcF38MF3/4F054G7p4GwZ8GzKIGgKMGs6kGtrUGqM4G7eYGAAA=

To use this deck, copy it to your clipboard and create a new deck in Hearthstone

1 Like
6

2x Pupil instead of Sandbox Scoundrel and just keep shadowstepping!

That’s probably what I’d go for

Sandbox Scoundrel does have some crazy potential when coupled with Sonya, but it costs you 5 mana to get the mini for the combo - how often can you afford to play that without breaking your elemental chain?

I’m just theorizing here, I’ve no clue how that deck feels in game - you probably know how each card “feels” on average, at least that’s pretty much the only thing I have going for me when I have to decide what to do.

There’s probably better options, but who knows?

Honestly, I feel the best when I copy/paste a deck which someone played to hit high ranks, because then I know it works, and if it doesn’t - well, that’s on me. That’s the only way I can comfortably experiment with piloting to hone those skills. If I also experiment with deckbuilding, I can never be sure if it’s the deck, or me, that is bad.

Which is why I need the answer to this question, somehow xD

1 Like
7

You really need the scoundrels for the lamplighter pop-off.

The deck still does inevitable damage with sonya scoundrels lamplighter and even bounce around to push 60

You need it for when druids try to out armor you or the occasional odyn warrior.

Being able to otk with 4 or more lamplighters is key to the deck being any good at all

The tempo and early wide board pressure are reasons i prefer this style of otk deck over others

Its easy enough to just play an elemental off of the initial scoundrel discount.

You dont always have to value playing the frost spells off of the 3 cost initial ele

The biggest issue is if you dont get your draw cards you can be dead in the water

8

Well, you do have 2x pupils for that

It gives you more flexibility and tempo

Against faster decks without armor, you can spend it to copy some card draw to speed up the cycle, while against the slower ones you can patienty wait and build up until you can copy a shadowstep

It’s honestly the only thing I see could be replaced from your deck. If you start ranking up, soon enough you won’t reach turn 6-7 to play your combo at all xD

Anyway, I said no theorycrafting and here I go breaking my own rules

I’m off to test a new insanity version I just lost to, it might be broken xD

9

So the combo goes:

Sonya-scoundrel-2xlamplighter-breakdance lighter-scoundrel- 2x lamplighter

There are other variations of course like scoundrel-sonya-breakdance scoundrel into lamplighter combo

Or for more damage using bounce around off of the 0 mana scoundrel into repeating

So the issue isnt about having enough to do, pupil throws off the ele chain more than it helps really. You just need to be able to draw your combo pieces and the deck can sometimes feel lacking in the draw department not in the potential damage devision lol

But like you said, its hard to find any data for decks with such a low playrate. I think it has a niche currently in this meta, maybe a small one but i see it

10

How? What am I missing here? xD

In order to play that combo, you first have to throw 5 mana on a 4/3 do nothing one turn.

Playing both Pupils costs 4 mana, so even when played both in the same turn, you still have 1 more mana to go

11

You need the scoundrels to pop off with the combo in one turn. Using pupils and shadowsteps you spread it out over multiple turns with lower lamplighter damage output

And you dont copy the lamplighters with sonya so you get less of them that way

You can combo usually in between turns 7-9 if you can draw the pieces.

5 mana scoundrel also allows you to play an elemental along side of it its not that bad.

The gameplan is inevitable over the top damage, the early tempo just gets you to that point. The scoundrel is not a sacrifice, it advances your win condition even if it is a little awkward at times

Removing scoundrel would lower the damage potential of the deck significantly and would also actually make it slower

12

Alright, man, I’ve no clue in that case

All I know is that this talk possibly saved me some ranks - the guy I took insanity from lost 2-3 in a row, he’s tanking

Not a good sign xD

1 Like
13

I actually really enjoy playing vs insanity lock xD. You have to have balls to pull off the plays necessary to win with that deck vs a class like rogue xD

1 Like
14

I mean, it takes balls to play Warlock in general xD

Self-mutilating, living on the edge… my kind of a guy xD

EDIT: yeah, no, the deck is as bad as it’s fun to play. A lot.

EDIT 2: hahahha 0-5 with the deck, but I’m still going xD It’s unbelieveable how unlucky a person can be. In 4/5 games, not a single crescendo drawn (last 14, 13, 12 and 13 cards - well, not last, but the middle ones - if they were last I’d be able to pick them up with fracking)

I just can’t decide if the deck is any good when I’m this unlucky.

1 Like
15

  1. Until the major theme of how to win changes. So if it’s designed to win with sif, it’s all sif mage.

  2. No, not in any published forum. You would really only care how the cards fare in relation to the cards in your deck, so game wide card stats like we have now wouldn’t really tell you what you want to know here. The level of analysis to reliably and validly answer your question here is beyond what it’s worth to data sites.

16

Insanity is still fine but not top tier. Yesterday I’ve lost to painlock, bonk shammy and pirate DH (lol) simply because domino never showed up during first turns but then I won handbuff pala easily plus two BSMs and both match ups are unfavoured for the deck.

So it is less consistent then before - that’s true. Still if you are not dead but turn 5 or 6 you are most likely finishing your combo prep for turn 7 and 8 :slight_smile:

Just in case - I’ve dropped reverberation for two dominos and playing legendary spell instead of Fizzle which actually made me play slower before instead of helping.

1 Like
17

Yeah that’s a good call, all of those changes - reverb was useful in Druid + Priest meta, Fizzle isn’t used since Tidepool pupils came, and Symphony is always good

Yeah, that’s what happens when both you and your opponents lowroll xD

You win unfavorable, but lose to favorable matchups, and since there’s usually more favorable matchups, you lose ranks

But I know to just ride it out without blinking twice xD

EDIT: It’s official, deck is broken xDD I finally won against Thijs, after 5 years or so xD

2 Likes
18

Congrats :slight_smile:

Btw - is there a way to maintain MMR somehow? I’ve went to legend 3000 this month before the miniset just to queue into 5 BSMs on the first day which has driven me down to 4500 and then some magic happened - I haven’t played for two days and found myself being 5999. I’ve went back to 5k after some games and parked it just to open the game again this morning and find myself back at 6k. Is it my current MMR which I seem to have tanked a bit last month or is it down to number of new BSM legends and just this month’s nonsense?

Previously one day without play would be something like +200 at these ranks but not +1k difference evening to evening.

I could even compare it to being 328 rank during Stormwind and I never touched ladder since last win due to everyone in top1k has been playing shadow priest. So during the whole month I’ve went only down to 1300 or so and still got my 11 SB back then. So this is somewhat weird and I have to fix my MMR (that was on another account though so those are not related). :slight_smile:

1 Like
19

It’s a little bit of multiple effects:

a) you hit early legend, so you got a better rank simply because people with better MMR haven’t done it yet,
b) after the mini-set, many players returned to play to try new decks, and many of them played more

Just play :slight_smile:

20

A sample of 1; and you think it’s conclusive statistical knowledge; it’s not even enough as an indication. It’s enticing for bragging for social attention and boosting an ego, but terrible application of math that is ultimately misleading; it’s misleading first to you; you don’t want to base your conclusions on bad math if you want to improve.