Because HSReplay hides their turn length data not in the Meta section, nor the Archetypes section, but the Decks section, it’s very difficult to answer this question even if one pays to get past the paywall. A precise answer is not possible (because HSReplay doesn’t make the turn data available for decks with less than 200 plays), and even a reliable estimate using the popularity of archetypes, the popularity of decklists within each archetype, and the turn data for each decklist is extremely calculation intensive.
That said, I have done what I would presume you have too much of a life to do, and calculated it for you. If you’re a math nerd, here’s the link to the spreadsheet I used:
Popularity data was taken from all servers over the past 7 days, while turn data is since Deadmines release, all according to HSReplay. HSReplay only provides raw turn data per decklist in multiples of 0.5 so it’s safe to assume a margin of error of ±0.25 turns.
Here are the core results:
The median number of turns for the player without the Coin is 8 across all of Diamond and Legend. The median number of turns if you have the coin is 8 across Diamond, and 7 across Legend.
The average number of turns goes down as rank goes up.
Top 1000 Legend games average 7.36 turns for both players.
Legend not-top-1000 games average 7.52 turns.
Diamond 4-1 games average 7.66 turns.
Diamond 10-5 games average 7.81 turns.
The lowest average final turn for any decklist was 6.5 (this time around — I’ve seen 6.0 before about a week ago for Zoo Warlock). Many aggro decks achieve a 6.5 turn game end average, as well as some Garrote Rogue and Anacondra Druid lists in Legend. Combo decks don’t average kills that quick in Diamond, even when it’s the exact same decklist, presumably due to pilot error.
The highest average final turn for any decklist was 11.5, for Big (N’Zoth) Priest.
It’s not quite a useless card. Some Warlock decks have the tools to push average game length to 8.5 turns while maintaining a winrate over 50%.
But it might as well just be a 9 mana 3/2 weapon with “Battlecry: Gain 5 Armor.” Even if you do make it to the turn after you play Jaraxxus, the odds of making it to the turn after that, so an Infernal can attack, are virtually non-existent.
And that’s at D10-5, as you uprank it gets worse.
Edit: I forgot about the combo with Tour Guide. Get an immediate Infernal that way and you might actually get to attack with it. Maybe.
Especially as this has been being contended and exaggerated so much this expansion, I truly appreciate you taking the time to do this research. Some will tell you the average game length now is 5 or 6. It’s very nice to have a resource to point to.
I could estimate Wild as well if you’d like. I’ve been playing the format recently and my guess is it’d be about half a turn faster — but that’s just a hypothesis.
Is there no turn data for lower ranks or was it just not worth the time?
I wonder if the faster games is due to concedes (better players see an unwinnable game more quickly) or just better gameplay. I wonder if there’s ever been metas where things went longer and higher ranks.
There is, but it’s important to understand that the normal HSReplay tools for reaping data just don’t work here and that makes the process far more tedious. For instance, because turn data is stored on the site by decklist and NOT by archetype, I had to click on literally EVERY decklist of Aggro Druid, even if it was off by a single card, then do a weighted average of every Aggro Druid list to get the average turn data for Aggro Druid overall, as an archetype… and I did this four times, for the two Diamond tiers and the two Legend tiers.
More importantly at lower ranks deck variance for sure is a thing but mistakes increase exponentially in game play. As you said it’s just not data that can give you any constructive info.
Exploit Druid is 4,5 and Pirate warrior is 6,0, and q hunter is 6,5 that’s 75+% of the meta in higher brackets^^ (checked fo gold I dont have premium) . The problem is that in wild the skill level is lower unless you play at very high legend than in standard. This leads to even more favourable looking turn lengh, since match lenghes are shorter the better the players are (side note in standard there are more otk decks, thus the lengh doesn’t change much; meta dependent) and in “dumpster” legend there are far more meme decks than in standard.
Another problem with hsr is that they don’t have enough data for wild.
This is actually really sad. It basically means all the really neat, fun and powerful 8-mana and up cards they print are essentially unplayable except for priests and warlocks trying to beat the curve.
The game needs a certain level of aggression to be healthy. But they’ve vomited out so many overpowered 1-3 cost cards nearly every expansion that they have effectively painted themselves into a corner.
I hope AV has some low-cost anti-aggro tools coming for most classes or the problem is just going to snowball.
Its really interesting to see this kind of data provided for the HS community so we know how bad it actually is. Tbh im not surprised its like as bad but turn 8 is still really bad cause it basically prevents any card that costs 5 or more and doesnt immediately do anything or protect you to be useless. Unless your 8-mana elemental that searches a spell summons a Mo’arg, you arent doing anything with it.
Anyways kudos to you for getting this data done which was probably more painful to collect that stubbing your toe against a dining table which inexplicably has knives attached all around the leg you hit and then falling on a pile of legos.
Not a good conclusion to draw. Even outside of ramp Druids where 8-Cost cards are commonly played, we have Doomhammer Shamans holding attack buffs until their second turn with Doomhammer out, we have Mr Smite not just generated by the Juggernaut but in the deck as well, etc. If you had said “any card that costs 6 or more”… okay, that’s close enough to true, but 5s are a little more forgiving.
Just don’t draw more than one of them tho because ain’t nobody got time for that
(yes, me)