This unfortunately is true, Generally the chance of getting a good hero, and a good quest for that hero/race and a reasonable goal ie not something that was going to take 10+ turns was pretty low.
But an average hero and a good quest was much more common and you could happily turn that into a 3rd or 4th.
Armor list is on this page. It is what they consider a teir list
So I have received data from hsreplay (EDIT) for latest patch. Dont know if its considered âreliableâ info though - it should be data for top 5%
Just posting it because I was reminded by Scr0tieMcB I have made calculations based on outdated ancient data 
T1 - 1
T2 - 8
T3 - 22
T4 - 59
Chance to roll atleast 1 hero of T1 + T2 (2 slots / 4 slots)
19,1% / 34,89 %
Chance to roll atleast 1 hero of T1 + T2 + T3 (2 slots / 4 slots)
57,28 % / 82,19 %
Chance to roll all heroes from the worst T4 (2 slots / 4 slots)
42,72% / 17,81%
I am kinda shocked that site considers so many heroes as underperforming though.
Do you think it would actually be better to do âhero rotationsâ here and there? Like rotating the worst performing heroes every month or something like that. So instead of 60 underperforming heroes we would have only like 45 instead.
Donât suppose you want to run the numbers based on the armor allotment in the link I gave ya? Also, want to share the math on how you are getting to these numbers?
No problem. But I prefer to actually see the performance of the hero based on data than on armor HS gave them.
Based on armor
T1 - 26
T2 - 23
T3 - 14
T4 - 10
T5 - 7
T6 - 4
T7- 6
I will try to show you how its calculated but I suck at explainingâŚ
Lets say we want to know what are the odds to roll T1 hero with battlegrounds pass:
But first we are going to check the odds NOT TO roll T1 hero:
All heroes - N=90
T1 heroes - m=26
Hero slots (4 with battlepas) n=4
We want to know the chance NOT TO roll T1 hero - k=0
Then just use any âcombinations calculatorâ out there. Its way easier than to calculate it yourself.
Usually it has 2 rows - ânâ - number of objects and ârâ - sample size
So we are going to find these 3 numbers. Lets name them C1; C2; C3
Now we finally get to probability:
P = (C1*C2) / C3 = (1 * 635376) / 2555190 = 0,24866
So probability of NOT ROLLING T1 hero is 24,866 %
So probability of rolling atleast 1 t1 hero is 100% - 24,866% = 75,134%
Ok, so now what are the totals for T1-T7 for 4 picks and 2 picks? Where I am going with this is I want to make a own forum post on the odds linked to this thread (or perhaps you do it since you are dude doing the actual work). It is VERY interesting to see how much of an actual gap there is between Pay to Win vs Free to Play.
TIer based on armor value (number of heroes / number of total heroes)- chance to pick at least one hero of X tiers - 2 vs 4 hero âslotsâ
T1 (26/90) = 49,66% / 75,13%
T1-T2 (49/90) = 79,53% / 96,04%
T1-T3 (63/90) = 91,24% / 99,31%
T1-T4 (73/90) = 96,60% / 99,91%
T1-T5 (80/90) = 98,88% / 99,99%
âŚ
Though no idea how is their armor related to performance. For example Gallywix is Tier 1 based on armor but tier 4 (worst tier) on hsreplay based on avg.placement. Similar to Toki being tier 3 based on armor value but being the very best performing hero on hsreplay based on avg.placement.
Are you using free data or paid data?
The average placement changes dramatically based on which players are using the hero.
I got paid data for top 5% MMR
Thanks! Carry on!
Itâs not too shocking because players are still trying to figure out what works and what doesnâtâŚso the only Tier 1 and 2 heroes are the ones that are so good that even experimenting gets you to top 4 easily.
There are some heroes that are extremely good and will net you Top 4 often IF you know what youâre doing and not experimenting (like Greybough) but HSReplay will show it as a bad Tier.
We should probably wait until the data starts smoothing outâŚhowever, with that being said, thatâs not going to change the obvious imbalance that 2 heroes vs 4 heroes selection creates. Your numbers likely wonât be too far off.
For example, your 42.72% chance of getting 2 bad heroes with 2 selections is what Iâm currently showing. Iâll have Tier 4 minions for both selection about half the time. Sometimes Iâll get a Tier 4 and Tier 3, but Iâm rarely going to see a Tier 2 or Tier 1.
Also, you have to consider that Putricide MUST be in all lobbiesâŚso the chance youâll hit that is 1 in X no matter what, where X is the number of open Hero slots from all players (16-32). 1 in 16 (if all players donât have battlepass) is not bad odds at hitting a PutricideâŚwhich is easily Tier 1 or 2 IMO
ALSO, itâs important to note that a Hero might be Tier 1 or 2 ONLY if a certain combinations of tribes are banned/not banned.
For example, Greybough is top tier IF mechs and/or beasts existâŚdespite the fact that HSReplay shows him Tier 4
Thatâs the cool thing about the premium membership, it shows you the Tier rating after bans.
These stats are a big yikes. You absolutely get annihilated on the chance of getting a t1 or t2 hero.
Ironically he is super busted with Undead due to the taunts being created and allowing minions that are triggering off deaths to stay alive and activate more often. I have done if twice now and itâs really strong as it is with the beast leaper builds.
Update: I just got Greybough offered twice in a lobby with Quillboars and that is a ridiculously disgusting build. Reborn into taunted rebornâs with a golden Scrapsmith means endless gems to slap on the undead and negate their whole weakness early on. A single greatboar is all you really need and a Sinrunner for the late game. If you can manage to find a Sister that is even better. Greybough is a beast in this meta with undead.
ATM the only Tier 1 Hero is Putricide but that was always known because of the broken combinations you can create. It really should just be Tier 0.