Doom Bolt Sun Spot Pyromancer

Grim Dawn Classes, Skills and Builds
21

YUP but if one gathers say 2k legendaries, a pattern should start to emerge.

22

Hmmm… Not sure what you mean by a pattern but in general that’s way too few drops I think since there are like 1k high level legendaries. You need tens of thousands legendaries to even get them all.

23

Yeah but still… I have almost 4k >lvl80 legendaries in total and for some I have 1 and others 25 :smiley: (For example that stonefist someone was looking for recently :rofl: )

1 Like
24

And then people make purple only builds, because those are easier to get than MI’s :thinking: :see_no_evil:

1 Like
25

Are there 1k of them? Well if we gather 10xnumber of legendaries, then we just perform a population / chi squared test and determine what the probablity is that we got our outcome compared to equal drop rates.

26

But they include Magi rings :clown_face:

1 Like
27

If you want to know the drop rates, wouldn’t it be easier to look at the game files? :sweat_smile:

28

I mean hunting those last legendaries can make it so hard. For one player it will be some legendary X that they will need 10k drops to get. And other player will have 10 copies of this X legendary in 1000 drops :frowning:

The first player will make a thread complaining that X doesn’t drop and it’s a bug :stuck_out_tongue:

29

But digging through code will not get me any drops!

1 Like
30

Just noticed the ones with 20+ are actually level 84 items. For 94 I have at most 15 or so I think. (Hard to do statistics as I don’t know how to access the IA database :stuck_out_tongue: )

31

Again, imo you’d need like 10 x number of legendaries drops to get them all even if they had all equal drop chance :stuck_out_tongue_winking_eye: so probably still too few drops to make conclusions

32

My goal is not to find them all, because it is RNG. My goal is to see if there is a pattern, if some seems to be more common than others. One can do statistics with what finds and compare to hypothesis. How do you think particle physics works? :stuck_out_tongue:

33

I’m just saying that if you get like 10 copies of X and 1 copy of Y in your 10k drops, your conclusion that X has higher drop chance won’t be very strong. I know it’s not very intuitive.

1 Like
34

As I wrote, it just boils down to perform a chi-squared calculation. Is quite easy to do and interprent

35

What is your goal with it? Measure drop chance of all legendaries or test equal drop chance hypothesis maybe? Do you disagree with anything I’ve said in particular?

For a good drop chance info we would need like 1000 players gathering data maybe :thinking:

36

image
image817×60 5.96 KB

Also it was mentioned here, and I have read it and heard about it so many times - that jewlrey seems more rare than say legs/shoulders/boots

37

Hmmm… a goal like this

seems ok for one player. I thought maybe you wanted some more granular / specific conclusions.

1 Like
38

As an example, I have now rolled a regular six-sided dice 60 times (10x number of outcomes) :game_die:

  • 1: 9
  • 2: 11
  • 3: 10
  • 4: 9
  • 5: 8
  • 6: 13

We now compare this data with the hypothesis that each side has equal probability.

Chi-squared value is 1.6

D.o.f is 5

p-value is 0.90

Which means that the hypothesis that each side has equal probability is valid with 90% certainty which is pretty good considering we “only” had 10x number of outcomes sample size.

no lol then I would need to stay in SR until GD 2 is released :smiley:

39

:+1: I’m ok with that. I thought you wanted to make a useful drop chance table for all legendaries or something and just wanted to spare you

40

You’re not going to get 10x the sample size in legendaries though :sweat_smile: sadly