Rarity Chances from Weapon Cases

[Crosscade]

Hey all,

Just out of interest, I collated data from several weapon case opening videos online to determine what kinds of chances you actually have to get something good when opening a crate. I only recorded data from videos that had at least 5+ crates being opened, to remove the distortion effect from single case rare drop videos (i.e. you have no idea how many cases they opened to get that one rare drop).

After watching quite a few videos from different channels, and recording 347 drops, these are the results:

• Mil-spec (Blue) Drops: 78.8% (287)
  
• Restricted (Purple) Drops: 16.2% (59)
  
• Classified (Pink) Drops: 2.5% (9)
  
• Covert (Orange) Drops: 1.6% (6)
  
• Exceedingly Rare (Yellow) Drops: 0.8% (3)

There didn't really appear to be any leaning towards StatTrack on any rarity. Of the 364 drops, 29 of them (8.0%) had StatTrack.

The two lower rarities, Consumer (White) and Industrial (Light Blue) only drop from in game drops.

Within each rarity, there didn't appear to be any major weighting, with each weapon dropping a similar amount. The only exception to this was the Galil DDPAT which seemed to be about 50% to 100% more likely to drop than the other Restricted. This is likely just an outlier effect.

There may also be too many Coverts or too few Classifieds in my numbers. Only one video showed a Classifed Deagle Hypnotic dropping but there were 5 Covert AWP Lightning Strikes drops, so these percentages could be out a bit.

Obviously these numbers will not be exact, but given the sample size I believe they should give a pretty good insight into the drop rarity and worth.

If I had to take a guess on the actual numbers, I would assume 79% Mil-Spec, 16.5% Restricted, 2.5% Classified, 1.5% Covert and 0.5% Ultra Rare, and a 8% StatTrack Chance.

I didn't look into the condition of the weapons (Battle-Scarred to Factory New) as that was not readily available from the videos, and wasn't the primary reason for my investigation.

[pinosewarrior89]

SentientHAL eredeti hozzászólása:

What rarity is grey? Cause that's what my Lichen Dashed SSG 08 is.

Pretty much the ♥♥♥♥♥♥♥♥ quality. No joke

[rykzet of spacepickshovel]

75 case, i only have asiimov. nothing more than this.

[b4N3_fTw™]

lol ur the boss man for this u definitly deserve to unlock a rare special item
+1

[Julian]

Hello Everyone,

For those curious: if you were going to open 364 chests yourself, there is a 95% chance of your drop percentage values being within these intervals.

Mil-Spec: [74.65%, 83.04%]
Restricted: [12.42%, 19.99%]
Classified: [0.88%, 4.06%]
Covert: [0.34%, 2.96%]
Exceedingly Rare: [-1.05%, 1.75%]

Note 1: This is called a C.I. (Confidence Interval): Given the OP's data, if this experiment was duplicated infinite times, 95% of them will have their probabilities in those ranges.
Note 2: I'm assuming that each drop is independent of one another, and that the randomization algorithm matches a Multinomial Distribution.
Note 3: Yes, given the sample, it is possible for the rate for the Exceedingly rare drops to be negative. This is clearly unrealistic, so you can just assume it reads 0% instead.

Here's the explanation behind the maths... you can stop reading now if you don't care.

Here goes:

First of all, I'm gonna be using the following terms... google if you don't know:

• Binomial/Multinomial Distribution
  
• Normal Distribution
  
• Maximum Likelihood Estimate (MLE)
  
• Central Limit Theorem (CLT)

First of all, Binomial Dist is just the "trivial" case of the Multinomial Dist..
We can split the different types of weapons as different types of drops, each with their own Binomial Distribution...

For each drop Yi, Yi ~ Binomial(n, pi) i = 1... 5
where pi is the actual (unknown) drop percentage rate
and n is the number of drops

I will use the notation pxi to denote the MLE
By the C.L.T (central limit theorem),
(pxi - pi) / sqrt(pxi(1-pxi)/n) ~ Normal(0, 1)

[insert algebra here... too long so ill skip]

Result: the C.I. is pxi +- c*sqrt(pxi(1-pxi)/n) for i = 1...5
where c is the constant from the Normal Distribution at 0.975 (ie c = 1.96)

For example, the Mil-Spec case,

Y1 ~ Binomail (364, p1)
MLE = px1 = 287/364 = 0.78846
C.I. = [0.78846 - 1.96*sqrt(0.78846(1-0.78846)/364), 0.78846 + 1.96*sqrt(0.78846(1-0.78846)/364)]
C.I. = [0.7465, 0.8304]

Same goes for the others.

Conclusion: This data supports the OP's guess as to what the actual probabilities are.

Real Conclusion: I wasted a good half hour's time. Maybe more.

PS. Let me know if i made an error somewhere...

[Profile Name]

Painiyff eredeti hozzászólása:

Hello Everyone,

For those curious: if you were going to open 364 chests yourself, there is a 95% chance of your drop percentage values being within these intervals.

Mil-Spec: [74.65%, 83.04%]
Restricted: [12.42%, 19.99%]
Classified: [0.88%, 4.06%]
Covert: [0.34%, 2.96%]
Exceedingly Rare: [-1.05%, 1.75%]

Note 1: This is called a C.I. (Confidence Interval): Given the OP's data, if this experiment was duplicated infinite times, 95% of them will have their probabilities in those ranges.
Note 2: I'm assuming that each drop is independent of one another, and that the randomization algorithm matches a Multinomial Distribution.
Note 3: Yes, given the sample, it is possible for the rate for the Exceedingly rare drops to be negative. This is clearly unrealistic, so you can just assume it reads 0% instead.

Here's the explanation behind the maths... you can stop reading now if you don't care.

Here goes:

First of all, I'm gonna be using the following terms... google if you don't know:

• Binomial/Multinomial Distribution
  
• Normal Distribution
  
• Maximum Likelihood Estimate (MLE)
  
• Central Limit Theorem (CLT)

First of all, Binomial Dist is just the "trivial" case of the Multinomial Dist..
We can split the different types of weapons as different types of drops, each with their own Binomial Distribution...

For each drop Yi, Yi ~ Binomial(n, pi) i = 1... 5
where pi is the actual (unknown) drop percentage rate
and n is the number of drops

I will use the notation pxi to denote the MLE
By the C.L.T (central limit theorem),
(pxi - pi) / sqrt(pxi(1-pxi)/n) ~ Normal(0, 1)

[insert algebra here... too long so ill skip]

Result: the C.I. is pxi +- c*sqrt(pxi(1-pxi)/n) for i = 1...5
where c is the constant from the Normal Distribution at 0.975 (ie c = 1.96)

For example, the Mil-Spec case,

Y1 ~ Binomail (364, p1)
MLE = px1 = 287/364 = 0.78846
C.I. = [0.78846 - 1.96*sqrt(0.78846(1-0.78846)/364), 0.78846 + 1.96*sqrt(0.78846(1-0.78846)/364)]
C.I. = [0.7465, 0.8304]

Same goes for the others.

Conclusion: This data supports the OP's guess as to what the actual probabilities are.

Real Conclusion: I wasted a good half hour's time. Maybe more.

PS. Let me know if i made an error somewhere...

Thx for making me do math in my lesuire time!

but on a real note, this is really good, thx!

[Mostlyghostly♫]

This thread is a life-saver.

[HaXdAbAx]

I have opened around 40 boxes and this is what i got that was "special" : gut knife case hardened BS, ak-47 vulcan FN, AK-47 redline FT, Deagle heirloom statrak MW, nova antique MW x2 and scar-20 cynex mw. Feel pretty lucky if.

[Bulleta©™®]

As you should!

[partyboy]

I guess I've got pretty lucky then, I've opened 20 and I've got


Mil-spec (Blue) Drops: 35% (7)
Restricted (Purple) Drops: 25% (5)
Classified (Pink) Drops: 20% (4)
Covert (RED not Orange) Drops: 5% (1)
Exceedingly Rare (Yellow) Drops: 15% (3)

[Festive Wind]

Thanks, I knew the approximate chance of the exceedingly rare item but not of any other weapons. You should transform this into a guide! :D

(If not, I'll make it and list you as a contributor.)

[werx]

Crosscade eredeti hozzászólása:

Hey all,

Just out of interest, I collated data from several weapon case opening videos online to determine what kinds of chances you actually have to get something good when opening a crate. I only recorded data from videos that had at least 5+ crates being opened, to remove the distortion effect from single case rare drop videos (i.e. you have no idea how many cases they opened to get that one rare drop).

After watching quite a few videos from different channels, and recording 347 drops, these are the results:

• Mil-spec (Blue) Drops: 78.8% (287)
  
• Restricted (Purple) Drops: 16.2% (59)
  
• Classified (Pink) Drops: 2.5% (9)
  
• Covert (Orange) Drops: 1.6% (6)
  
• Exceedingly Rare (Yellow) Drops: 0.8% (3)

There didn't really appear to be any leaning towards StatTrack on any rarity. Of the 364 drops, 29 of them (8.0%) had StatTrack.

The two lower rarities, Consumer (White) and Industrial (Light Blue) only drop from in game drops.

Within each rarity, there didn't appear to be any major weighting, with each weapon dropping a similar amount. The only exception to this was the Galil DDPAT which seemed to be about 50% to 100% more likely to drop than the other Restricted. This is likely just an outlier effect.

There may also be too many Coverts or too few Classifieds in my numbers. Only one video showed a Classifed Deagle Hypnotic dropping but there were 5 Covert AWP Lightning Strikes drops, so these percentages could be out a bit.

Obviously these numbers will not be exact, but given the sample size I believe they should give a pretty good insight into the drop rarity and worth.

If I had to take a guess on the actual numbers, I would assume 79% Mil-Spec, 16.5% Restricted, 2.5% Classified, 1.5% Covert and 0.5% Ultra Rare, and a 8% StatTrack Chance.

I didn't look into the condition of the weapons (Battle-Scarred to Factory New) as that was not readily available from the videos, and wasn't the primary reason for my investigation.

This is awesome, I love how somone put some time into this so that others can weight their desision on whether or not to do a 2.49$ gamble.

[The Pink Panther]

i opened about ten cases and only got mil-spec

[Rexwiz]

HiD. KCiV eredeti hozzászólása:

nice so basically sell the cases and buy your skins

Exceedingly Rare (Yellow) Drops: 0.8% (3)... those are terrible odds, you could spend thousands of cash and NEVER get one. Much better off buying what you want in the market. We talking about winning the lottery odds..1 in 14 million. What the ♥♥♥♥ Valve ?

[Rexwiz]

http://www.gambleaware.co.uk/

[Dr. B ©™]

that is a lot of work u have done....nice work~