Confirmed: perk gems stack until 2000%

[DarthDuck]

This is a question that I've seen floating around and I finally set aside an entire afternoon to test it by opening the same chest over and over again and then reverting to a save point. I did two tests, opening 50 chests each time.

Constants: OG, Rebirth 216, Saydhi's large chest, gold +37%, and all OG items unlocked.


Test 1 - 500% gem+ (Adamant Armor)

gem results: 16
cash results: 34
item results: 0


Test 2 - 6,036% gem+ (Adamant Armor, Jade Ring, Apoc, Nodus, and Ruin)

gem results: 43
cash results: 7
item results: 0


analysis: With 500% gem+ only 32% of my large chests gave me a gem. With 6,036% gem+ 86% of my large chests gave me a gem.

6,036 / 500 ≈ 12.
43 gems / 16 gems ≈ 2.7.

This means that the 6036% was exaggerated and functionally only 2222.22%, when compared to using only 500% gem+. This indicates that gem+ (and all other perk gems) cap at either 2,000% or 2,222%. Considering that a large sample of chests was used in this experiment, it's unlikely that this was just based on streaks in the data and that the gem+ really could offer a 6,000% benefit. Further, I parsed subsections of the data, and the the proportions were nearly the same in each subsection as well as the same as the total sample of data.


comments: I noticed no difference in the quality of gems despite the vast difference in gem+ that I was using, implying that gem+ only increases the likelihood of gems dropping but not the quality. The only gems of remote interest were '6 shield on combo' which I got three of, 47 health, and 183 magic. Then after 5 hours and 24 minutes (given I was multitasking), I got a 181 attack gem on chest #97. It's a good thing I decided to test 100 chests! Anyway, the fact that 95 of the gems were completely useless (the only other attack gem I got had 100 attack) confirms my already strong conviction that gaming chests to try and get what you want is a complete waste of time. In the time it took me to get that good attack gem (which was completely lucky considering how close it was to the end of my experiment) I could have killed Raidriar up to three times, completing 15 rebirths.

Edit: Something else that can probably be drawn from this data is that with everything unlocked it's more unlikely to get item results. I remember getting items much more frequently while I still had some unlocked, and 0/100 item results is quite interesting considering that many complain about getting too many item results from large chests. It was not part of this test, but in general when I just open chests without any perk gems I rarely get items. And after documenting this 0/100 item result I can safely say that it doesn't just seem that way.


conclusion: Gem+ certainly does stack, but apparently not past either 2,000% or 2,222%. Also, the amount of gem+ does not improve the quality of gems, only the quantity of gem results. This means that you would be wasting time by putting on more than your best two gem+ pieces before opening a large chest, assuming that your total is less than about 2,000%.

[BLADET]

great study and well presented, thanks Darth! can say it makes me a bit sad tho. gem farming simply ain't worth it and there ain't much (more) to it (then what we already know). we've probably reached the cap on most of them already, and the ones left is probably just +1-5 more than what is listed now. makes me wonder what Chair thought when they placed those caps, feels like their underestimated either the game or the players. many of the best gems, or very close to them, can be found in early rebirths. indicating they had NG-players in mind. but at that stage one doesn't have the money, unless of course, one either buy IAPs or hack it...

once again thanks Darth!

edit: one thing that I do like is that I've got some really good gems from medium chests lately. like Magic +180 and Attack +193! maybe drops from these will continue to get better with rebirth #? then again, have a feeling that cap on lvl 604 that NG-players have mentioned probably effects OG too, but as titan lvl 1300... just have to wait and see, still far away from that.

edit 2: as a comparison I just got a Magic +113 in the large chest in VoT...

[Aussi3]

Your'e an IB2 scientist!! Great finding's and results, Nice work again

[=Silver=]

Good study, Darth
But methodologically, conclusion about "exaggerating" is made basing on the proposition that you have a linear function in approximation. It could've been just a non-linear, diminishing returns dependency (a kind of parabolic one, e.g.)

You need at list third independent value between 500 and 6000. The more - the better. If only one - take a middle value. It'll allow to build a curve and extrapolate.

You still may talk about cap, if it'll be a kind of asymptotic function

[DarthDuck]

@Silver, I was waiting for someone to call me on sweeping that point under the rug, but let's pretend that I opened 50 more chests in the exact same location with 2,000% gem+ and I got exactly or near-exactly 4 times as many gems as from the sample with 500% gem+, would that be enough to convince you that it was linear until intersecting a hard cap?

Also, since I was already able to walk away with a nice attack gem for my efforts, would you be comfortable if I did another test on a different chest such as my next time at the Saydie chest (so I would have a shot at getting a 6 shield on combo for my Tetrach)?

While that's not ideal it should still be controlled enough to address the possibility of diminishing returns right? Either way please allow me until Jan 21st to do a follow-up test because I have some important rl deadlines to deal with.

PS. I used the word 'exaggerated' before I crunched the numbers in the analysis stage (expecting diminishing returns).

I wasn't seeking after a cap; that discovery was quite unexpected. All I wanted to do was figure out if they stack at all and I did determine that.

For now I think the thread is still appropriately named since it doesn't say "only stack to 2,000".

[=Silver=]

@Silver, I was waiting for someone to call me on sweeping that point under the rug, but let's pretend that I opened 50 more chests in the exact same location with 2,000% gem+ and I got exactly or near-exactly 4 times as many gems as from the sample with 500% gem+, would that be enough to convince you that it was linear until intersecting a hard cap?

If these values would lay on one straight line (or about) - yes

Then again, it maybe infinite diminishing, asympthote (if not linear) or just "artificial end", set by coders. If third, it may contain its own algorithm including random generator, or analytic equation, or both.

Also, since I was already able to walk away with a nice attack gem for my efforts, would you be comfortable if I did another test on a different chest such as my next time at the Saydie chest (so I would have a shot at getting a 6 shield on combo for my Tetrach)?

While that's not ideal it should still be controlled enough to address the possibility of diminishing returns right? Either way please allow me until Jan 21st to do a follow-up test because I have some important rl deadlines to deal with.

PS. I used the word 'exaggerated' before I crunched the numbers in the analysis stage (expecting diminishing returns).

I wasn't seeking after a cap; that discovery was quite unexpected. All I wanted to do was figure out if they stack at all and I did determine that.

For now I think the thread is still appropriately named since it doesn't say "only stack to 2,000".

Darth, you did a great job. I'm not trying to say that it proves nothing. I think you proved the stack principle, and had the evidence on diminishing. The question was just about the nature of the cap... If it exists

[ztripp]

@ Darth great job and good find!!! Thanks for your hard work and time spent!!!

[mammoth]

wow very impressed. thanks darth. i always wonder why i keep getting cash results with a combined +5500% perk gems. gem farming becomes tedious after a while because as yo noted, it's very hard to get good gems.

[Harry-Son]

Only just seen this thread, but incredible finds Darth and well worth mentioning. I too keep getting poor gems such as X amount of damage on p.parry and X amount healed on p.parry. Those sort of terrible triangle gems that I have no time for. Still holding out for the 6 shield on combo gem.....

[nwcs]

I've sure learned a lot reading this forum! I was getting rid of certain perk gems but now I think I'll keep them. Just wish the rebirth in OG provided better gems in the store...