Gambling and Horadric Cache Legendary Research Projects Continue

Gambling and Horadric Cache Legendary Research Projects Continue

Fans in our Diablo 3 community forum have been comparing legendary drop rates from gambling and Horadric Caches, and the collected info is getting interesting.

The Legendary Gambling Drop Rate thread has a lot of user-reported data which Memster is compiling into a big Legendary Gambling spreadsheet that you can see right there. The only thing holding it back is insufficient data, so do keep track of your gambling trials and hits/misses and add your reports to the thread. FOR SCIENCE.

Jackpot sighted in Clan chat.
Jackpot sighted in Clan chat.
The Horadric Cache data collection thread isn‘t as robust, and no one has turned it into a big multi-colored spreadsheet, and there’s not much point in that now that Blizzard has committed to buffing the legendary drop rates from Goodie Bags… but it’s still worth a look and listen. Most players cite legendary drop rates from Caches of about one in ten, but there are plenty of people with ugly anecdotes of going 50 or 60 or 80 caches in a row without a single legendary item. (At least we know there’s no hidden script running a minimum bag fail-rate algorithm, eh?)

Also note that the vast majority of legendary items found from Caches are the Horadric Cache-only legendaries. It’s not like Caches can pop all legs and sometimes, you’ll get one of the bag-only ones. Those are virtually the only legs you ever get from them, which makes the Ring of Royal Grandeur not so impossible to hunt, since if an Act One cache drops a legendary ring, it’s almost certain to be that one.

Speaking of hidden scripts and odds and drop finds…

Conspiracies Grow Like Mushrooms*

One byproduct of Diablo 3 being online-only and not allowing modding is that the nuts and bolts of the game engine remains out of reach of players. This was not the case with Diablo 1 and Diablo 2, and with the full game data fans were able not only to create some fantastic mods, but to dig through the code and figure out how everything was calculated, including item drops. (Massive tutorial for D2 item generation in the wiki.) With that data, math-savvy fans were able to build bots that could calculate the exact odds of any item in the game being dropped by any target in the game. More to the point, fans found numerous bugs, both large and small, which directly affected item drop odds.

I doubt that many of the RNG conspiracists know this whole back story, but I think the fact that we simply don’t know how the game calculates drops directly fuels a lot of the conspiracies that Xanth documented in a recent article. Does Diablo 3 have secret legendary drop timers that run (correctly or not) from the time you find item X, or that tend to bestow all the best drops in a party game on one character? Probably not, at least not on purpose, but who really knows? The devs? Hopefully, but there were many details (buggy or not) about item drops in Diablo 2 that no one knew until fans found them.

Few of us (I certainly didn’t) realized just how unbalanced and screwy the legendary drop rates were in DiabloWikiDiablo 3 vanilla until we compiled info on thousands of legendary drops from dozens of players, and saw just how common and uncommon some of the items were, and how little the frequency of items seemed to have in common with their power or value. (This was, intentionally or otherwise, a direct continuation of Diablo 2’s drop odds, since D2’s wildly-unbalanced positioning of items in the Treasure Classes made some uniques roughly 1/1000th as common as other uniques of equivalent value.)

So when players compile drop results from gambling and we see that some item DiabloWikislots appear to roll legendaries 1/10th as often as others, it’s probably just DiabloWikiRNG and and the fallacy of small numbers… but I am not at all ruling out bugs in the code. Now if you’ll excuse me, I need to adjust the fit of my tinfoil hat.

* Best in the dark when fed lots of bullshit.

Related to this article
You're not logged in. Register or login to post a comment.

27 thoughts on “Gambling and Horadric Cache Legendary Research Projects Continue

  1. I opened 25 Horadric Caches a week ago and got 2 Rings of Royal Grandeur. Everyone treated me like I had the Devil’s luck.

    They were the only two legendaries that dropped, incidentally.

  2. I personally would like to see the loot drops evened out more since D3 post 2.0 has become more of a linear loot progression since trading has been removed.

    I personally would love to see an option where legendary drops weren’t weighted at all. Say there’s 200 legendary items; give them each a 0.5% chance to drop. This lets players realistically get all items over hundreds of hours of play and removes the “ugh, not another” issue players constantly encounter.

    To recap:

    1) Remove weighted RNG on legendaries.
    2) Make the smart loot system factor in currently equipped gear to ensure smoother upgrade curves

    • “removes the “ugh, not another” issue players constantly encounter.”

      It wouldn’t remove it, but it would statistically lessen it. Some people would still have “bad luck”.

      “Make the smart loot system factor in currently equipped gear to ensure smoother upgrade curves”

      I assume by this you mean that items that you already have on your character should drop more rarely or shouldn’t drop at all. I think that’s a really bad idea. What if you already have a few pieces of your class Set, but they rolled badly, so you want to find better versions?

      It could also promote some really problematic player behavior, because we would have incentives to use less than ideal items to find the ones we want. That’s basically what the MF stat did as well and many players + the devs clearly didn’t like it in the long run, which is why it’s all but gone now.

      • My second point “smart loot factors equipped gear” does not mean the game lessens drops you already have because that would be stupid as you say.

        What I did mean is I believe the game should have a sophisticated system that would for instance, determine my Wand is 1800 DPS currently and therefore drop a legendary wand (when RNG rolled a legendary wand) that would be around 1850-1950 DPS or so. This would be a nice upgrade but not too amazing. Next time around when I got a leg wand drop it would be higher and so on.

        The system would prevent two things: extremely bad Wand drops and extremely good Wand drops. Removing uber drops may seem lame, but one thing that really ruins the reward curve is when a fresh level 70 player lucks out on a top-tier top-rolled Wand, for instance.

        Getting a massive 40% Damage boost with a “jackpot” Wand drop may seem fun at first, but long term this means the next 20 leg Wands that drop will be almost surely inferior to the “jackpot” Wand. Jackpot elation is fun for a few hours maybe, but finding nothing but inferior Wands for the next 200 hours is NOT fun, even if players don’t realize it.

        My system, although more of a “managed” loot system, would still require players to play for hundreds of hours, but it would ensure players would know they would get upgrades if they keep playing. Right now, most players hit a plateau after maybe 300 hours on a character…and the heavily weighted RNG system does nothing but discourage the vast majority of people IMHO.

        In the end I believe D3’s itemization system is too weak to support true weighted RNG systems like D2 and PoE employ. Rather, it would be nice to just ensure we got consistent incremental upgrades to our gear every 20-30 hours.

        • “one thing that really ruins the reward curve is when a fresh level 70 player lucks out on a top-tier top-rolled Wand, for instance.”

          Would it really be better if you’d have to spend the same amount of time to reach that wand, but you’d get constant, miniscule upgrades on the way? Also, what do you consider an upgrade? Sure, in the case of DPS the situation is simple: more is better. But what about individual stats? How should the game know what do you want? And no, not everyone wants the same stats, despite what some people would have you believe.

          “Right now, most players hit a plateau after maybe 300 hours on a character”

          In my opinion, this would happen in literally any system you can think of. The only question is when. Nothing lasts forever. I consider 300 hours game time per character pretty good to be honest. By the time most people reach that we’ll get new content patches and an expansion anyway.

          • First, thanks for commenting and making me think more HardRock. I think the answer to your question, “Would it really be better…”, yes it would be very much better to find incremental upgrades rather than nothing for long stretches.

            Game design is all about psychology and managing expectations. Giving constant fun rewards every 10-20 hours is vastly superior to long stretches of nothing and then a big reward.

            Anyway, all my blabbering won’t change a thing anyway, but I would like the items to scale way higher like the console version does…this along with my “Managed Loot” system would ensure 1,000 hours or more per character of constant upgrades.

            But really…I’d just settle for Blizzard implementing a Set Class drop timer every 20-30 hours…that would at least be superior to nonstop TOTAL CRAP BLACKTHORNES. 😉

  3. RNG is not random, it is calculated. This calculation depends on various factors such as calculation method, CPU type/vendor, server structure, etc.

    My personal experience:
    I was implementing multiplayer mode to one game. I had pool of 4 optional objectives and I needed to pick only one of them. The script was created in Lua, so I used Lua random number generator (math.random with math.randomseed(engine_ticks)). When I tested it I realized that this generator is generating only 3 of 4 numbers (number 3 was missing), so I ran a loop with 1000 iterations. Only 1,2 and 4 showed up. Later I tried to move randomseed() from function body to script header and it worked great (even the probabilities were ~.25 for all 4 numbers).

    There may be many issues with RNG in D3. If you have noticed, e.g. Crit. chance is calculated correctly (this calculation depends on your character instance). But what if drops are based on calculations within the whole server(one virtual server or server shard for e.g. ~100 players). So 100 players using the same RNG may result in not various results. They should test this issue on live servers with many players online (not on their local development server or PC).

    This also may not be the case, but if your character is experiencing the same drops over and over and over (and no other item types), it’s not a bad luck. The RNG is simply “broken”.

    • Whatever you may think, D3 team consists many good coders, so assuming that they can’t make proper random method (which is not so difficult, uncle google is always helpfull) is quite surprising.

      • “Whatever you may think, D3 team consists many good coders, so assuming that they can’t make proper random method (which is not so difficult, uncle google is always helpfull) is quite surprising.”

        Wait did you just answer a post that involved a complex description of script, with “D3 guys are so smart…..they can google how to code?” Holy crap you tard. Quit brown nosing Blizzard you twat

        The fact is yes, I know i get the same drops over and over. Each toon I create gets the same group of gear time and again , I would call that broken. It doesnt happen to everyone, but it seems to start at character creation. Example- My DH gets belts, belts , more belts. Havent seen a crossbow or helm or god forbid, set gear.
        Monk – Bone saber, 8 times….really? I need a chest piece but yet to see one in 100 hours of play.

        • Hands down, best comment ever 😛

          Though you’re right, my answer was pretty lame. What I truly meant is that random is really simple, nothing difficult, though yes, you need to pay attention to what you’re doing. More complex implementations (like ones using Gaussian Distribution) made me google out solution, but still that was not too difficult. Besides abillity to use internet in your advantage is highly expected, no one knows everything. So I should rather say – “Even if they didn’t knew how to do random, they would easily find the solution.” Though it sounds a little like I would have no faith in them, so that was pretty lame.

        • Yeah, I got already like 7-8 Nemesis bracers with my char… meanwhile a guildie scavenged 4-5 Cindercoats. Would be so easy to trade with him :/

      • Of course D3 team consists of many good coders (no question about it) but every one can make mistake. My personal experience showed to me that many companies are testing their products in different environment than production environment (in some cases it is not possible to test on production env., like bank DB for example).
        As I said before, some calculations are correct (Crit. change) and some are completely ridiculous (the-same-legendary 8-hit combo 😀 ).

  4. “if your character is experiencing the same drops over and over and over (and no other item types), it’s not a bad luck. The RNG is simply “broken”.”

    Or that the drops are weighted, which is admittedly the case here.

    Sure, there may be bugs in the system as well. As you’ve said, a programmer can mess up any function, including RNG. However, the closed nature of the game engine makes reliable client-side testing extremely time consuming. We would need controlled drop studies spanning weeks or preferably months to reach definite conclusions about their system. The best we can do is post our anecdotal experiences when we think there’s a problem and then let Blizzard decide how to proceed. They can do quick drop tests like you did with your RNG probably in minutes or at most hours, but we as players can’t. This is thorn in many people’s side, because they don’t trust the devs when they say that everything is working as it should.

    • “This is thorn in many people’s side, because they don’t trust the devs when they say that everything is working as it should.”

      Yup….if there is anything I have learned, its Blizzard loves lip service…

  5. D2’s 1.11b best items were made with runewords which could be accumulated. This meant that even if you had poor RNG with rune runs you always got something and always felt like you were getting closer to godly shiny. When the best items in the game are in 1 bag and no progress toward is allowed then each time you fail you are statistically no better off getting it than you were yesterDayMonthYear.

    Hence the disappointment for many that runewords aren’t in the game and never will be. The only option is crafting as it has replaced runes being the more inventory friendly option. However, it still doesn’t really provide that sense of accumulating toward godly shiny. Crafting system needs upgrading to be the next runeword system, not just a mystic imo. We need a serious content patch. The 1.11b of D3.

    • “We need a serious content patch. The 1.11b of D3.”

      I’m hoping for the same. The way I see it RoS created a good foundation for the game the same way LoD (1.07) did for D2. I think the new features are solid and if we get new Legendaries in the coming months (similarly to the new Runewords in D2) that could keep the game fresh for quite a while. Also, I’m sure a little friendly competition in the form of Ladders will help, at least for some.

  6. Ill start running some bounties and gamble my ass off and update it in this post. On a good day I tend to gamble away a couple of 1000s of bloodshards, so it should provide somekind of sample size.

  7. Weird thing happened to me yesterday. Went trough 5 t3 rifts, full party, no legs. Then i did one t3 solo rift and dropped 6 legendaries in ONE map. Most of them from containers. Hmm. There is definitely something going on with those “timers”.

    • Not necessarily. First of all, we’d need to know how often this happens to players, not just to you, because when tens or hundreds of thousands play the game this is bound to happen to someone.

      Second, the drop timer won’t prevent this from happening, it’s not even meant to. The only thing it does is slowly increase your chance of getting a Legendary after a set amount of time has passed after your last drop.

      • Something similar to Hemlax’s experience happened to me the other day. I don’t remember how much of a “dry spell” I had before it happened, but I got 4 legendaries in one rift map as well.

        My theory is that the “timer system” they have in effect is similar to the DH’s SharpShooter passive. It slowly increases the chance of getting a legendary as HardRock said, but also gives a small time frame before it resets back to the base chance. My guess is that the duration of that period of increased chance is similar to the duration of shrine effects.

        • “My theory is that the “timer system” they have in effect is similar to the DH’s SharpShooter passive.”

          That’s the first feasible idea I’ve read so far. If the timer indeed works like that it’s probably not intended though. Reading Travis’ original description of this new feature I think it’s very clear they wanted this drop bonus to reset after the player found a Legendary.

          Like I said above, there could very well be bugs in the system, but a statistically insignificant amount of players reporting “chain-drops” is very far from a proof. This is definitely something the devs should investigate, if they didn’t already. This new drop timer is a new feature after all, so it’s more likely to have bugs.

    • This is the issue with all this anecdotal evidence.

      Depending on what people *want* to believe they’ll either say:

      1. Yeah, well, drops are still controlled and you getting 6 legendary items that fast happens so little that it doesn’t make a difference in the long run. If you play 9 million hours, you’ll see the patterns. Just try it yourself!

      2. Drops are not timed, and RNG was RNG, giving you a ton of items. Hooray.

      It’s a totally useless discussion, but who doesn’t like a good conspiracy?

  8. Yesterday I started chasing that ring.
    Yesterday I got 2 rings from 5 act 1 caches.

    Lucky me. Now of to find the thorns(got the helm)set and Sanguinary vambraces…

  9. I’m in the no-ring category with hundreds of caches…booo. Bring on the cache buff!

    The feast or famine legendaries is so hard to quantify. Everyone notices the peaks and valleys acutely but our brains aren’t wired to handle all the boring in between that is required to make any realistic conclusions.

    • Yes, I agree that we tend to only remember peaks and valleys. That’s why I’ve just recently started keeping a “journal” of my Legendary drops during my bounty & rift runs. I keep track of the source of the drop & what it is, act/wp/area, elite/mob kills, and time spent. Unfortunately, the kill counts are not very accurate since the game won’t update them without ending the game first. Does anyone know a way to get the game to update the profile kill count mid-game?

Comments are closed.