Is stats pulling rigged?

  • Yesterday i bought a number of EOJ items from the npc in Chassizz trying to get certain stats. In theory there is 1/4 chance to get a needed stat on each items. In practice there were only 5 "good stats in at least 28 pulls. Someone told me he had even worse luck with mass pulls. It's harder to provide reliable statistics on shell items but a number of people mentioned quite disturbing "coincidences" there too.


    So the question is what are the real probabilities for each stat in eoj, shell, honor, pom pulls?


    Please note the emphasis in the previous sentence. All statements about database values, lua files etc are simply irrelevant. Players get actual items with actual stats, not someone's assumptions about how the game should work.


    @Game Maters. Could you please do some tests? Create a char, refill its currencies as necessary, make hundreds of pulls and share the statistics. Many players will be grateful for this.


  • RoM Welten used to have the stat tables posted for the various groupings, idk if they still do or if they are updated. But one thing I do remember is that the tables are different and the tables for EoJ and Mementoes pulls were much different than those for shells or honor. Your chances of getting a usuable stat on EOJ or Memento pulls are miserable, not anywhere close to 1 in 4, If you got 5 usable stats in 28 pulls I'm thinking you did pretty well.

  • The stats from EOJ regular rewards seem to have the same stats odds as pulling with shells and honor points. Plate has a chance for normal mode instance Stamina/Defense stat, but sometimes it takes over 5000 to actually get that stat. I wouldn't recommend pulling from best rewards unless you have lots of EOJs to spare. Last zone to have good stats from the best rewards was Splitwater.

  • @Game Maters. Could you please do some tests? Create a char, refill its currencies as necessary, make hundreds of pulls and share the statistics. Many players will be grateful for this.

    You can get an insight how to properly conduct a statistic here [https://en.m.wikipedia.org/wik…stical_hypothesis_testing] and see why pulling 28 times is not enough to have answer (Also please see [https://en.m.wikipedia.org/wiki/Binomial_distribution] ).


    Also saying "good stats" isn't very precise.

    We could computer how likely it is to get 5 good stats out of 28 if we knew which item you are talking about and what you consider good stats.

  • @Game Maters. Could you please do some tests? Create a char, refill its currencies as necessary, make hundreds of pulls and share the statistics. Many players will be grateful for this.

    You can get an insight how to properly conduct a statistic here [https://en.m.wikipedia.org/wik…stical_hypothesis_testing] and see why pulling 28 times is not enough to have answer (Also please see [https://en.m.wikipedia.org/wiki/Binomial_distribution] ).

    Yes, I think this is why he asked a GM to do it 100s of times.

  • Thank you for answering, at least it removes the uncertainty whether messages are read at all.


    Yes, I'm well aware that 28 pulls are not enough to make reliable conclusions for this situation. There were no claims made, only expressing the

    doubt.


    Sadly i don't have unlimited resources to run proper tests. That's why I asked GMs to do it. Of course, the best approach is counting each stat and estimate the "fairness" hypothesis.


    PS. Calculating the probability to get 5 of 28 may be a good exercise for a high school student. Of course it is completely unrelated to the issue in question.

  • Well, you asked us for tests, so please tell us the item you pulled and which stats you consider good stats.

    Does it honestly matter what Utu considers a good stat and what's not a good stat? I'm sure that if you guys who are able to have unlimited currencies because you are with the company running a game would be able to pull 100-10000 of one item and see the odds of one stat combo being made is far easier (and quicker) than us players attempting to do that same thing with our LIMITED currencies we have available. That was the whole reason for his post in the first place. He's not asking for theoretical assumptions or calculations. He's asking those with GF to do real testing to see if certain stat combos are more prone to being pulled over other ones. If mattk/int comes up 5 out of 1000 times while int/dex comes up 95 out of 1000 times it could be assumed that mattk/int has a lower REAL LIFE chance than int/dex combos do. Now do you see the problem here and why you're not offering a solution or help in the matter?

  • Well, you asked us for tests, so please tell us the item you pulled and which stats you consider good stats.

    Does it honestly matter what Utu considers a good stat and what's not a good stat? [...] Now do you see the problem here and why you're not offering a solution or help in the matter?

    It does matter.

    Given a loot table with 6 stats. And someone says after 12 pulls I didn't get a good stat. What does the player consider a good stat? 1 of the 6 stats, or 3?

    Also without providing us the item name we cannot test anything at all.

  • While Utu asked for 100 pulls..,



    Please pull 1000 for each type of gear and put the result of each pull in a table.


    And like Utu said: 'Many players will be grateful for this'.



    Of course we will share our analysis of the results also here on the forum to help the community of this great game! 8):thumbup:

  • going from testing 1 item 100 times to all items 1000 times is kind of a big increase xD, but this would indeed make for way more accurate statistics.


    Anecdotally I myself have never felt that the odds are off. Ive gone on plenty of dry streaks, but ive heard of many players that got lucky many times in a row too (perhaps because its special and more likely to be worth mentioning?). Like for example i get really lucky on shell stats, but unlucky on proof rolls. I myself am then inclined to say my luck sucks (lost alot of money in proofs), but overal its just average.


    From all the gambly things in rom stat odds are my lowest concerns, but would still be very interesting to see the results.


    If utu can just clarify what item and stat(s), it would be really cool if some testing (1000?) with result table could be posted.


    PS. i would also be interested in the HD-rate on pulls, idk if someone knows the numbers here?

  • One thing i forgot to mention last time. It was a pleasant surprise to see that game team has people versed in statistics.

    My initial message was intentionally made in layman terms to draw attention, good it's no longer needed.


    Now we can head to test scenario discussion. The number 1000 looks good. Asking to make more pulls of a single item seem unreasonable. On the other hand, it's more than enough for high certainty conclusions. But i will understand if you don't have the time for all and will go for a small number.


    The minimal setup to leave players satisfied (with tests conducted, not necessarily the outcome) should involve the following.

    A "physical" item for 13 proofs. Any would do.

    A "magical" item for 13 proofs.

    One plate, one chain, one leather, one mage item for 300 shells

    Pvp "dawn" earrings except rebirth

    A healer item from dreamland.


    This seems to cover most cases and is feasible to implement. EOJ pulls testing is nice-to-have but it definitely has lower priority.


    Looking forward to see any results.

    Thank you in advance

  • Having someone do 1000 tests for every type of gear is asking for much, but the question is what are the real probabilities of these pulls? Can you confirm that the tables for possible stats and their probabilities shown as equal on the RoM Welten database is correct?

    Looking at the database tables for EOJ in Chassizz, it looks like the normal shell/honor stats (what alander said). I've never had a problem with shell stats. Sometimes you're unlucky and the stat you want doesn't show up for a while, but people have complained of rarely ever getting a good stat. So, is 1/4 the actual chance for a specific stat with Shells/Honor/EOJ?


    Korin , if you're testing, please try plate gear and tell us how many of each stat you get.

    Restraint Handguards of Dawn (EOJ)

    Subversion Handguards of Dawn (Shells)

    Steel Boots of Dawn (Honor)


    It's unlikely that it'll be exactly 25% for each stat, but with enough trials it should be accurate enough or we should be able to draw our own conclusions if it shows up to be dramatically different than expected. I don't think they're rigged, but would be happy for this thing to be put to rest.



    I checked all of the "New Pantheon" Mem gear sold in Chassizz (cloths, leather, chain, plate, weapons, and accessories) and they all have the same table on the database for possible stats. I think we can agree that none of these stats are usable:



    Defense of the Pantheon
    Stam/MP
    Fire of the Pantheon
    Str/Matt
    Strength of the Pantheon
    Dex/Matt
    Ice of the Pantheon
    Int/MP
    Moon of the Pantheon
    HP/MP
    Star of the Pantheon
    Def/Matt
    Weeds of the Pantheon
    Str/Matt
    Barbarity of the Pantheon
    Str/Wis
    Stone of the Pantheon
    Patt/MP
    Strike of the Pantheon
    Dex/MP
    Breakthrough of the Pantheon
    Patt/Mdef
    Mountain of the Pantheon
    Int/Str
    Woods of the Pantheon
    Int/Dex
    Forest of the Pantheon
    Int/Patt
    Tide of the Pantheon
    Wis/Patt
    Lake of the Pantheon
    Wis/Dex
    Water of the Pantheon
    MP/Matt
    Remnants of the Pantheon
    Wis/MP
    Obscurity of the Pantheon
    Mdef/MP




    Is this correct? While we're on the subject, who were these stats meant for? I understand that we can't have all good stats (that would be too easy), but 0/19 of the stats usable? Unless you're buying the mem gear to actually wear, there's no reason for someone to ever buy it.

    Signature coming soon. Maybe.

  • Please pull 1000 for each type of gear and put the result of each pull in a table.

    seal-of-disapproval-pet-not-impressed-1360795792b.jpg


    [...] The minimal setup to leave players satisfied (with tests conducted, not necessarily the outcome) should involve the following [...]

    We do test that, and I'm not going to test everything again just because someone was having bad luck and feels something is odd.

    IF you name me an item and a hypothesis that can be tested (either probability* or the distribution**) I am willing to conduct a test,

    else not.


    * For example "I don't think stat X has probability Y"

    ** For example "I don't think that all 4 stats are equally distributed".

  • It's getting better and better with each your post. @Korrin is probably a professional card player who enjoys opening his trump cards one after another ))


    Based on drop issues during the "watermelon" even I was under the assumption that game team doesn't recognize the necessity of testing actual items beforehand. It feels really nice to be wrong about this.


    Since the tests have been already conducted could you please share some results? How many pulls were made? Do stats in pom, shell, honor, eoj, dreamland items seem to equally distributed?


    PS. I don't think any particular item has bias while others don't so no need to go that way.

  • @Korrin is probably a professional card player who enjoys opening his trump cards one after another ))

    Well, kind of, but not really xD


    Based on drop issues during the "watermelon" even I was under the assumption that game team doesn't recognize the necessity of testing actual items beforehand.

    The watermelon seed loot thingy works different than stat pulling. Maybe SaitoHajime is kind enough to explain it to you, maybe not ^^


    Since the tests have been already conducted could you please share some results? How many pulls were made?

    No, we do not share internal info. If you would like to help testing Runes of Magic you can consider applying to the tester team.

    Do stats in pom, shell, honor, eoj, dreamland items seem to equally distributed?

    Take a look into loot tables. Sometimes stats are equally distributes, sometimes they aren't.

  • First, Watermelon event was mentioned because it generate funny circling dialog when Gameforge was saying about correct database values and players were complaining that actual drops were not what they should be. This kind of interaction surely can inspire certain ideas about game team. But this is not issue in question and let's focus on the stats.


    First, where can one see the "official" loot tables? I mean those published or confirmed by Gameforge . I hope you are not soliciting players into inspecting client files and violating ToS section 3.3 or taking third-party information for granted.


    Second, we are again not on equal footing here.

    Have you tested distributions of all drops? Or at least those proposed earlier?

    Players cannot make thousands of pulls for free. I got suspicious after spending almost EOJ points cap for chain and leather items in Chassizz. Have you tested those? What was the expected distribution? Did the real distribution pass statistical criteria? These questions don't seem to dispose your "really internal" info and answers will be much appreciated.

    Or maybe you tested just one particular distribution. In that case request for deeper inspection is reasonable. But shell items would be a higher priority as well as PoM gear.


    As you can see answering that way is getting the discussion nowhere. Although it is a good way to bury the conversation when one party has to make may guesses without prior knowledge and another can happily give short comments without any actual information. Just compare the size of your last response and the number of options i had to consider making mine.

    I hope we can do better than that.


    Looking forward to anything you consider helpful to keep players informed.

  • First, where can one see the "official" loot tables? I mean those published or confirmed by Gameforge . I hope you are not soliciting players into inspecting client files and violating ToS section 3.3 or taking third-party information for granted.

    You can just look in it up the Romwelten database http://www.rom-welten.de/database/view.php?id=214987, no need to break the ToS.


    First, Watermelon event was mentioned because it generate funny circling dialog when Gameforge was saying about correct database values and players were complaining that actual drops were not what they should be.

    This is an example of a bug with the drop chance, so please point them out if you find any issues, we will check it then.


    I got suspicious after spending almost EOJ points cap for chain and leather items in Chassizz.

    7xMEEs2.jpg

  • Last night I was trying to get a stamina/defense stat from the EOJ best rewards. I pulled from the plate gear 27 times without seeing the stat I wanted. Maybe the stat isn't possible to obtain from Chassiz, but I'll try it another day to confirm it. It's also not the first time I tried +20 times in one day not getting the stat from the best rewards.


    A couple of years ago it was the first time I saw a Strength/Physical Attack drop from the last boss of normal mode Sun Temple after +25 runs. Before that I always thought it was some myth that got started to encourage people to run normal mode if they couldn't do hard mode.

  • The best addon to use ingame is Dungeonloots. You can search any item and double left-click it to see the drop rate of the stats in the loot tables. Of course, the observed drop rate will be vastly different unless you do like 10,000+ runs.


    For example I was running bethomia to farm warlock ISS. I did 18 runs without dropping a single one. Given that the chance of dropping a mage upper is 16.6%, so the chance of not dropping a single one is just under 4%. It's easy to get frustrated and feel unlucky when that happens, but in reality each run is independent from the one prior, so you aren't any more likely to drop the mage upper in each subsequent run as opposed to an earlier one.