[Technical] How rare is Rare?

Since the game is down, I thought this would be a good time to analyze the data I've collected on all the chests I've opened (I started collecting data when I was at level 9 in the campaign).

So, how rare are Rare/Epic/Legendary items, anyway?

Short answer: depends on what kind of chest you're opening.

Long answer:

These are the basic kinds of chests in the game:

Purchased:

Simple/Ornate: purchased for 18 gold, contains 2 items, no guaranteed rarity, no treasure​

Magnificent: purchased for 30 pizza, contains 1 Rare or better and four no guarantee​

Epic: purchased for 150 pizza, contains 2 Epic or better and 3 no guarantee (I think, I didn't buy any)​

​

MP:

Brown: 2 items, no guaranteed rarity​

Gold: 4 items, rarities are different for each​

​

SP:

Mid-Dungeon: 2 items, no guaranteed rarity​

End-of-Dungeon: 4 items, 1 guaranteed Uncommon or better​

Treasure Hunt: 1 specific Epic first time, same as End-of-Dungeon on repeats​

Quest: 3 items, 1 guaranteed Rare​

Level 1: these are just weird​

(I don't know if the Epic in Treasure Hunt chests replaces the guaranteed Uncommon or not; all 11 of mine had an additional Uncommon or better in the chest, but that should happen 90% of the time anyway.)

NOTE: The percentages below are per item, so the chance of getting an Epic in an End-of-Dungeon chest is: 3 * 0.64% + 2.40% = 4.32%

Purchased Chests: 762 data points
(188 Simple chests, 189 Ornate chests, 2 Magnificent chests)
Common: 77.17%
Uncommon: 18.11%
Rare: 4.20%
Epic: 0.52%
Legendary: 0%

No Guaranteed Rarity: 1409 data points
(SP and MP chests)
Common: 69.77%
Uncommon: 25.55%
Rare: 4.05%
Epic: 0.64%
Legendary: 0% (not exactly zero)

Guaranteed Uncommon: 208 data points
(End-of-Dungeon chests, MP chests)
Uncommon: 79.33%
Rare: 17.31%
Epic: 2.40%
Legendary: 0.96%

EDIT: I calculated this based on the assumption that adventure final chests guaranteed one Uncommon. However, it now seems that most actually give two guaranteed Uncommons, so I need to recalculate.

Guaranteed Rare: 127 data points
(MP chests, 2 Magnificent chests)
Rare: 72.44%
Epic: 22.05%
Legendary: 5.51%

Guaranteed Epic: 11 data points
(club rewards for Treasure Hunts)
Epic: 72.73%
Legendary: 27.27%

 

So, getting treasure hunt along with club will net you 11 epic+ items
Which is roughly same as getting 5 Epic Chests which = 750 pizza

To me it seems like getting treasure hunt + club is tremendous value

 

Oh thank GOODNESS. I've been wondering whether the spread were anything like classic D&D percentages, or were skewed to give out too many rare items. Because Common is above 50%, it's actually "common," which avoids the most blatant mistake.

Interestingly, the rest matches pretty closely to the classic, being:

Common: 75%
Uncommon: 20%
Rare: 4%
Very rare: 1%

We might as well assume it is literally the same chart. Well, except that we don't know what percentage Legendary should be.

 

5% chance of legendary for MP isnt too bad. Pretty easy to get 2 chest a day, so 1 legendary per 10 days of casual play. Plus 4 epics.

 

Hey Some more data
64 epics chests
20 out of 128 were legendary

edit: ps. this was on the test server. I did NOT spend real money on this

 

Saw this post bumped and clicked in just to recommend using the Test Server to figure things out. Glad someone else had the idea and actually acted on it. Though having data from multiple people may not be a bad thing. Could possibly even figure out if there are differences in drop chances among rarity tiers (ie - more likely to get a legendary level 7 item than a legendary level 18 item)

 

Definitely the best way to spend pizza if you want to play the SP campaign.

 

So, what are the chances of getting a certain item type at each guaranteed rarity?
(Attempting a table...)

______Purchased: 762 - None: 1409 - Uncommon: 208 - Rare: 127 - Epic: 11
arcane item: ____ 10.76% _____ 9.58% ________ 10.58% ___ 11.81% _ 18.18%
arcane skill: ______ 6.69% _____ 5.18% _________ 5.77% ______ 0% _____ 0%
boots: __________ 8.14% _____ 6.96% _________ 6.25% ____ 7.87% _____ 0%
divine armor: _____ 5.91% _____ 3.62% _________ 4.81% ____ 2.36% _____ 0%
divine item: ______ 8.53% _____ 9.01% _________ 9.62% ____ 6.30% ___ 9.09%
divine skill: ______ 5.77% _____ 4.05% _________ 6.73% ____ 2.36% _____ 0%
divine weapon: ____ 6.96% _____ 4.26% _________ 2.40% ____ 7.87% ___ 9.09%
dwarf skill: _______ 2.62% _____ 0.78% _________ 1.92% ____ 0.79% _____ 0%
elf skill: _________ 1.05% _____ 1.28% ___________ 0% _____ 1.57% ___ 9.09%
heavy armor: _____ 5.77% _____ 4.54% _________ 5.29% _____ 3.94% ____ 0%
helmet: _________ 3.54% _____ 3.19% _________ 1.92% _____ 5.51% ___ 9.09%
human skill: ______ 1.57% _____ 1.35% _________ 0.96% _____ 2.36% ____ 0%
martial skill: ______ 5.64% _____ 5.18% _________ 4.81% _______ 0% ____ 0%
robes: __________ 6.04% _____ 4.33% _________ 4.81% _____ 0.79% ____ 0%
shield: __________ 6.69% _____ 5.32% _________ 1.92% _____ 8.66% ____ 0%
staff: ___________ 7.74% _____ 6.60% _________ 5.77% ____ 10.24% ____ 0%
treasure: __________ 0% ____ 16.75% _______ 16.35% ____ 20.47% _ 45.45%
weapon: ________ 6.56% ______ 8.02% _______ 10.10% ______ 7.09% ____ 0%

So it's not my imagination that I'm getting too many treasures and not enough race skills.

 

One other thing that makes treasure club valuable is when completing quests. The quests are available once you complete the campaign, but the final chest for each guarantees at least a rare card.

Obviously it also helps with the MP rewards, plus the treasure hunt adventures. If you're playing a lot, it adds up quickly.

 

I did some calculations for MP chest farming, assuming you have purchased the club:

If you do 1 win/day, after a week you will get: 1 Legendary, 3 Epics, 11 Rares, 5 Uncommon and 15 Commons
If you do 3 wins/day, after a week you will get: 2 Legendary, 7 Epics, 24 Rares, 20 Uncommons and 39 Commons
If you do 6 wins/day, after a week you will get: 3 Legendary, 11 Epics, 38 Rares, 43 Uncommons and 74 Commons
If you do 11 wins/day, after a week you will get: 4 Legendary, 16 Epics, 57 Rares, 68 Uncommons and 142 Commons
If you do 20 wins/day, after a week you will get: 8 Legendary, 28 Epics, 65 Rares, 115 Uncommons and 274 Commons.

If we assume an average game duration to be 20 minutes, and 2 games needed for a single win (in a long term it must be true in ELO system), that would mean 40 minutes of your time per win.
Please note that 11 wins/day is more time than a full time job, and 20 wins/day would take more than 13 hours (!)

Are my estimations totally off and the most MP games are done much shorter? Also, can a good player expect to win significantly more than 50% games for a long time? I am still a beginner in MP.

My plan is to aim for 1 win/day, as that has best ratio of items/time consumed. Next best is 3 and 6, with 4,2,7 following. All others are definitely suboptimal.

----

And of course, even better is to work at a job, get $10 per hour, convert to 330 pizza, and then buy chest/items in shops 1650 gold can give you a lot.

 

Purchasing chests for pizza:
If you still want/need some rares, you should purchase the 30 pizza chests. Only if you are after epics/legendaries, you should focus on the 150 pizza ones.

$99 / 3800 pizza will get you:
a) Magnificient chests: 15 Legendaries, 61 Epics, 203 Rares, 127 Uncommons and 355 Commons
b) Epic chests: 21 Legendaries, 56 Epics, 3 Rares, 19 Uncommons and 53 Commons
c) Ornate chests: 32 Epics, 127 Rares, 792 Uncommons and 2217 Commons

Buying Epic chests instead of Magnificient chests will only convert 5 epics into legendaries, from your total of 76 E+L items at the expense of 200 rares that you will lose, and 100 U+300 C items.

The cost for one E/L item is therefore 99/76, or about $1.30.
(If you only want Legendaries, then those will cost you $4.82 each (via Epic chests) or $6.51 each (via Magnificient chest).)

 

Avg game time should be significantly lower than 20 mins. 15 mins is closer but thats still on the long side of estimates. I find median to be aprox 12-13 mins.

It also partially depends on your own party make up. Certain parties tend to win faster when they do and die faster when they do. I would assume 3 mages takes longer to win/lose than 3 warriors.

If you don't smurf, doubtful. You need to be right at the very top, i would guess at least top 50, for the game to have difficulty matching you with similarly skilled players consistently. Still, even the best players in beta only had a win rate of 60% or so in general. 1-3 might have gotten close to 70%

 

That fits pretty well with my observations. If i had to guess, i'd say that it's a complex set of "rolls" to determine if an item will get upgraded from its base state. These numbers are just guesstimates, but it's like a 30% chance of getting an item upgraded to the next level. After that it's a 25% chance to see if that item will get upgraded again, a 20% chance for the next level, etc. So if you have no guaranteed rarity, it would break out something like this:
Common: 70%
Uncommon: 20.8%
Rare: 7.5%
Epic: 1.5%
Legendary: 0.2%

If you have a guaranteed rare, it looks like so:
Rare: 70%
Epic: 22.5%
Legendary: 7.5%

Etc.

Obviously, under this system, the higher your guaranteed base, the more likely you are to get a top tier item. That doesn't quite match Kalin's data, but it's pretty close for a guess.

One thing i learned on the test server: it sorts your items by rarity so the items only get better as the items are revealed one by one. If your first item is upgraded all of the ones of the same guaranteed rarity will also be upgraded. Also: pizza chests do not contain treasure - only actual items. Not sure about the others, but i suspect that holds true for all purchased chests with a guaranteed rarity.

 

I think the sorting happens after the chest is filled. In beta I had a treasure hunt where the guaranteed item was in the third slot, with an extra epic in the fourth.

 

By the looks of things, a good rule-of-thumb is that each rarity level is 5x less common than the one before. (eg. 'uncommon' is 5x rarer than 'common', and rare is 5x rarer than 'uncommon' and 25x rarer than 'common'. etc.)

Also, based on the low number of epic racial skills found, my guess is that the rarity of the item is determine first, and then the actual item is randomly selected from all possible items with that rarity. ie. All items of the same rarity are equally likely to come up; but since there are fewer total epic racial skills compared to other item types, epic racial skills in general are less likely to come up. (Just a guess.)

 

what exactly is the difference between a quest and the end of a dungeon? I'm confused?

are some dungeons quests? if so which ones? are some dungeons quests, but only once? (the dungeons that once you beat them the first time they unlock more of the map?) if I repeat these do they keep giving me the rare? or only an uncommon?

 

@namad, a 'quest' is a special challenge that involve re-playing a module with some kind of restriction to make it harder. The option to partake in quests is only unlocked after completing the entire single-player campaign. Quests typically involve things like 'use only elves', and 'no party deaths'; and completing each quest guarantees a reward of at least one 'rare' item. There are several quests for each module (ie. adventure), and each quest can only be completed once.

 

thanks this clears things up for me, quests are something i get later glad I wasn't misunderstanding anything.

 

I'm playing part 1 of "Gladiatorial Arena, Qualifiers" over and over again to get items. According to Kalin, this is a "Mid-Dungeon: 2 items, no guaranteed rarity" stage, which has a Legendary drop percentage of 0%. However, I just scored a legendary item! I almost can't believe it. Based on this data point alone, it seems Mutak's 0.2% drop percentage sounds more accurate.

 

Kalin's drop rates are based on actual data - mine are based on guesswork extrapolated from data. As such, Kalin's rates are 100% accurate but not necessarily definitive or predictive, while mine are predictive, but have no guaranteed accuracy and are no where near definitive.

Kalin's rates probably have a margin of error of plus or minus 5%. Mine have a margin of error of plus or minus 100%. Mine might be perfectly accurate but there's no way to confirm that except by doing what Kalin is doing.