I have been playing with BX's procedural dungeon generation, and it's really fun. Creating a dungeon level is such a wonderful way to spend a spring night. But there was a detail that I could not figure out about the monster lairs: Treasure was only to be found in monster lairs, yet the dungeon generation seemed to work with the smaller number of monsters only (and they don't have any assigned "% in lair" attribute like in other editions). Finally, by reading an unvaluable Delta's article this very night, I understood that there was nothing to be understood: there is an ambiguity in the rules that seems to be resolved by assuming that all monsters in the dungeon have an "% in lair" of 100%; or, which is the same, all dungeon encounters are balanced towards their own treasure by using their largest number appearing.
Of course, it might be strange that you open a door in a dungeon and you find there 30 men guarding their #A treasure. So, maybe, it could be a good idea to assume that they have made a lair in that section of the dungeon, and they are tactically and organically disposed along a space of otherwise empty rooms. And now that this is settled, I want to expose my conundrum with treasure types. I already spoke about how I use the averages and make two rolls to make a small variance and decide the predominant shape of it. I found recently an anonymous chart that had done a similar work, also translating the magic item chances to d6:
I still want to find another way. Maybe is just habit, or something in me likes to find the formula that obsoletes the chart. Because, in the end, the three factors (amount of riches, amount of jewelry among the riches and amount of magic items) are kind of proportional and it feels natural to tie them to something in the monster. One OSR game for kids, DAGGER, uses monster's HD + 1d6 against this table: 1-3 no treasure, 4-5 coin purse, 6-7 sack of treasure, 8-9 treasure chest. Dungeon World did the same thing but with the monster's damage die instead:
This example looks so bad because the list is totally off from what you would expect in a D&D's treasure, even at high rolls. But serves to illustrate the concept. Probably for a BX variant the best chance would be tying it to HD; always remembering that many monsters do not carry treasure not even in their lairs. Then, adding explicative notes in monster description for special cases: Dragons always have more treasure, as do Men's lairs, to a crazy extreme. Take a look at Delta's table and be astonished:
I went to my faithful tome of Pits and Perils (Not a clone of any D&D edition, its its own thing, but similar enough to make style comparisons) and here are the notes:
Treasure types are divided into four (I, II, III and IV) in respect to treasure amount. Each one decuplicates the previous one. Those are at the same time divided into type A treasures (natural treasures such as monster body parts or spider webs that take X turns to be harvested, more info on the monster's description) and type B treasures (your classic coins and gems)
With every type increase, also the jewelry and magic item chance increase (also the magic item quality). With the table as it is, there is only the chance of 1 magic item per monster, but it is trivial to hack this to give it more variance: instead of 2, 3 and 4 in 6 chance for types II, III and IV, you roll 2, 3 or 4 six sided dice, with each roll of 1 being one item.
The four types approach is also tempting. I like that the details of treasure are provided in the monster description too, both in amount and in nature. But maybe not even those four types are needed: I think that treasure amount could be written in the monster stats as:
Treasure: Y/N?
If yes, check how much by making a roll based on monster's HD. If there is an arterisk, check description for specifics (add or take money, items, magic...)
There is also a very different approach, which is the one that I am using at the moment. I took it from this very interesting piece by a mysterious author called Lungfungus. I actually went to print and bind that book alongside The Implied OD&D Setting by Wayne Rossi because I love to have that kind of books at home.
The concept I am testing is to standarize the average treasure amount of a treasure room in a level 1 dungeon, taking into account monster, trap, empty and special rooms (I give special rooms 1 in 6 chance of treasure, just like an empty one). Using my own parameters, I calculated that this number is 292 gp per room containing treasure. This number is the same, no matter if the room has a monster, a trap, or neither. Then I randomize the amount with a table, ensuring that in average, the result is still 292 gp;
1 - 25% treasure
2 - 50% treasure
3 - 75 % treasure
4 - 100 % treasure
5 - 150% treasure
6 - 200% treasure
(for example, in a d6 table, the average result will go towards 292 or any other set number, as long as all the percentages sum 600%, as 600 split between 6 results makes for the 100%)
Multiply the given number for the dungeon level you are currently at, to make it grow proportionally to danger. This progression is the same used in AD&D.
This method has the particularity that makes all monsters equal towards treasure, unless you specifically decide against it. So, for example, by giving them individual treasure. This also makes monster-guarded treasure equal in size to trap-guarded or non-guarded one. It is not as crazy as it sounds, if you picture it as that the particular monster that put it in there doesn't feel the need to be guarding it 24h, 365 days a year. The box is already on a deadly dungeon, and he might be lurking around as a wandering monster, or waiting in another room.