OSR: Treasure Curves & Generation Procedures

Work continues on the Treasure Overhaul, and that means math.

Part 1: Gem and Jewellery Generation

The book will include treasure tables, so I've spent quite a bit of time digging into their function, their challenges, and their deceptive math.

Treasure Type A is bandit treasure. Sure, in AD&D it's also associated with the Lich, Locathah, Troglodytes, and the Giant Squid, but it is, in most system that include treasure tables, A is "the chest at the heart of the encampment" treasure. It's a useful point of comparison. We'll ignore coins and magic items for the time being.

OD&D: 50% chance of 6x1d6 gems /  50% chance of 6x1d6 pieces of jewellery
AD&D: 60% chance of 4x1d10 gems / 60% chance of 4x1d10 pieces of jewellery
OSE:  50% chance of 6d6 gems /  50% chance of 6d6 pieces of jewellery*
TO: 50% chance of 6d6 gems /  50% chance of 6d6 pieces of jewellery*

*The average of 6d6 and 6x1d6 is the same: 21, but the distribution changes significantly. OSE's treasure results fall in the middle of a curve instead of spreading out. 

 Yanping Wang, Sand

Gem Procedure

I started with:

For each gem or batch of 5 or 10 gems: 
1d20
1-4: 10gp
5-9: 50gp
10-14: 100gp
15-18: 500gp
19: 1,000gp
20: 5,000gp

For each gem or batch with a roll of 15+, roll an additional d20. 
1-16: No effect
17-19: x2
20: x10

This procedure looks good. It passes the back-of-a-napkin math test. But it doesn't pass the complicated math test. 

When simulated, the results look like this:

I've charted them alongside OD&D, AD&D, and OSE's gem results for Type A treasure.OSE has a small gem value table and no way to increase a rolled value. It's tidy, but it's flat. 

You can see that the Treasure Overhaul's results spike way higher than the reference systems. Around 15% of results are above 20,000. The general shape looks OK, and the lower end of the range lines up nicely, but it clearly needs some tweaks. 

After some guesswork and tests, the revised procedure looks like this:

For each gem or batch of 5 or 10 gems:
1d20
1-4: 10gp
5-9: 50gp
10-14: 100gp
15-18: 500gp
19: 1,000gp
20: 2,000gp

For each result of 15+, roll an additional d20. 
1-16: No effect
17-19: x2
20: x3

Tweaking the top of the modifier table to x4 results in an overshot. I could fiddle around with the 19 and 20 results on the main table to get the perfect top-of-the-range match, but that pushes the centre of the curve away from OD&D and A&D.

This method does not have a chance to produce extremely high-value gems, unlike OD&D/AD&D's recursive lottery, but it's a surprisingly close match, and it only requires 2 rolls (or one throw). Gem values are capped at 6,000gp, which is plenty for spell purposes (especially in a batch of 10), and the GM is free to combine gems.

I wanted to keep 6 possible outcomes on the d20 table so that each outcome is associated with a different tier of stone, or with one emblematic D&D / spell component stone.

	TO Gems (Original)	TO Gems (Revised)	OD&D Gems A	AD&D Gems A	OSE Gems A
Average:	4,542gp	3,405gp	3,243gp	3,765gp	2,208gp
>3x Avg:	7%	12%	10%	0%	8%
>2x Avg:	11%	17%	14%	17%	24%
# of Gems Avg	10.5	10.5	10.5	15	10.5
Gem Avg.	433gp	324gp	309gp	251gp	210gp

This gem procedure is fairly elegant; you just roll 2 d20s at once and only reference the second one if the first is 15+. I even made the gem type/description tables d12s and d6s, so you can roll a d12 and a d6 in the same handful and get all the results you could possibly need at once, without interference. I could have used a d100 instead of one of the d20s, but testing confirmed that d100 rolls are annoying for batched results.

Censer c.~1500

Jewellery Procedure

As mentioned in my previous post, jewellery is the meat-and-potatoes of OD&D and AD&D treasure table results. 

I started with:

For each gem or batch of 5 or 10 pieces of jewellery: 
1d20
1-5: 200x1d10gp
6-14: 500x1d10gp
15-20: 1000x1d10gp

For each result of 15+, roll an additional d20. 
1-16: No effect
17-19: x2
20: x10 x3

I'd like to use the same modifier table for both Gems and Jewellery, the so table was adjusted accordingly. It turned out to be exactly what the Jewellery table required. I ran a few tests with tweaked values, but it turns out that the results are nearly perfect as-is, post gem revision.

I'll continue to test this system, just in case the highest value needs to be tweaked from 1,000gp to 900gp, but so far the results are very encouraging.

 

	TO Jewellery A	OD&D A	AD&D A	OSE A
Average:	35,524gp	32,858gp	35,902gp	9,342gp
>3x Avg:	10%	9%	10%	4%
>2x Avg:	25%	23%	19%	26%
# of Pieces Avg	10.5	10.5	13.2	10.5
gp/Piece Avg	3,383gp	3,129gp	2,720gp	8,90gp

 

Part 2: OD&D Treasure Comparisons

Here's a comparison of all the major OD&D treasure types. 200 runs each.

I wish OD&D named its treasure types instead of vaguely implying their function, but, thanks to this chart, some of the principles become more clear. Courtney Campbell broke down the AD&D charts here. Here's my take on OD&D. They're not as baffling they first seem.

A is Men (non-nautical, non-desert) and Centaurs. It's a Bandit Lair / Treasury. 

B is Ghouls, Wights, Hydras, and Nixies. Things that kill adventurers by the dozen. It's Loot / Burrow / Gullet items. The magic items are "Weapon, Armour, or Misc. Weapon." Presumably they didn't work.

C is Gargoyles, Lycanthropes, Minotaurs, Pixies, Gnomes, and Ogres. Ogres add +1,000gp. It's Pocket Change. If it doesn't include jewellery (a 75% chance) it's basically worthless.

D is Orcs, Hobgoblins, Gnolls, Trolls, Mummies, Cockatrices, Manticores, Purple Wurms, and Dryads. I'm calling this Traders and Raiders. Portable, practical wealth.

E is Wraiths, Spectres, Gorgons, Wyverns, Elves, Griffons, and Giants. Giants add +5,000gp. It's Better Pocket Change. I'm not totally sold on this as a unique treasure type. More magic items than C, but otherwise, it's just C + gold pieces + better chances.

F is Vampires, Basilisks, Meduae, and Chimeras. It is heavily jewellery-weighted, with 2-24 pieces. and the magic items stipulate "no weapons". This is Glittering Trinkets. Even an immortal vampire eventually gets tired of being stabbed by decorative mantlepiece swords.  

G is Dwarves. Gold and magic items. It's a Dwarven Trove. There's nothing else it could be. Gold, more gems than jewellery, and a good chance of magic items. You can clearly see the steps of the 1d4x10x1,000gp in the blue coin section.

H is Dragons. It's a Hoard. Lots of everything, but (as previously discussed) more gems and jewellery than coins. 

I is Rocs. It's Shiny Things. Jewellery and gems only; stuff a magpie the size of an elephant would loot.

Some of the categories still don't make any sense to me. Why do Griffons get scrolls? Why aren't weapons scattered around a Basilisk, and why do Chimeras need all that jewellery? But despite a few oddities, these charts really helped me get a grip on OD&D's internal logic. AD&D is... another matter.  

Final Notes

I hope this offers some insight into the rigorous (and peculiar) methods that are applied behind the scenes of the Treasure Overhaul. There are plenty of things that I'm happy to eyeball or write from scratch, but if I'm going to include treasure tables, I want them to be as sturdy as possible.   

EDIT: One potential reason why a lot of calculated average values for OD&D Treasure Type A (or conversions based on it) online and my values are so different might lie in the deceitful 6d6 vs 6x1d6 roll. 6d6 and 6x1d6 have the same average of 21... but think about the distribution. The chance of rolling a 36 on 6d6 is 0.2%. The chance of rolling 36 on 6x1d6 is 16.6%. And when we're talking about gems, where each gem can suddenly increase in value... 

OSR: Testing Treasure Tables & Treasure Values

Treasure tables are, for better or worse, part of old-school RPGs. Since the intention is to make the Treasure Overhaul a very useful and broadly compatible book, it's important to understand the subtleties of historical treasure generation. The average treasure value of a given type is relatively easy to calculate. It's also deceptive. Averages are a good start, but if you only look at averages, you can easily miss the big picture. 

To make sure I understood the bones of OD&D and AD&D's famous treasure tables, I cobbled together a spreadsheet that runs through 100 at a time. No, you can't see it. It's awful. I'm reasonably confident it's correct, but it's not elegant.

OD&D Treasure Type A1 

Bars are cumulative. 200 trials.

I ran many, many trials to confirm the general trends, but charts with 200 results are easier to read. 

1. Null results

Around 8% of rolls for Type A1 result in 0gp. Around 20% result in less than 1,000gp. 

There's an argument to be made that unexpectedly high-value results (>150,000gp for A1) only feel "real" if there's a chance of unexpectedly low results. I'm still thinking this concept through.

Take random encounter tables. Nobody insists that if a Random Encounter Table includes 3d6 level-draining PC-slaughtering Wights, it's only fair that it include 3d6 helpless Marshmallow Cuddlebugs. The low result isn't required to make the high result feel "worth it". There's usually a spread of results and deadliness, sure, but in OSR games, even a handful of Goblins can be dangerous. People would complain if a random encounter table was 10% empty. 

On the other hand, a random encounter table is only used some of the time. A treasure table is used every time (in theory). So a 0gp treasure is just moving the probabilities of a no-encounter roll to a different step.

A 0gp result can also help with worldbuilding. When I rolled up my OD&D Hexcrawl (available on Patreon) I had a surprising number of 0gp results to explain. That's part of why I created the spreadsheet. I wanted to see if I was unlucky or incompetent. Turns out it was probably both. Hooray!

Antonio J. Manzanedo

2. Gems and Jewellery

The structure of a treasure table, and cultural depictions of dragon hoards, suggests coins are important while gems and jewellery are accessories. The results (and plenty of documented wisdom) shows that this isn't true. Coins are a consolation prize. Gems are a lottery ticket that could boost a small treasure into an enormous one. But the meat-and-potatoes of the OD&D treasure table is jewellery.

Taken from Greyharp's Single Volume Edition.

Treasure Type A1 has a 50% chance of including 3d6 pieces of jewellery. If the treasure includes jewellery, it's all but guaranteed it will dominate the result. You can see the cutoff on the chart.

To match the usual at-table procedure, the sheet rolls jewellery and gem values in batches of 10. E.g. if I rolled a 36 on 6x1d6, I'd roll for 10, 10, 10, and 6 gem values. All you need is one batch of 10 hit 10x1,000gp and you've got 100,000gp. 

To put it another way, for Treasure Type A1, if you roll Jewellery, you get a minimum of 6 pieces. At the minimum value of 300gp per piece, that's 1,800gp. That's worth more than the maximum roll for Copper and Silver coins. The jewellery table is the driving value behind treasure in OD&D. You could leave coins out entirely and barely notice.

Values of each type of treasure, separated. 500 trials.

3. The Gem Lottery 

In OD&D, gems have a base value based on a 1d100 roll, and then a 1-in-6 chance to upgrade to the next highest value (or a list of higher values that can't be initially rolled), up to the incredibly unlikely 500,000gp. Gems are the lottery tickets of the OD&D treasure table. 

But "standard" upgrade, of 10x 1,000gp gems to 10x 5,000gp gems or 10x 10,000gp gems still only puts the Gem results in the midrange of Jewellery results. You can see that on the chart. There are a few results where Gems form the bulk of the treasure, but the solid and consistent mass of Treasure Type A's value comes from Jewellery.

4. Average Value Check

Just to make sure I'm not totally crazy, let's run through some basic averages for OD&D jewellery.

OD&D Jewellery (Type A1) Average Calculation
Roll	Value	Value	Rate	Product
20	(3d6x1000)	1050	0.2	210
80	(1d6x1000)	3500	0.6	2100
100	(1d10x1000)	5500	0.2	1100
		Average Per Piece
				3410
		Average Pieces Per Type A1
			(6x1d6)	21
		Average Per Type A1
				71,610
My Calculations (500 runs)			Rate	0.5
	36,358		Average	35,805

That's pretty darn close, and it matches the values found by others. But this does suggest that Delta's famous XP/GP tables are incorrect, at least for OD&D (if I'm reading them correctly).

OD&D Treasure Type H

Bars are cumulative. 200 trials.

Ah, the classic Hoard. In A1, Gems and Jewellery have same number of pieces (6x1d6). In H, Gems have 1d100 pieces, while Jewellery has 1d4x10 pieces. This provides more-or-less the same jewellery values, with the added boost of a high gem roll giving a higher chance upgrading gem values. In A1, there are at most 4 "batches" that can be upgraded. In H, there are up to 10.   

It's interesting that the top ~20% of Type A1 is above 100,000gp, while ~30% of Type H is above 100,000. Type H feels larger, with its 1d100 gems and 6x1d10x1,000gp, but in practice it's all jewellery-dominated. A has a 50% chance of 6x1d6 pieces; H has a 50% chance of 1d4x10 pieces. The difference between 6x1d6 and 4x1d10 is not terribly significant. 

Dragons, the standard hoarding creatures in OD&D, do have a built-in age-related multiplier. "Very Young and Young Dragons are unlikely to have acquired treasure. Sub-Adult Dragons will have about half the indicated treasure for Dragons. Very Old Dragons can have as much as twice the indicated amount." 

AD&D Treasure Type A 

In AD&D, Treasure Type A is used for for the Lich, Locathah, Men (Bandits), Troglodytes, and the Squid Giant... for some reason. It's a mixed bag.

Bars are cumulative. 200 trials.

As usual, AD&D's procedure is OD&D + a layer of complexity.  

I did not implement the stone increase/decrease base value table. It's more complicated than OD&D's. I looked at it, and I tested out a few worst-case scenarios, but it's a horrible and recursive thing and I just can't be bothered. It provides for a runaway doubling of gem value (as in OD&D) but it's so hard to implement cleanly (with my very limited skillset, at least). As my grandfather said, just before his homemade self-driving car took him over a cliff, there are times when you should code things properly and times when nested IF statements are good enough. 

Jewellery in AD&D also has a runaway increase. If the piece has gems (on a 51+), roll 1d8. On a 1, add 5,000gp and roll 1d6. If that's a 1, double it to 10,000gp and repeat, to a maximum of +640,000gp. There's a craftsmanship value adjustment step which I didn't implement.

The point is to confirm that, in AD&D as well as OD&D, the "average" treasure value doesn't tell the whole story. Jewellery dominates the chart.  

Final Notes

Treasure tables are fun. They are also very difficult to implement with at-table utility in mind. For a pre-generated dungeon, where the GM can sit back, roll, and type things into a calculator, they work just fine. But at the table, with the pressure of game flow bearing down on them, and the added pressure of fast math, they're a slog. 

I've also created a small OD&D hexcrawl and 3-level standard dungeon. They're available on Patreon. I sometimes post previews of the Treasure Overhaul on Bluesky.