Monday, April 23, 2018

Underworld Overhaul, Pt. 1: Monster Level Tables

We noted previously that numbers for monsters and treasure, in the dungeon in OD&D, were at their root determined by DM fiat. Now we'll spend a few posts trying to massage what is there into a more formal and evidence-based system.

The first part here is probably the easiest: Re-aligning the monster level tables via some underlying rule. Fortunately, we have the fairly well-justified Equivalent Hit Dice (EHD) values available from our previous Arena/Monster Metrics program, and it is relatively straightforward to turn those into a rule for distributing monsters among the 6 "monster levels" in OD&D. Based on inspection of the original book tables, comparing to our EHD values, and also balancing with various experiments in the Arena program, we propose the following as a pretty good definition (see also on GitHub):

Further below, you'll see a complete listing of underworld OD&D monsters parsed into the six levels according to this metric. Most monsters stay at the same level as they appear in the Vol-3 book (p. 1011), although there are some notable changes. Also, monsters are distributed fairly evenly across the six levels, with around 16 monsters each on average; fewer at Levels 1-2 (about 10 apiece), and more in the catch-all Level 6.

To compare to the book system, you can see an old analysis here, in which shaded rows indicate the original monster tables. The book tables have cutoffs at about EHD 1/2/3/4/6/10. But with better monster metrics, you can see in the link that 3rd level has a big problem with how few monsters it keeps under that rule. If we used that book rule to assess the current expanded monster list, then the problem would remain; only about 9 or 10 monsters at levels 2-3; and an enormous dump of 35 monsters all at level 6. So this strongly indicated that we needed to massage the book pattern a bit for a somewhat better distribution.

In the new list below, the third column in the list below also shows the kill ratio attributable to each monster within a level, based on about 2.4 million combat simulations of 4-man fighter parties assaying into a standard-type dungeon, and encountering bands of monsters at equivalent EHD values. (This was generated with the current Arena program using these command switches: -n=10000 -v -z=4 -rk.) Monsters in each level are sorted by increasing kill ratios, so that the whole list runs uniformly from lower to higher danger.

The kill-ratio organization also allows us to look for outliers at the start or end of a table, which may argue for incrementing or decrementing the EHD value of any monsters "on the bubble", so to speak, and this was done manually in a few cases. (Specifically, EHDs for the following were hand-tuned based on this analysis: Large Spider, Giant Beetle-Bombardier, Hell Hound, Harpy, Hero, Swashbuckler, Gray Ooze, and Zombie.) As one example, the Zombie (at the newly-recognized 2 HD) is inevitably going to stick out as either the strongest Level 1 monster, or the weakest Level 2 monster; we've set the EHD at 2 so it appears at Level 2 and smooths out the tables sizes a bit.

With that in mind, here are the proposed new Monster Level Tables:

#Level 1 MonstersKill %
1Giant Rat3%
2Giant Centipede4%
6Large Spider9%
9Giant Beetle, Fire13%

#Level 2 MonstersKill %
4Giant Frog7%
5Lizard Man8%
6Gray Ooze10%
7Giant Ant, Worker11%
8Giant Hog13%
9Giant Lizard15%

#Level 3 MonstersKill %
2Giant Toad3%
4Giant Tick4%
5Giant Weasel4%
6Giant Beetle, Bomb.4%
12White Ape9%

#Level 4 MonstersKill %
3Giant Snake4%
5Giant Beetle, Boring5%
6Giant Wasp5%
7Owl Bear5%
11Gelatinous Cube6%
13Giant Ant, Warrior7%
16Hell Hound15%

#Level 5 MonstersKill %
2Hydra, 6 Heads2%
4Displacer Beast3%
5Giant Beetle, Stag3%
6Giant, Stone3%
8Giant Slug6%
10Giant, Hill6%
11Giant, Frost7%
14Minotaur Lizard9%
15Invisible Stalker9%
16Giant Scorpion9%

#Level 6 MonstersKill %
2Giant Beetle, Rhino.0%
3Giant, Cloud0%
4Giant, Fire1%
5Hydra, 10 Heads1%
6Fire Lizard1%
8Umber Hulk1%
11Golem, Flesh1%
12Carrion Crawler1%
14Phase Spider2%
15Giant, Storm2%
16Dragon, White4%
17Dragon, Black4%
19Dragon, Green6%
20Dragon, Blue6%
21Mind Flayer6%
22Dragon, Red6%
23Ogre Mage6%
28Purple Worm9%


  1. I am looking at the 0% kill ratio for the Level 6 monsters.
    Do you happen to have kill ratio numbers if the Ettin, Giant Rhino Beetle and Cloud Giant were Level 5 monsters?
    Are they just too deadly at level 5 and pushovers at level 6?

    1. That would be interesting, especially as wandering monsters are not always encountered only on their assigned level.

      I'm just looking at the troll, it's a bit disappointing that such a well known and feared monster is actually the weakest on level 5! Perhaps when it makes an appearance on level 4 it really is as terrifying as we'd hope.

      On the other hand I'm entirely happy with a zombie being the weakest on level 2, that seems fitting.

    2. I suppose the deadliness may go up as you take into account number appearing? If I recall, and reading back past Monster Metric entries, the above kill% is for 1 monster vs. party.

    3. The current version does party-based encounters, and this simulation was done with 4-man parties (which has been found to be a nice size for advancement purposes).

      The number of monsters encountered is scaled to the party, dungeon level, and EHD. For example: say there's a 5th-level fighter in the list. For an expedition he gets together with 3 other random 5th-level fighters, and presumably they adventure on the 5th level of the dungeon. They roll on a Monster Level Matrix (like OD&D Vol-3, p. 10). Let's say they generate an encounter with Trolls (EHD 9). Then we compute a balanced number to be NA = 4 (party size) × 5 (dungeon level) / 9 (EHD) = 2. (That is, party EHD is 4×5 = 20, and monster EHD is 2×9 = 18, which is as close as we can make them.)

      The Ettins, et. al., at 6th level get kills, it's just that the percent rounds off to 0%. For example: In a run of 2.4 million total combats, 6th level monsters score a total of 4,945 kills. Ettins scored 14 kills (for which 14/4945 = 0.28%); compare to Purple Worms with 456 kills (456/4945 = 9%).

      If we moved these types to the next lower level, they would tend to stick out as outliers in the other direction, which is undesirable in its own way. For example, with the current categorization, you'd have to lower Ettins from EHD 13 to 10 (to get into Level 5). In that case, they account for 17% of all the kills at Level 5, which is almost twice the kills of any other monster at that level (Spectres being pushed down to 9%, Giant Scorpions 8%, etc.), which doesn't look right to my eye.