I like to roll 1d20 and 1d10, subtract the lowest roll from the highest. Oh, and in this scenario the 0 on the d10 actually means 0. If the roll would make the HP = 0, it is instead the number on all dice involved. And that means the percentage chance is weighted toward the low end but is still swingy as hell (1 has a 9.5% chance of popping, 20 is at .5%). Now I only do this for games that intend a low starting HP, you only have a 27.5% of making it higher than 10HP starting, but semi-fantastic beginning HP are possible. Since HP are an abstraction, no ability scores should ever affect it.
Games I'd use this with that don't have classes (reduce HP advancement to 1/level if it exists at all):
Into the Odd
Knave/Maze Rats
Mork Borg
For other games with classes, I'd add 6HP to starting warriors, 0 HP to starting casters, 3HP to everybody else, and again reduce HP gain with advancement to 1/level, maybe 2/level for warriors
6/25/2023 EDIT: I have since discarded all of these notions (and these silly minimalist trash games are no longer OSR). I generally say if you roll less than 1/2 your HD, then add 1/2 your HD to the roll, so PCs are always in the upper half of their HD.
Example: Fighter rolls d8 for HP, gets a 3. Since it's less than (d8/2 =) 4, you add 4 to the roll and therefore this PC begins with 7 HP. So the possible rolls for a Fighter become: 1=5, 2=6, 3=7, 4, 5, 6, 7, 8. Low roll is 4 which only has 1/8 chance, then 5/6/7 each have 1/4 chance, and finally max HP has 1/8 chance. I call this a fake non-linear distribution model. Extrapolating all results into a table:
|
Roll |
d12 |
d10 |
d8 |
d6 |
d4 |
|
1 |
7 |
6 |
5 |
4 |
3 |
|
2 |
8 |
7 |
6 |
5 |
2 |
|
3 |
9 |
8 |
7 |
3 |
3 |
|
4 |
10 |
9 |
4 |
4 |
4 |
|
5 |
11 |
5 |
5 |
5 |
- |
|
6 |
6 |
6 |
6 |
6 |
- |
|
7 |
7 |
7 |
7 |
- |
- |
|
8 |
8 |
8 |
8 |
- |
- |
|
9 |
9 |
9 |
- |
- |
- |
|
10 |
10 |
10 |
- |
- |
- |
|
11 |
11 |
- |
- |
- |
- |
|
12 |
12 |
- |
- |
- |
- |
I think this strikes the correct balance between the widespread houserule of max HP at first level and the generally considered standard roll and pray method from TSR editions. In this case there are variable HP and you can still be somewhat screwed, but the distribution can ease the burden somewhat and eliminate 1 HP starting characters. It also has the benefit of still letting PCs roll and feel good when they get a max value, as they still have to roll that naturally by this system.
