I'm making a horror homebrew. The character's stats range from 1-5, 1 being very little talent, 5 being high talent.

My question is how often should the character with the 5 stat succeed at the given task? What about the character with 1 in the same stat doing the same task?

I want there to be some tension, and have success not be taken for granted, even at a high level of talent. But I also don't want it feel hopeless even trying to do anything. A fine balance.

It's a tricky line to walk, I agree.  I generally find the following points are needed for both a plausibly "realistic" and potentially exciting probability weighting:

1) In general, people with an average level of competence working on appropriately "average" tasks in that area of competence tend to succeed in practice far more often than they fail, given appropriate preparation time, context, and lack of unusual obstacles.  So "average" ability (I don't know if this is score 2 or 3 on your scale) can be relied upon with, I would say, 95% or better consistency for moderate success vs. a task of "average" difficulty.

2) In contrast to this, even people of exceptional competence won't succeed at exceptionally difficult tasks with the same consistency -- they will typically do better than 50% but seldom better than 75%-80% reliability.

3) In horror specifically, one of the things that puts tension into the challenges is that PCs are hindered not by attempting exceptionally difficult tasks, but by attempting normally mundane tasks under exceptionally difficult circumstances. Running up the stairs without tripping is easy; doing it when you're dizzy, exhausted, wounded and panicking from the sound of the Masked Killer's heavy boots tromping closer and closer is harder than you ever expected.

If your 1-5 Stat represented an additive D6 die pool, for example, and 3 was your baseline average, rolling 3d6 vs. Difficulty 5 ("Routine") will equal or beat it 98% of the time.  But throw in -1d for Injury, and -1d for Panic, and suddenly it's only 1d vs. 5 -- a 2/3 chance of failing.

Conversely, a 5d pool will beat Difficulty 10 ("Challenging") the vast majority of the time, but still has a reasonable chance of failing at Difficulty 15 ("Exceptional") and less than a 50% chance vs. Difficulty 20 ("Heroic") -- and a 5d pool with the same penalties for Injury and Panic will only have a 50% chance of succeeding even vs. what is normally, for him, a "simple" Difficulty 10 task.

So I think the key is not so much to create a unique probability distribution in itself, but to make sure there are plenty of possible circumstantial modifiers to make things more difficult.  For optimum results, make sure that none of the modifiers is in itself significantly large, but that they just keep piling on top of one another making things worse ... and worse ... and worse....

It's my first homebrew, so I was trying to keep it simple by having every task be the same difficulty, with probablity being decided by the person's talent, and possiblely consequences they currently suffer. My thought was using a nd6 dice pool, where n is the stat you're using. If you roll any 6s, then you succeed.

However, the percentage of success for a person with a 5 stat is only about 60%. I looked it up, and the d6 dice pool was designed for 10 dice. If I double the number of dice to match, where n stat rolls (2n)d6, then the 1 stat person gets a success rate of about 30%, which I think is a bit too high for a horror game. Correct me if my assumptions are wrong.

One thing about the consequence thing is that, even though I understand the horror of accumulating penalties, it's hard to add granular modifiers to the d6 dice pool, since every modifier affects the result a pretty significant amount.

In practice I'd say a success rate of 30% is bad enough, even in a horror game, to induce dread in most players forced to face it, especially since -- if every task is basically going to be the same "difficulty" -- it sounds like your PCs will only be expected to roll for stuff that's significant, dangerous and unusual anyway, stuff with significant consequences attached to failure.  Any PC action which doesn't meet this criterion could be treated as a simple "Can you do this? If yes, you succeed; if not, you fail" handwave.

One possible alternative mechanism might be the one used in the out-of-print Doctor Who game called Time Lord.  In that game, stats ranged basically from 1-6, while difficulties ranged from 1-10.  If your stat is higher than the difficulty, you succeed automatically. If it's equal to or less than difficulty, you roll 2d6, but subtract the lower die from the higher -- thus rolling a 2 and a 4 gives +2, a 6 and a 3 gives +3, etc (ties equal a roll of +0).  Your stat + rolled bonus must beat the difficulty to succeed.  (In the original game, ties equalled failure; I've done a few homebrew riffs which designate a tie as a complication, a partial success with skewed side effects.)

For you a simpler way to do it might be to say that any test of ability where a character's survival, health or sanity is on the line is a Crisis Check: you add your Stat to a roll of (high d6 - low d6) and must get a total of 6 to Escape. Failure by varying degrees imposes cumulative bad consequences, up to death.  So somebody with an average Stat 3 needs to roll +3 or higher, which succeeds about 33% of the time; somebody with a Stat of 5 needs to roll +1 or higher, which happens about 83% of the time; and somebody with a Stat of 1 needs to roll +5 (the highest possible roll), which only happens about 6% of the time.  Does this sound more like the distribution you were thinking of?

Yeah, basically they roll for actions that a normal person might fail at, but aren't impossible.

Something that someone else suggested was a 2d6 + skill level roll with a target number of 10, which ends up as below.

Skill  TN 10    
1      16.7%  
2      27.8%    
3      41.7%  
4      58.3%  
5      72.2%

What do you think of this? I think it's simple enough, and if I need to change the probablities, I can do so by changing the target number.

Quote from: saevikas;1114708Yeah, basically they roll for actions that a normal person might fail at, but aren't impossible.

Something that someone else suggested was a 2d6 + skill level roll with a target number of 10, which ends up as below.

Skill  TN 10    
1      16.7%  
2      27.8%    
3      41.7%  
4      58.3%  
5      72.2%

What do you think of this? I think it's simple enough, and if I need to change the probablities, I can do so by changing the target number.

Looks pretty straightforward -- reminds me of Barbarians of Lemuria, if you haven't seen it.

Out of curiosity, what is your baseline "average" Stat value?  Is it 2 or 3?  If PCs Stats tend more to the 2-3 range than the 4-5, some sort of Drama Point mechanism where they can survive a failure or two if they have to would probably help. Either that, or there should always be emphasis on players finding ways around rolls they look likely to lose, rather than making them guess/gamble on which rolls they can afford to lose and which they can't.  (To riff on the old "say yes or roll the dice" gag, this game could be something like, "Never ask to roll the dice, or you're in trouble.")

Btw, the math was wrong on that, that's for target number 9.

I'm thinking of having a mechanism where they can reroll a failure, at the cost of some of their health. It'll hurt when health inevitably becomes precious. I'm not sure what the average should be, I'm thinking about a 3.

Welcome aboard saevikas!!!

Quote from: saevikas;1114632I'm making a horror homebrew.

Define "horror" as you view it for your game.

Horror is a massive genre and you need to declare where your game fits in the genre so players understand the tropes and expectations for your game. Is your game Texas Chainsaw or Poltergeist? Same director, rather different movies.

Quote from: saevikas;1114632My question is how often should the character with the 5 stat succeed at the given task? What about the character with 1 in the same stat doing the same task?

If grandma has STR 1 and a bodybuilder has STR 5, there should be significant differences in success for tasks involving muscles. AKA, if the task is smacking a serial killer with a bat, grandma is only going to impress the villain on a critical and that bodybuilder is going to beat the teeth out of the crazy killer most of the time.

Unless your horror game is cinematic, not realistic.

Which would be perfectly fine, but it's a choice you need to decide upfront in game design.

BTW, do you own Kevin Crawford's Silent Legions? It's a goldmine of ideas for horror gaming.
https://www.drivethrurpg.com/product/145769/Silent-Legions