"Probability Benchmarks for new Mechanics" Topic

When designing dice mechanics do other people have benchmarks for probabilities of common attacks/actions? What are they? I come up with some common attacks that need to fit to certain probabilities and then fit my dice mechanics around them in the simplest way possible that will achieve all my requirements.

For example, in my ww2 rules, I try to make the maximum probably of a tank gun hitting another tank at the closest range band be ~50-60%. For small arms firing at cover and long range band, I need the probability of casualties to be 5% or less.

It could be for any era, do musket era players like a particular rate of musket casualties between line battalions?

Probability of hit for how many shots? If you're using 5% for one rifle shot / round at 'long'(?) range against a target in cover everyone will be dead in no time. You need to establish what period of time this firing is happening over.

The period of time being a turn. It's arbitrary how many shots are fired in a turn since it is boiled down to one roll resolution.

5% chance of X men inflicting X casualties per turn.

I will just make a general comment. Muskets were not very accurate pre ACW. I think a lot of games allow for too many hits. But I would research the firing accuracy, range, and tactics for each era. You can usually find some decent data to go by and come up with an historical result.

I believe the Austrians in the Napoleonic wars fired only a couple of practice shots a year in order to save powder. The French practiced more.

I've rarely seen what a game defines as a "range band". Since guns with a flatter trajectory and higher muzzle velocity I'd use the rounds time of flight in half seconds to determine hit chance.

Look on page 175: PDF link

So a gun with a muzzle velocity of 800 mps would have a range band at 400m, 800m, 1200m, 1500m, etc. Call them short, medium, long and very long.

If you look at the 50% dispersion (far right column) at 1500m you'll see it is 1.0m in height (high or low from aim point). This should mean you have slightly less than a 50% chance of hitting a target 2.0m high like a T-34/85 IF you know the exact range to the target or you are already ranged in on it (it's been hit one of more times). Smaller targets are harder to hit and large targets easier. Good crews have a better chance and poor crews worse. Some guns may be inherently less accurate. A rangefinder might allow for the first shot to be ranged in.

The big variable is that crews normally had a 20%-25% range estimation error on the first shot unless using a rangefinder. So a target at 1000m might be estimated from 800m to 1200m and most likely miss. The could decrease the chance to hit on the first shot by about 50% of the ranged in value.

To simulate real tank gunnery tactics the first ranging shot at 1500m (very long range band) might be 20% chance to hit because of the range estimation error, the second shot (after a bracketing correction on the missed ranging shot) might improve to 35%. By the third shot you'd be ranged in with a 50% chance to hit and it never gets any better.

If your ranging shot hit then you are now ranged in for all successive shots even if some shots did miss because of the natural dispersion of the shot.

At very close range of 25m to 100m a shot often missed because the gunner hurried the shot and the parallax difference between where the gun and sight are pointing. You could have the turret in the cross hairs but the gun is pointing just off to the right of the turret and you miss.

I'd advise looking at Panzer War Micro Armor Rules: panzer-war.com

5% chance of X men inflicting X casualties per turn.

You could use a binomial table to get the results quickly.

A British War Office Report showed that an infantry section would produce about 1% per minute causalities against an enemy section in a trench.

So as Gauntlet said, it's all about the time scale you are using. In the tank gun example I gave you'd need to use about 10 second turns to get a realistic rate of fire.

It's going to get real complicated.

Wolfhag

Assuming the game is the thing, in a game, what "feels" right (or fun) to the player is more important than what is accurate for the history or capabilities of troops, weapons and equipment.
The player wants a decent chance of "doing something significant" when he attacks, and his opponent wants a decent chance of surviving that attack— but in both cases, they typically expect something to happen. If it doesn't happen, and doesn't happen too frequently or consistently, then players may dislike the game, even though it's as accurate to real world probabilities as possible. And that goes for things happening too easily as well. The chances shouldn't feel like "never", nor feel like "always"— even if they actually aren't either.

This isn't to say one should design a game to be easy— the player should face a challenge of maneuvering his forces and properly selecting the suitable forces to face their foes, or figuring out how to overcome apparent insufficiencies and improbable odds. But math done to historical precision isn't always what it's cracked up to be; leeway for the "feel" is going to be the better choice. How does one determine the feel is right? Testing it out— experiment with different probability structures till you have one that seems right to you.

Again, the above assumes that what one is trying to create is placing an emphasis on "game" over "simulation." If one wishes to produce the latter, then my comments really don't apply.

Yeah Parzival, I'm definitely going for feel over simulation. My numbers are just numbers I make up because they will influence players to do what feels believable and is fun.

The 5% example is something that can happen but will never be relied upon for your strategy to hold together. Fire under that condition is for suppression mainly.

The purpose of my post has nothing to do with realism. I just wondered if other people have simple numbers for common actions that they use as constraints for the mechanics.

Gauntlet,
My mistake. I took benchmark to mean "evaluate or check (something) by comparison with a standard" with mostly historic verifiable data and examples being the standard.

There is nothing wrong about going for "feeling" but I think the benchmark for "feeling" is what you, not someone else thinks it should be. If you have a realistic simulation the players cannot relate to it won't give them the right "feeling". The visuals go a long way to creating that.

The purpose of my post has nothing to do with realism.

So what is it that you and your players are wanting to "feel" in the game? Normally feedback from players is whether or not they "felt" the game was realistic or not.

Wolfhag

I think the question really is not the hit probability, but the tactical effect probability. If you think that a well-trained infantry battalion should be able to stop a cavalry charge by fire, you need the numbers which will accomplish that--which in turn involves your morale rules, and maybe other things. We used to figure in CLS that only about half the figures removed from the table were actually casualties: the rest were "knocked loose from the unit" or were "helping an injured comrade to the rear." Tank on tank is one thing. Battalion vs battalion or brigade vs brigade, fire has to be considered in conjunction with morale.

For example, in my ww2 rules, I try to make the maximum probably of a tank gun hitting another tank at the closest range band be ~50-60%. For small arms firing at cover and long range band, I need the probability of casualties to be 5% or less.

It could be for any era, do musket era players like a particular rate of musket casualties between line battalions?

Gauntlet:

Could it? Based on what, pray tell? This strikes me as in the neighborhood of WAG, or Wild A.. Guess. Wolfhag had the reasonable approach by going to the Army's statistical answers to that question… I mean they were and are seriously interested in your question. If you are interested in portraying something historical/from reality that is one source. after all, it is a statistical question: How often does X occur…in the real world?

You basically have 3 choices:

1. Do whatever feels right. Often designers go with percentages that work for the game, then call it 'historical'. [lots of examples of that.]

2. You can do what you are doing here, group think/crowdsource, or find out what the majority of gamers think 'feels right.' If you are designing this for the public to some degree, it is reeeealy hard to come up with something that most players will say 'feels right' among a group of even four or five. Many designers look at popular games and use some variation of the statistical outcomes found there. Better odds of finding a 'feels right' zone for your probabilities. It does produce a 'same old--same old' feel though. It is no wonder that many games look like many previous designs.

3. OR you can do what simulation designers [and the military do], they build the statistical base from real events, most often recorded. That will get you much closer to reality, a one-to-one relationship between your hit ratio and real combat.

I have found that player intuition, or general intuition for that matter, about the frequency of an event or the reason for the event too often ends up being counter-intuitive. We do this all the time. For instance, often people will stop putting chlorine in their pools when it starts smelling like chlorine, but in reality, it means you don't have enough chlorine in the pool.

My own efforts in developing a statistical base for the Napoleonic wars, I have found a number of things that do not match most all Napoleonic War games. For instance, how often do you think one side or the other will retreat from a firefight? 20% of the time, 50%? Regardless of the casualties, looking at a random set of 50 firefights across 20 years of war, less than 10% of firefights saw one side retreat without some other event intervening. That is less than 5 of 50.

Statistically, 50 random examples is pretty good with that stark a result. Units engaged in firefights would only retreat if charged, outflanked or the other side was reinforced. A straight-up firefight would just continue until one of those events.

Not what one would expect, and certainly not what most all rules sets portray.

It's your game and you can design it anyway you want with any probability schemes. If there is a desire to have those probabilities reflect reality at all, you first have to have some idea about what the actual probabilities were.

I think the question really is not the hit probability, but the tactical effect probability.

Robert P. :

I agree. It is near impossible to link casualties [hit rates] to unit behavior [i.e. when units take 20% casualties, they will retreat 50% of the time.] We know troops behave all sorts of ways that can be only marginally related to actual hits/casualties. And that is assuming you have a reasonable idea of the casualties suffered in the first place, which most often you don't.

Outcome [behavior] frequency is a better basis for combat effects. [It is also WAY easier to establish.]

I guess I must have badly mangled my question.

My question was about game design.

When you design a game, you decide what you want to happen in different situations (could be based on real life or just what feels right). Then you design the mechanics to give you the result you envision.

For example suppose I decide that my game is going to use Wolfhags 1% per minute example and say a turn is 5 minutes. I can use 5% chance per turn as a benchmark to test my dice system to make it will give me roughly that result in those conditions on average.

I have come up with dozens and dozens of possible ways to use dice but most of them aren't able to satisfy all of my benchmarks simultaneously.

I was curious if anyone else has "standard probabilities" to quickly check new dice systems.

For a basic example I'll say I roll 2 different colored dice. One is for cover and requires a 6 to hit a slit trench, one is for range and requires a 5 or 6 to hit at this range.

That gives a 5.5% chance, close enough to my chosen "standard".

Of course, this system is too simple since it doesn't account for the weapon firing or status of firing unit. But it's an example.

It gets more complicated when you have interactions between parameters like range affecting armor penetration and accuracy separately.

Here is a link for bench marking small arms fire accuracy and effectiveness: PDF link

PDF link

The small arms vs infantry rules I use are when a target is hit there is a D6 cover save by the defender based on the amount of target exposure (1-6 feet exposed). Standing in the open a 1 is No Effect or a light wound, prone behind cover or in a trench is a 1-5 for No Effect.

Using a cover save eliminates the need to modify firepower values and gets the defender in the action.

Wolfhag

I had to up the 'to hit' in my WIP skirmish game (to 4+ on a D10) because anything lower would be too deadly at short range (although you can lower it some by 'aiming'.) On the plus side, most firearms have 'unlimited' range (not really a problem on a tabletop ;-) and I also got rid of almost all weapon stats/tables (much, much less to memorize… ;-)

Wolfhag, the cover save is a good mechanic functionally, but I don't like that it almost doubles the amount rolling necessary.

I'm a sucker for fast mechanics.

Gauntlet,
Well it depends on what you want the game to reflect. In my game cover saves are not a burden anyone as the shooter and target player both roll at the same time or solo roll two dice at once, not a big deal. It speeds up the game too as the shooter is not searching for (and forgetting) firepower modifiers.

My gaming experience and play testing shows that saving rolls, while not necessarily realistic, they do result in generating suspense in the game because the result is unknown. Players find cover saves suspenseful and entertaining and there is a historical basis for them. It makes players think about cover too.

Also, when a shooter rolls for the hit location the defender also rolls a die at the same time for the chance of a ricochet. While not common, historically rounds that should have penetrated did occasionally ricochet.

For me, playable mechanics that contribute to the entertainment value of the game and have a historical basis that players like it I keep them.

Are you bench marking against other abstracted game mechanics or historical outcomes?

Wolfhag

An opposed roll is a great way to implement a saving-roll like mechanic.
Both players are involved.

Yeah Parzival, I'm definitely going for feel over simulation. My numbers are just numbers I make up because they will influence players to do what feels believable and is fun.

Gauntlet:

I take it you were just asking other gamers what numbers they 'make up' for combat ratios. That's fine, but you originally asked:

When designing dice mechanics do other people have benchmarks for probabilities of common attacks/actions? What are they?

And you got answers about what others can and do use as benchmarks. Regardless of what kind of probabilities are used, it is still game design. It is all about what the game ratios represent. Going for fun is a given, regardless of the source for those probabilities. Designing for 'feel' is fine, but a rather nebulous goal at best, even for only a few players. Often, it is simply going for what players 'expect' from past experience rather than any feel of reality.