PDA

View Full Version : dice rolling odds - reroll



slaanghoul
27-08-2008, 06:52
OK, I am math guy when it comes to WHFB . .. I try not to think and just play with the odds.

So I'm going to ask a question about reroll. Reroll is a good thing, but sometime I'm not so sure, so i want to get a few ideas from you guys here.

Battlestandard for demons and undead. The reroll is nice when you fail by a bunch. However, what if you fail by just one or two. . . should you take the lost or reroll it and take a chance of loosing worst!

Example: Demons LD7 lost combat by 4 = need 3 or more to past. First roll got a 4 . .. u loose one model. I would just take the one model lost and move on. To me . . . . how i think of reroll break test is I always base that my next reroll will be a 7. I just use 7 as a base and go from there.

Using 7 as a rule; another example. Demon LD9. lost combat by 1. = need 8 or less. I roll a 9, loose 1 model. I would reroll it .. because I think my odds will be in my favor and I will either roll 8 or less or worst roll a 9 and be the same. However, there is a chance of rolling 10 - 12 and take a bigger lost. Seems like a good risk.

So the question to you guys, what is the best logic and odds to figure this. Is my rule of 7 correct?

WLBjork
27-08-2008, 06:57
As a fast rule, yes.

I think Avaian has a calculator on his site somewhere which will calculate the odds for you.

Oh, and the Undead Battle Standard works very differently to the Daemon Battle Standard (mainly because the undead work differently to the daemons).

Braad
27-08-2008, 07:00
Depends how important that model/unit is. If you are about to loose 1 wound, but that means half of your wounds on something big and shiny, then you could re-roll. If it was one wound on a big wound on something with 20 models in it, then let it go.

I think this is more a situation how it happens and then decide, then to come up with some fancy rule to follow. Sometimes you just need to take risks, but most of the times this also depends on the situation at hand.

Loopstah
27-08-2008, 09:13
You don't get a re-roll with Undead, seeing as they don't take Break tests.

Tarax
27-08-2008, 09:15
You do not have to re-roll the dice. It's all up to you.

Whether the chances are for or against you I can not tell and frankly I'm not bothered.

This isn't as much a rules question, more a tactical question.

Good luck, and remember: The odds are always against you, even if statistically not. Just look at how well steeds perform compared to the riders. ;)

Condottiere
27-08-2008, 09:18
You don't get a re-roll with Undead, seeing as they don't take Break tests.I don't play Undead, only against them - is there anything printed?

death__lord
27-08-2008, 09:19
I don't play Undead, only against them - is there anything printed?

Yes, undead only looses one less wound from CR if in range of a BSB.

T10
27-08-2008, 10:14
OK, I am math guy when it comes to WHFB . .. I try not to think and just play with the odds.
(...)
So the question to you guys, what is the best logic and odds to figure this. Is my rule of 7 correct?

Well, if your roll is higher than 7, then a re-roll is likely to make it better. If it's below 7 then a re-roll is likely to make it worse.

But as Braad says, it's more a question of wether you think the loss of wounds from an otherwise ok roll is worth risking a re-roll.

Simply put: You have to think.

-T10

Badbones777
08-09-2008, 03:12
As a fast rule, yes.

I think Avaian has a calculator on his site somewhere which will calculate the odds for you.

Oh, and the Undead Battle Standard works very differently to the Daemon Battle Standard (mainly because the undead work differently to the daemons).

Yeah 7 is generally considered to be the average of 2d6. Though that said, GW dice (or rather the TYPE of dice that GW sell, the ones with slightly rounded egdes) apparently roll 1's more frequently than they should!

Lord Dan
08-09-2008, 03:48
It's more important to consider the context of the break test than to contimplate potential models-lost alone. For instance:

-Could losing 2 models instead of one take away a rank bonus? Outnumbering bonus?
-Is it important that you not lose even one model, so as to hold a rank bonus or the more important outnumbering bonus?
-Can you afford to lose up to 4 or 5 models instead of just one, with the potential to roll no wounds at all with your re-roll?
-If you hold them longer in combat, will you put yourself in a more advantageous situation? (For example, do you have other units on the way?)

Warhammer has math involved, but so little of it has to do with math. I always thought statistics were silly to contemplate anyway, when the chances of you rolling a 12 for every re-roll is exactly the same as you rolling double 1's. What you "should" roll doesn't matter with dice.

DeathlessDraich
08-09-2008, 09:26
Warhammer has math involved, but so little of it has to do with math. I always thought statistics were silly to contemplate anyway, when the chances of you rolling a 12 for every re-roll is exactly the same as you rolling double 1's. What you "should" roll doesn't matter with dice.

It's balancing chance/luck with probability.
If many dice are rolled then, it would be better to rely on probability than on chance but if one dice is rolled then probability is next to useless in assessing the outcome.




Example: Demons LD7 lost combat by 4 = need 3 or more to past. First roll got a 4 . .. u loose one model. I would just take the one model lost and move on. To me . . . . how i think of reroll break test is I always base that my next reroll will be a 7. I just use 7 as a base and go from there.

Using 7 as a rule; another example. Demon LD9. lost combat by 1. = need 8 or less. I roll a 9, loose 1 model. I would reroll it .. because I think my odds will be in my favor and I will either roll 8 or less or worst roll a 9 and be the same. However, there is a chance of rolling 10 - 12 and take a bigger lost. Seems like a good risk.

So the question to you guys, what is the best logic and odds to figure this. Is my rule of 7 correct?

1) Yes, (7) would be the modal roll for 2-dice rolls and a good benchmark.

2) Probability of rolling a (7) with 2 dice is highest compared to other rolls but is only 16.7%.
The next highest Probability is 6 or 8 -> 13.9%.

So I tend to use the 6,7 or 8 benchmark :p.


3) For a BSB and normal break test, probabilities of succeeding at different Lds:
Ld..............Probability
6...................66
7...................82
8...................92

slaanghoul
09-09-2008, 15:09
It's balancing chance/luck with probability.
If many dice are rolled then, it would be better to rely on probability than on chance but if one dice is rolled then probability is next to useless in assessing the outcome.




1) Yes, (7) would be the modal roll for 2-dice rolls and a good benchmark.

2) Probability of rolling a (7) with 2 dice is highest compared to other rolls but is only 16.7%.
The next highest Probability is 6 or 8 -> 13.9%.

So I tend to use the 6,7 or 8 benchmark :p.


3) For a BSB and normal break test, probabilities of succeeding at different Lds:
Ld..............Probability
6...................66
7...................82
8...................92

Nice, are you sure of this? So if I need to roll 8 or under, I have 92% chance?

Can you please list the probability from 2 to 12 if you don't mind. I always thought that 7 was about 75% and 8 was around 80%, wow. . .. eight is really good! Not too worry about putting unite with LD 8 away to the flank from my general anymore. I'm never too worry about break test because I like to pick my fight, but when somebody is shooting and magic that cause panic test . .. that is the problem for my Empire army. But 8 is golden for panic test.

FigureFour
09-09-2008, 15:59
Nice, are you sure of this? So if I need to roll 8 or under, I have 92% chance?

Can you please list the probability from 2 to 12 if you don't mind. I always thought that 7 was about 75% and 8 was around 80%, wow. . .. eight is really good! Not too worry about putting unite with LD 8 away to the flank from my general anymore. I'm never too worry about break test because I like to pick my fight, but when somebody is shooting and magic that cause panic test . .. that is the problem for my Empire army. But 8 is golden for panic test.

He's talking about with a BSB reroll. It's closer to 80% without the reroll.

narrativium
09-09-2008, 16:09
The actual probability, or approximation by percentage? Let's try both...

There are 36 outcomes from 2D6, condensed into 11 scores and 11 corresponding probabilities of rolling that score or under. A reroll has the same probability of achieving each score and you only want to use it if the Break test fails.

So, rolling a 7 or less on 2D6 has a 21/36, or 7/12, or 58.3% probability. Re-rolling the failing 5/12 leaves a success probability of (7/12) + ((7/12) * (5/12)) = 119/144 or 82.6% probability.

Here's the full list, in order: score (or less) needed - true probability - percentage - reroll probability - percentage:
2 - 1/36 - 2.78% - 71/1296 - 5.48%
3 - 1/12 - 8.33% - 23/144 - 15.97%
4 - 1/6 - 16.67% - 11/36 - 30.56%
5 - 5/18 - 27.78% - 155/324 - 47.84%
6 - 5/12 - 41.67% - 95/144 - 65.97%
7 - 7/12 - 58.33% - 119/144 - 82.64%
8 - 13/18 - 72.22% - 299/324 - 92.28%
9 - 5/6 - 83.33% - 35/36 - 97.22%
10 - 11/12 - 91.67% - 143/144 - 99.31%
11 - 35/36 - 97.22% - 1295/1296 - 99.92%
12 - 1 - 100% - 1 - 100% (waste of re-roll, certain event)

7's better than a 75% chance with a reroll because the odds of rolling a 7 aren't 50%; it's the median value, so the odds of rolling higher than 7 are equal to the odds of rolling less than 7, and therefore the odds of rolling 7 or less are better than 50%. More than 50% of failed rolls will also therefore succeed on the reroll.

WLBjork
10-09-2008, 05:02
narrativium, you need to be clearer that those probablilties are to score that value or less.

Completely confused me when I first read it, thought you'd messed it up totally!

Gokamok
10-09-2008, 10:41
....and just to clarify: You must use the re-roll from the BSB if the first test is failed (ie. above the value required not to lose wounds/break);)

BRB, p. 83 "Re-roll break tests".