Question about Luck advantage: Do you re-roll all three dice?

[Ulzgoroth]

Originally Posted by Refplace

How is it worth less?
Odds are good you have at least one average and one low value (in this case a high number for a to hit roll say) so rolling that one bad die has a better chance of benefit.

I haven't broken it down to analyze, I admit, but consider that the set of rolls you want to use Luck on is not going to have the same distribution as the set of all rolls.

For instance, if you roll an 18 and have Luck handy you may very well want to use it. If you re-roll one die of an 18, you're not going to get below a 13 and even re-rolling it twice you still have a 1/9 chance of being stuck with a 17 or 18. If you reroll all the dice of an 18 twice you're strongly expected to to better than a 10 and vanishingly unlikely to be stuck with a 17 or 18.

[thulben]

Originally Posted by Peter Knutsen

Luck with some Cosmic slapped onto it.

It's not an at all unreasonable idea. It's just more powerful than what the designer originally envisioned Luck as doing, and therefore it needs to cost more.

So, I whipped up some code for a simulation and here we go:

Spoiler:  

      STD    LCK    NEW
 3:   0.5%   1.4%   3.6%
      0.5%   1.4%   3.6%
 4:   1.4%   4.1%   7.9%
      1.9%   5.4%  11.5%
 5:   2.8%   7.8%  12.0%
      4.7%  13.2%  23.5%
 6:   4.7%  12.0%  14.4%
      9.3%  25.2%  37.8%
 7:   6.9%  15.9%  14.8%
     16.2%  41.1%  52.6%
 8:   9.7%  18.2%  13.5%
     25.9%  59.3%  66.1%
 9:  11.6%  16.2%  11.3%
     37.5%  75.5%  77.4%
10:  12.5%  11.9%   8.6%
     50.0%  87.5%  86.0%
11:  12.5%   7.2%   6.0%
     62.5%  94.7%  92.0%
12:  11.6%   3.6%   3.8%
     74.1%  98.3%  95.9%
13:   9.8%   1.3%   2.2%
     83.8%  99.6%  98.1%
14:   6.9%   0.3%   1.1%
     90.8%  99.9%  99.2%
15:   4.6%   0.1%   0.5%
     95.4% 100.0%  99.7%
16:   2.8%   0.0%   0.2%
     98.2% 100.0%  99.9%
17:   1.4%   0.0%   0.1%
     99.5% 100.0% 100.0%
18:   0.5%   0.0%   0.0%
    100.0% 100.0% 100.0%

How to read the above: The first column represents the standard "roll 3d". The second is the standard luck advantage. The third is "re-roll any single die two times". Just to be clear on what I mean by that, let's say that I roll "6, 5, 4". Re-roll the 6, get a 1. Re-roll the 5, get a 2. That's it.

For any given sum of the dice, there are two rows in the table. The first is the chance of hitting that number exactly, the second is the chance of hitting at most that number. I wasn't expecting the modified luck to be so powerful, but there you go.

[Ulzgoroth]

Originally Posted by thulben

How to read the above: The first column represents the standard "roll 3d". The second is the standard luck advantage. The third is "re-roll any single die two times". Just to be clear on what I mean by that, let's say that I roll "6, 5, 4". Re-roll the 6, get a 1. Re-roll the 5, get a 2. That's it.

I think that's a weird interpretation of 'reroll any single die 2 times'

[Kuroshima]

Being able to reroll up to 2 dice is basically the same as roling 5d6, and keeping the 3 lowest (slightly worse, but not by that much). Instead of programming Monte Carlo simulations, I prefer to use Anydice ;)

Here's a graph of what each kind of luck looks like

[sir_pudding]

Originally Posted by Kuroshima

Here's a graph of what each kind of luck looks like

The "lowest of 5d" variant appears to have a significantly lower mean than normal luck.

[Kuroshima]

Originally Posted by sir_pudding

The "lowest of 5d" variant appears to have a significantly lower mean than normal luck.

0.42 lower mean is certainly appreciable, but when both kinds of luck lower the mean by at least 2.5, I would not call it "significantly lower". Also, Std dev is higher by 0.43 for Monte Carlo luck.

[sir_pudding]

Originally Posted by Kuroshima

0.42 lower mean is certainly appreciable, but when both kinds of luck lower the mean by at least 2.5, I would not call it "significantly lower". Also, Std dev is higher by 0.43 for Monte Carlo luck.

Yeah, I'm just surprised the mean is lower at all. I suppose if I had thought about it that would have been obvious, but my gut feeling was that standard luck was generally better overall.

[Ronin Rabbit]

You guys are awesome. I'm just the type who'll come up with an idea and make a quick and dirty rule that's good enough for gov't work. You guys do the math and show your work. I wish I found this board five years ago lol.


So mind if I throw another monkeywrench in the works?

Bare with me on this-- I don't have rules nearly as down as I used too-- but I thought I remembered reading that you could use advantages or disadvantages as enhacements/limitations on other advantages and limitations. Something about the percent is equal to the cost of the advant/disadvant?

That got me thinking about guys that are "lucky" at one thing but not everything. I wouldn't want to mess with applying a luck enhancement on a skill, but maybe applied to a talent you could create a character who was lucky within a certain field.

If you guys don't like that idea... how would you represent someone who's lucky but only within a small field of activities?

Would you just limit the luck advantage?

If you go with the talent thing, then do you allow players to buy the enhancement on multiple sets of skills?

[vierasmarius]

Originally Posted by Ronin Rabbit

Would you just limit the luck advantage?

This. Aspected Luck (-20%, B66) is great for experts who rarely make mistakes (letting them reroll those inevitable 1-in-216 Crit Fails). A Talent is generally of the right scope to serve as an Aspect, but a particularly narrow one could qualify as a worse limitation.

[Ronin Rabbit]

Originally Posted by vierasmarius

This. Aspected Luck (-20%, B66) is great for experts who rarely make mistakes (letting them reroll those inevitable 1-in-216 Crit Fails). A Talent is generally of the right scope to serve as an Aspect, but a particularly narrow one could qualify as a worse limitation.

See, this is another of those nice things 4e had that 3e didn't... at least you didn't name another book that, to quote the kids from Dork Tower, MUST BE MINE lol