10 Magic Strength Pretenders....

[Graeme Dice]

Thufir said:
I don't suppose anyone has ever posted a formula for how magic duel works?

Many times. 1d6 + skill vs. 1d6 + skill

The highest wins, the lower dies. On a tie, both die.

The probabilities are found in this thread:
Magic Duel and other dice roll charts

[Graeme Dice]

Graeme Dice said:

Many times. 1d6 + skill vs. 1d6 + skill

The highest wins, the lower dies. On a tie, both die.

The probabilities are found in this thread:
Magic Duel and other dice roll charts

If you are 5 points higher, then you will only die 2.7% of the time. That's about the same as having a MR of 22.

[Thufir]

Graeme Dice said:
If you are 5 points higher, then you will only die 2.7% of the time. That's about the same as having a MR of 22.

Excellent info and links, Graeme, thanks a bunch! And my apologies for not searching before asking

I find this encouraging, at a 2.7% chance with a 5 pt difference, it sounds to me like perhaps the astral combat mage pretender is a viable strategy. I still don't know enough about how magic duels unfold in MP game play to make an informed judgement, but it's quite helpful to know the formula.

[archaeolept]

i'm only going w/ the numbers zen mentioned. perhaps Magic Duel as currently implemented isn't working quite right. the numbers given by Saber Cherry were for non open-ended die rolls, for instance...

Frankly, I would suggest keeping your 500 pt Ghost King w/ 3 astral somewhere a long way from anyone else.

[Thufir]

archaeolept said:
i'm only going w/ the numbers zen mentioned. perhaps Magic Duel as currently implemented isn't working quite right. the numbers given by Saber Cherry were for non open-ended die rolls, for instance...

I haven't done the math (yet), but being open ended does not dramatically change the odds. It's certainly possible that this is not implemented the way the post indicates, but it seems unlikely that this is bugged (perhaps the implementation has been changed, or misinformation was propagated along the way). It's easy to test though, and I'll do that later today.

Frankly, I would suggest keeping your 500 pt Ghost King w/ 3 astral somewhere a long way from anyone else.

From a game design point of view, I think it's eminently reasonable that a 3s Ghost King is vulnerable because of having chosen astral. The thing I've been questioning is whether a 6s+ pretender is unduely vulnerable to astral hit squads.

[Thufir]

So, I've done some testing on magic duel now, and the results were a little bit confusing. I did 2 sets of 20 tests, one with S3 vs S8 and the second set with S7 vs S8

Here's what I've got:

S3 wins -- S8 wins

  0          20



S7 wins -- S8 wins -- Tie (no deaths) -- Tie (both die)

  2          16            1                   1

If my testing method is truly random (which I am begining to doubt), then the second set of results is surprising, in that S7 should win roughly 1/3rd of the time according to the formula for magic duel.

The one thing I'm uncertain of in doing my testing, is whether or not I am truly getting a random sample by reloading from saved files. Can anyone (ideally a dev ) tell me how Illwinter is setting the seed for their random number generator? I'd like to find a testing method where I'm sure that I'm getting a true random sample before doing more tests.

[archaeolept]

lol.

i especially like the "Tie (no deaths)" since, as presently understood, that shouldn't be possible

i also really suspect the random number generator is quite quirky anyways

but good work thufir. perhaps try a different situation in a different game...

also, if you're using 2.13, it could be that the devs have altered the method...

[Nagot Gick Fel]

Thufir said:
The one thing I'm uncertain of in doing my testing, is whether or not I am truly getting a random sample by reloading from saved files. Can anyone (ideally a dev ) tell me how Illwinter is setting the seed for their random number generator? I'd like to find a testing method where I'm sure that I'm getting a true random sample before doing more tests.

Don't bother with testing Magic Duel in 2.13. The formula is slightly bugged and you may not get accurate results.

[PDF]

Nagot Gick Fel said:

Don't bother with testing Magic Duel in 2.13. The formula is slightly bugged and you may not get accurate results.

Holy Sh*t ! When will this spell behave like something that can be expected ?