SFR, Inc.

[otto8064]

I just wanted to get a little dialogue started to see what different methods players use to rate dice on playability. I spent an entire weekend creating my own formula and tweeking it, and I'm still making improvements. But it gives a fairly accurate rating based on my personal preferences in dice.

For six-siders I started by listing the five basic actions, and plugging in the likelihood of rolling that action

Oak Lord
Melee - 83% (5/6)
Missile - 17% (1/6)
Maneuver - 17%
Save - 33% (1/3)
Magic - 17%

then I made a list of every SAI in the game and made up some percentage modifiers, for instance for Smite, I have each face with smite count as melee plus each smite icon adds 2% to melee:

Oak Lord
Melee - 95% (6 smites x 2=12% + 83% = 95%)
Missile - 17%
Maneuver - 17%
Save - 33%
Magic - 17%
(I like to think of it as the OakLord is 95% reliable in a melee fight)

Next I figure out a "multiplier":
For Target Categories (Melee for heavy melee, Melee and Save for light melee, Missile for missile, Melee and Maneuver for cavalry, and Magic for magic) if the % is:
>= 33% the multiplier is 0.3
>= 50% the multiplier is 1.0
>= 75% the multiplier is 1.3
>= 83% the multiplier is 1.7
>= 90% the multiplier is 2.0
For Save and Maneuver categories if the percentage is:
>= 33% the multiplier is 0.3
>= 50% the multiplier is 0.5
Non target categories don't get counted (i.e. missile faces on a heavy melee unit).

so for the Oak Lord we have:

Melee - 95%
Missile - 17%
Maneuver - 17%
Save - 33%
Magic - 17% 2.3 (TC >= 90%= 2.0 + SC >= 33% = 0.3)

add every category % up (179) and subtract 83 for 6-sided dice (to account for ID icons) to get 96

Oak Lord
Melee - 95%
Missile - 17%
Maneuver - 17%
Save - 33%
Magic - 17% 2.3x96 =221 this is my rating for the Oak Lord

I do it a little differently for 10-sided dice but it still turns out fairly accurate, the biggest challenge I have is in assigning modifiers that are fair and accurate to the SAI's. Also, it really measures the qualities I want it to measure, personally I prefer dice that are specialized but that can still save and maneuver.

[DEEPBLUEB2]

Mark A. Wiker created a Racial Analyses

You might find it of intrest.

[otto8064]

Thanks Jim! thats gonna be helpful when I get around to integrating racial abilities to the rating. I mostly wanted a system that gave values to the SAI's, because some are mostly worthless and also because theres differences in how many icons are on a dice like only 3 cantrips on an ashbringer or 4 flies on a daybringer that I wanted factored in.

[otto8064]

for those interested, here's what I've got as the top 3 3-healths for each category:

Melee 3-health Avg184 Mean192
#1 - Vindicator 237
#2 - Oak Lord 221
#3 - Warlord 212

Light Melee 3-health Avg225 Mean215
#1 - Filcher 283
#2 - Harbinger 283
#3 - Noble Willow 232

Missile 3-health Avg208 Mean237
#1 - Pine Prince 327
#2 - Assailer 325
#3 - Devastator 283

Cavalry 3-health Avg269 Mean273
#1 - Leopard Rider 405
#2 - Mammoth Rider 351
#3 - Bear Master 294

Magic 3-health Avg164 Mean179
#1 - Magi 213
#2 - Eldar Dryad 212
#3 - Ashbringer 180

[chuckpint]

Bruce Parrello when he made DDAssist, tried to make up a rating system. It shows up as the "Points" column in the die finder page. It's currently broken because it takes into effect charging. The interesting thing is that most cavalry units have high points because of that. I've never tried to tinker with that code, but I suppose I could fix it sometime. It's been a long time since I've programmed in Pascal...
_________________
You can never have too many dice.
First Place at the first ever Daemon Dice sealed starter tournament.
Battlefest tied for first GenCon 2012
Single Race Champion GenCon 2008-2009, Sealed Box Champion GenCon 2007,2009,
My collection is 21,500 Dragon Dice, 20,000 Daemon Dice, and others (too many to count).

[ddicerc]

Bruce's article on "Points o EFfect" is on the DDRC web site (with Bruce's permission). It's a little outdated since there have been rule changes since it was written (not to mention Treefolk being released), but the basic concept can be found there.

http://mysite.verizon.net/DDiceRC/DDTP2.htm
_________________
Steve "DDice" Braun
The Diceman's Gaming Pages: http://ddicerc.weebly.com
2014-2015 U.S. National Champion
(Just when you thought it couldn't get any worse...)

[otto8064]

ok I see how he does ratings mathematically, but the problem I see is that it doesn't translate into 'playability', for instance Genie gets a pretty high rating but Galeforce is almost useless but actually improves the overall rating, and some SAI's like Web and Smite are equal in their rating value because they are both used in melee, since Web doesnt kill dice, it seems like it should contribute to Saving and not to Melee as much as a Smite does since a Smite kills with no questions asked.

[DEEPBLUEB2]

Do you give a rend more points since it can accumulate more hit points?

[chuckpint]

That's the problem with it all right. While the number of different rolls the icon is useful in, is an important measure, I agree it's impossible to say with a straight face that Web is just as good as Smite. Not to say you couldn't have some factor for playability.

There is another factor that comes into play here as well. That is what you want the army to do. If I wanted to play an army that is mostly a missile army, Smite would be low on my list. Stone would right up there, on the other hand. So it depends on what you are looking for. Bruce had in DDAssist a bunch of columns for all Maneuver, Magic (broken up by the 5 elemental colors), Missile, Melee, Save vs Missile, Save vs. Melee. You could give each column a weight from 0 to 1.
_________________
You can never have too many dice.
First Place at the first ever Daemon Dice sealed starter tournament.
Battlefest tied for first GenCon 2012
Single Race Champion GenCon 2008-2009, Sealed Box Champion GenCon 2007,2009,
My collection is 21,500 Dragon Dice, 20,000 Daemon Dice, and others (too many to count).

[otto8064]

exactly, thats why I had to create seperate modifiers for each SAI

like Jim asked about Rend, I have it counting as a face of melee and a face of maneuver then a little extra for each icon because it allows you to reroll, double strike gets that little bit extra too since it also rerolls but it doesn't count as a face of maneuver so the Leopard Rider ended up with a higher score than the Lion-folk

but since I'm the one who assigns values to the SAI's, its a direct reflection of the attributes I value most

[KilljoyKing]

You can actually figure out, to a certain precision, how much a rerolling face like Rend contributes to melee, thus showing just how truly borked no-limit rerolling is.


Rider has 3 ID, 4 Melee, and 6 Rend across its faces, so (13 Melee / 6 Sides / 3 Health = ) 0.72 per point. Two of these sides, however (2/6 = 1/3) give the chance of reroll. This reroll gives the same point value as an initial roll (0.72) plus the same chance for another reroll.

Maybe it'd be easier if I separated the SAIs...

Rider Roll Breakdown:
2 / 6: No Melee (Saves and Missiles)
1 / 6: 3 Melee
1 / 6: 4 Melee
2 / 6: 3 Melee + Reroll
This is still the same result as above -- just expressed in a different form (and omitting rare health, which is really just the same and dividing the whole deal by three). Writing out in full parenthetical form:
(1/6 * 3 + 1/6 * 4 + 2/6 * [3 + 1/6 * 3 + 1/6 * 4 + 2/6 * {3 + 1/6 * 3 + 1/6 * 4 + 2/6 * .... } ] )
... and it goes on like that. Quite infinite, of course, but two-sixths tapers out after so many powers. I mean, 2/6 to the fourth power is only one percent (1/81) -- possible, but that is really gonna stop affecting the final number.

Still ran it for one hundred plies (only because I rigged a script to do so, although the difference between ten and one hundred runs wasn't really noticeable), and I got.... 1.08.... wow. On average, the Leopard Rider will roll one melee result per health, or an average of three melee.

Similar results apply to other rerollers, but, y'know, none of them have such a high chance of reroll -- not even the Tigerfolk, and it's SAI doesn't even count as maneuver.

To give you a comparison, a Bear Master only has an average melee showing of 0.8 (0.666 without rerolls). That's quite a difference...


So the reroll modifiers have very set as to what they should or should not be; that's just probability and statistics, which you cannot discount as it is the entire driving force behind the whole study. Other SAI, on the other hand, are more tricky. How do you accurately measure the benefit from the no-saves of Smite? Or the undefendable blows dealt by Counter? How does picking your target with Bulls-Eye measure up?

Certainly, Counter is worth more than two counts each icon (save, melee, and a magic-only defendable hit?). Bulls-Eye, depending on the targets available, can be just as good as a 1/6 (or 1/10) chance of death.
Smite is similar to double the melee results; four unblockable hits are akin to rolling eight melee with four saves on the enemy side. But there is no guarantee that the enemy would have rolled said hits. Or, from another perspective, a small army attacking a larger or better protected one may only be able to damage it through such SAI! Dunno what the weight should be for it, but prolly more than a measly two percent.
_________________
Dragon Dice Musing and Such @ Reality's End

[otto8064]

I have rends set up as a 3% melee modifier per icon which for a Leopard Rider Rend equates to 2 faces melee (33%) and 2 faces maneuver (33%) plus the modifier (18%) so he looks like this:

Leopard Rider
Melee - 85%
Missile - 33%
Maneuver - 50%
Save - 33%
Magic - 17% 3/135 405

which makes him the highest rated 3-health

when I was assigning values to SAI's, icons like fly only get counted as a face of save and maneuver because that's all they count towards. The ones that got modifiers had that little extra oomph like rerolls or no saves.

[cliffwiggs]

[otto8064]

Clearly I'm going about this all wrong

Okay if I want the average 1-health to score 100, the average 2-health to score 200, and the average 3-health to be 300 etc. so that each score is divisible by their health points to be around 100...

and the average 6-sided dice has 1 icon per health per face...

then each primary icon is .17
each secondary icon is .16
and each tertiary icon is .15...

Example:
Dwarf Footman 1H
ID
1 Save .16
1 Save .16
1 Melee .17
1 Melee .17
2 Melee .34 =1.00

then a Rend icon on a Leopard Rider is .17 maneuver, .185 melee?
(.17 x 1.08333333 from cliff) or....

would it be .17+(.17x.6767676767)
which is the odds of rerolling another melee, with the possibility of it also being a Rend and rolling another melee etc. etc
for about .285????

(I have a feeling I'm gonna end up re-rating all the dice...again)

[cliffwiggs]

[otto8064]

well, getting back to my original post....how would you do it then?

[KilljoyKing]

Oh, sure, you can "solve for X" and use "junior high algebra" if you want to do it easily, but if I wanted to think, I wouldn't have become a programmer, now would I have?

And for the record, the result I got actually was repeating-threes for as far a precision as floating types have; I just didn't report them since, well, three-thousands of a melee isn't very meaningful. Hundredths are a good cut-off point.

That whole (3 + x) is just the same thing my infinite parentheses did, just without the algebra explanation. I mean, given this board's past experiences with math, I thought brute was the way to go.
Speaking of which, if you like math so much, where were you for the legal combinations problem from earlier this year?
_________________
Dragon Dice Musing and Such @ Reality's End

[cliffwiggs]

I don't look at individual dice. When I make an army I look high level what do I want my strategy to be. What do I want to be good at, what do I accept being bad at. Do I want to be consistent or have larger but less frequent rolls. am I worried about my opponent playing a certain strategy.

Then I find the race/spells/sai's that support those goals. How many dice have faces which are 'dead' to me. What if I don't get my terrain I want. what if I get stuck in missile instead of magic or melee.

so start with one thing you want to be the best at, then pick another you want to be 2nd best at and finally pick something you want to be the worst at. That will usually define what dice you want to use.

There are also trends in the community. everyone used to play all common armies. then it was all monster, then all of a certain race. etc etc

the current trend I saw at gencon is a mishmash. a few units from this race, a few mages from that race, a dragon hunter or two. BOOM.

but if I look at a die and it doesn't support what I want to do, I don't include it. When I play all lavaelf I tend to throw in a few dwarfs just for the consistancy and non-terrain based saves and gold magic.

thats a whole lot of words that don't really mean much, except that all dice are useful in the right army. There is no magic formula that says this die is always worse than that die.
_________________
Multiplayer Champion Gencon11/Battlefest Champion(tied) GenCon10/World Champion GenCon07/National Champion Origins05/Intermediate Champion GenCon02/Novice Champion Origins99

[cliffwiggs]

[cliffwiggs]