Do You Even Rank?

[Voom]

PvP time. I can't help but be competitive. I've played AT for more than two years so I think I understand its dynamics. The following will judge who has the best character based on combat stats alone (does not include small stones).

Abbreviations:
AP = Action points /Attack cost (round down)
AC = Attack chance
AD = Attack damage (average)
CM = Critical multiplier
ECC = Effective critical chance
CD = Critical damage = AD x (CM-1) x ECC
BC = Block chance
DR = Damage resistance
HP = Health points

Add up the points (round to 2 decimal places):
3 points per 1 AP
1 point per 20% AC
1 point per 4 AD
1 point per 5 CD
1 point per 12% BC
1 point per 1 DR
1 point per 20 HP

Score = Total points / Character level

I'll use my character as an example:
Voomster Lvl 18
AP = 10/5 = 2 => 2 x 3 = 6.00
AC = 153% / 20% = 7.65
AD = (17 + 24) / 2 = 20.5 => 20.5 / 4 = 5.13
CD = 20.5 x (2-1) x 22% = 4.51 => 4.51 / 5 = .90
BC = 72% => 72% / 12% = 6.00
DR = 2 => 2 / 1 = 2.00
HP = 47 => 47 / 20 = 2.35

Total points = 30.03
Score = 30.03 / 18 = 1.67

The higher the score, the better.

This will be fun so please calculate and share your rank!

(Edit: CM -> CM-1, 3 CD -> 5 CD, and calculations)

[rijackson741]

I think there are a few problems trying to compare characters using your calculation.

One is that your scoring is kind of arbitrary. I do not think an additional 20 HP is worth the same as an additional 20% AC. Given the option, I'd take the extra HP every time. CD must also be worth less than AD, because it only affects some monsters, but you have it worth more.

The second is that I think it's biased towards lower level characters (quite heavily, in fact). The denominator (the bit on the bottom) in your final calculation of Score doubles when your level doubles (obviously). To make the final score unbiased the numerator (i.e. the bit on the top) would also have to also have to double when your level doubles (for some kind of average build). But once you get to a certain point AP cannot increase, CM cannot increase, and ECC is extremely hard to increase. Even just looking at the other stats, if you went from level 20 to level 40, I doubt you could double all of AC, AD, BC, DR, and HP.

[Voom]

In other words, this a failed attempt at comparing stats with a single number. I see what you mean with higher levels being unable to double their stats fast enough. After all, you can't compare light-weight boxers with heavy-weight boxers. Well, I have no idea on how to scale the score with the increase in levels.

Arbitrary, a little. However, I used level up amounts and rarity. If you are allowed +5HP in a level up you are also allowed +5%AC, +3%BC, +1AD. So I equated them. As for the CD, I agree with what you mentioned.

[Voom]

If I was given the choice, I would take the 20%AC. I never found that I needed much more health in pretty much all situations.

[Voom]

Maybe I could apply some sort of exponential expression like x^2 or slower x^1/2, but the only way to figure out what kind. I would need a database of a very large number of character stats from many different builds at all levels to develop an "average" to base calculations. The one that is currently collected based on signature is not big and various enough. There is a similar exponential formula, that Oskar probably made, for combat mechanics concerning AC-DC calculations. Google code even shows its graph.

Anyways, if there is a better way to PvP, please share.

[rijackson741]

I guess that fact that you would pick AC and I would pick HP shows how subjective this is . But that's one of the good things about AT. You can't just reduce the best build to one number. To tell the truth, my main character doesn't have a lot of use for either in the current version of the game. With 246% AC, more AC doesn't get me a much higher chance of actually hitting anything, and nothing (not even a mob) can survive long enough to knock me down by 196 HP

As an aside, I think the multiplier for CD should be (CM-1), not CM, because that represents the extra damage you get from a critical hit.

Yes, I agree that, in theory, it would be possible to correct for level given a database of character stats and enough work. There is one here, in the old high score list: http://andorstrail.com/viewtopic.php?f=2&t=327. You could paste it into Excel, after which it will require a little manual fixing (I did this in the past, for other reasons, but unfortunately I seem to have deleted the spreadsheet). You could do a regression of level vs each stat, which could give a basis for a correction. It would be a lot of work though, and would not resolve the question of the relative merits of different stats.

Here's a spreadsheet with the graph you mentioned, along with a few other things: https://www.dropbox.com/s/2jgfka6u91v05 ... .xlsx?dl=0

[Voom]

I agree, it is a good thing to have such build flexibility and fluidity. And yeah, at a certain point adding AC will have less of effect (again, that's an exponential graph, or rather a transposed exponential growth or an inverted exponential decay graph). So I get the idea but idk if I can do it. I want to try. As for HP though, I think additional HP loses its effect some amount sooner. The thing is, IMO, the opponent either hits me or its dead. It has to be over with or I'm not serious about it. That's how I feel. You don't play when you're playing AT.

Ok, I see your point for CM. It makes sense.

I will check out the spreadsheet and mess around with the numbers. Thank you. Have a good New Year's!

[Zukero]

To be nitpicking, it is an arctangent. But I think what you meant was a logarithmic scale

[Voom]

To be nitpicking, it is an arctangent. But I think what you meant was a logarithmic scale

Honestly, idk. Arctangent looks like the combat calculation graph. Log scale is profoundly narrow, just imagining the the average score vs the average level as levels increase seems like a wider much slower equation. Idk if it will be possible to do it but maybe excel would help.

[rijackson741]

Here's that spreadsheet with the high score list: https://www.dropbox.com/s/dxaa1kwmlpq1k ... .xlsx?dl=0
The first tab has the raw list. The highlighted cells indicate a problem for one reason or another. The second tab has a pruned list with the problems removed, along with any entries for which there are no stats (i.e. nothing in the column HP, or the ones to the right of that).

It's not a great database, but it's the only one there is.