Skills and Stats: Some Assembly Required (Part 2)   Leave a comment

Welcome back to Gratuitous JRPG.  Last time we covered the wide world of skillsets, and what they cover.  Furthermore, we revealed the majority of the Sigil Crest attached template skillset involved in the game.  This post, however, we are going to take a break from this to finally start adding some numbers to the cast.  It’s something that has to be done sooner or later, and it goes into the current subject of finally implementing the player characters that is the ultimate goal of this multi-parter.

The implementation of statistical values in a JRPG is a more precise science than many failed amateur attempts at games would be.  And not only is not paying attention to this potentially murderous to anything resembling game balance, but it’s one of the most common problems.  Admittedly, however, it wasn’t the easiest to avoid back in the early days of RPG Maker.  It wasn’t until RMVX that we got particularly clear documentation on the system’s damage algorithms (RMXP had documentation, but it feels particularly unclear, and pre-XP RMs had zero documentation on how it calculated damage at all).  Naturally, with zero direction for how to get the numbers to do what you want them to do, your only option for getting solid balance was the hard one: Test and repeat until satisfactory.  And needless to say, this process could be described by some as “hell”.


And it is full of unclear algorithms and repeated testing against too-hard and too-easy enemies over and over.  And that’s before you learn that your stats will contribute only one sixteenth of themselves to a skill at most!

Needless to say, the clarity of algorithms is a large part of why I supported RPG Maker VX over XP and previous.  And it’s even easier with VX Ace–where they cut out the background algorithm (If you really wanted to change it in VX, you can reach into the code to swap it, but you only get two formulas to work with) and simply let the end-user determine damage calculations on an ability-by-ability basis, clear as day.  Why am I going over the differences between previous and current RPG Maker damage algorithms again, you ask?  Because with the added transparency for the end-user, we now have no excuse for not making a more statistically precise game.  We had it back in 2K where stats contributed amazingly little to performance.  But in VX and VX Ace, that problem and that excuse is gone, meaning that if we have a case of a generic enemy who is stronger than the final boss, or a boss who crumples in a basic physical from the mage, it is all our fault.

This is why careful consideration of what numbers we implement is so important.  The tools are there, and it’s made certain that we can use them with a notable degree of accuracy, so we should give them the respect they deserve and do so.  And as such, one would then ask where would it be the best to start?  This is perhaps the most important question to ask ourselves with regard to stats.  Not “how high do I want them to be at endgame when fighting the Great Elder God Flumfphlogloth,” but “how high do I want them to start?”  After all, before a game has an end, it has to have a beginning.  The answer to this, however, is simultaneously far more simple and complex than we’d like.

Technically, we have entire ranges of numbers that could work for our purposes.  Barring the damage formulas being extraordinarily specific and covering an identifiable and very narrow range, it will be notably difficult to estimate what is a “right” value in a vacuum.  In fact, the answer is going to frequently be that there simply is no one correct value, which will frustrate anyone looking for the one perfect answer.  This includes a very large number of people, myself included, who were hoping there would be a way to divine this set of values.  However, this does not mean that one cannot determine what are optimal statistical values in relation to one another for the purposes of your game.  This is difficult to explain, but as an example–if your Attack and Magic Attack formulas are similar to identical, you will likely want the offensive stats involved to generally be in proportion to one another for equivalent levels on characters.  However, sadly, number determination for a game is never going to be an exact science, and the best advice I am ultimately going to have is simply to eyeball the numbers.

Eyeballing the numbers, however, means trying to get a rough estimate of where they should be.  It is not, however, an excuse to pull off one of the bigger mistakes made by amateur game designers: making the numbers extraordinarily high for the sake of having big numbers show up everywhere.  The novelty of this likely has started with three games in particular: late-SNES/early-PSX Final Fantasy, Valkyrie Profile, and Disgaea–the lattermost of which still markets its huge numbers in its commercials, despite these values being irrelevant outside of postgame content.  Big numbers can wow some people, but ultimately will be harder to tighten a game around on for one, and can in fact be seen as a cheap trick to try and make something look more impressive than it is, like spinning rims on a cheaply-made car.  This is not to say that large numbers are bad, so much as that putting thought into one’s chosen numbers is well worth the effort.

Another two facets of selecting numbers for stats are much easier once one has the base value to work around down: spread and scaling.  Spread can be described as the differences in statistics between different characters at the same level, while scaling is the statistical change within a character over the course of the game.  With spread, the general idea is that you want to find a sort of happy medium based on what sort of game you are making.  The higher the statistics are, the wider the spread will need to be for the individual characters to be statistically distinct.  Too wide of a spread, however, can result in some characters being overly effective or ineffective in some capacity.  These tolerances may differ based on the game type one is making, however–an Action-RPG is more tolerant of narrow spreads due to the fact that one is always controlling one character.  A RPG with a notably large cast, or a Strategy RPG with heavy unit specialization is frequently more tolerant of wider spreads than normal, however.

Scaling can be handled in several ways, but can be summed up in a couple of ways–magnitude and pattern.  Magnitude is simply how much the character’s stats are expected to improve–assuming natural improvement through leveling, not equipment or statistic-boosting upgrades.  Pattern determines the distribution of the magnitude, which can have some interesting applications–and comes in three basic forms: linear, sublinear, and superlinear.  Most Final Fantasy games take a superlinear high-magnitude approach to statistic scaling (Final Fantasy 5 is a good example–start in double-digit HP, end in the 2000s), for example.  Similarly, one should not mistake linear growth with constant growth–which is a form of sublinear.  Linear in this context means that assuming a starting statistic is at value X, that character will gain X in that stat every Y levels.  Constant means that for a fixed value C, that character will gain C in that stat every level, regardless of the initial stats.  And notably enough, scaling interacts with spread in a few ways.  Generally speaking, for a given magnitude within reasonable bounds, a linear pattern will maintain spread, a sublinear pattern will narrow spread, and a superlinear pattern will expand spread.  Once the magnitude exceeds a given bound (this bound depends on the game’s equations and the original statistical spread), even linear patterns will start to functionally expand a spread.  Scaling also affects return difficulty–a high-magnitude scale can have enemies cease to be effective as soon as the end of the dungeon you use them in, while low-magnitude ones may have even the earlygame enemies a legitimate threat so much as halfway into the game or later, before factoring in equipment advancement, of course.  This is an important decision on how fast you want advancement to go, but it is important to keep in mind.

With these in mind, the question is easily “Where to start?” in regards to the current game.  Keeping in mind the general equations I have come up with prior:

  • Physical: POW*(PEN-ARM)
  • Magical: MNT*((FOC*C1)+C2-WIL)

the first reaction would be to keep the numbers as low as possible.  This isn’t a problem, I could easily run these with any number from the high singles on up as a starting point, but I don’t want the numbers getting too big.  However, I have difficulty deciding on a singular number until I look once again at the Sigil Crest statistical mods: they’re all in 5% increments.  With some quick testing in the RPG Maker VX Ace framework, I determine that it does not, in fact, round fractional statistical modifiers when applying separate percentile-based modifiers–simply truncating them for effect-related use.  In effect, the magic number here is the minimum possible value for these effects to be applicable as early as possible: 20.  This is as low as the current setup allows me to go.  While this is the bare minimum, there are indeed values I want to be higher than others–if I wish to include a Confuse or Charm effect, for example, or something equally insidious, I want to ensure that self-harm is indeed a threat rather than an annoyance.  As such, Focus and Penetration will have to be higher in the end–the deviation from base will have to be eyeballed in this situation, largely to factor for equipment–while not all armor provides ARM (in fact, it is largely the heavier armors that do this), all weapons provide a degree of PEN, and some items might increase one’s FOC.  Comparably, however, this allows ARM and WIL to have a minimum of 20.  Since MNT directly opposes WIL on status checks, it also needs to stay on the same scale.  Given that they’re similar in the equations, POW similarly has to stay on the same scale, though weapons may have a minor effect on it–a decision I’ll elaborate on a bit later.

Minimums for the non-HP/ST stats will tentatively be as follows:

  • Power: 20
  • Armor: 20
  • Penetration: 24
  • Mental: 20
  • Will: 20
  • Focus: 24

Now, with those, we can determine the minimum possible stats for the characters with a “1” in their slots.  Experimentation with formulas here points out that against a placeholder enemy defense value, single-point variance between ranks (20-25/24-29) at this point will be considerably negligible.  Looking at this, the initial instinct is to go with two-point variance between ranks (20, 22, 24, 26, 28, 30/24-34), but I am finding more appeal in proceeding with a 1.5 point variance set (20, 21, 23, 24, 26, 27/24, 25, 27, 28, 30, 31).  It’s an uneven progression, but furthermore allows some interesting calculations with scaling at level if I leave the decimal in for the initial math.

The question before full statistical mapping remains, however: what about HP and ST?  At first we can consider this to be an arbitrary value–especially ST, given that it factors into costs more than anything–until you remember that Wrack, a general spell described last post, targets ST specifically, giving the possibility for ST damage.  This means that ST still needs to remain proportional to HP, regardless of everything else.  Costs can be adjusted, but damage is going to be proportional in some manner to HP–which leaves HP remaining.  This at first would appear arbitrary–and in a game without anything resembling Confusion or Charm type control status, it in fact is, since enemy stats are based entirely on what effect you want them to get out of the player.  However, once the potential for self-damage via status or ability comes up, then it will need to be thought of in terms of roughly how much the player would be taking from that.  Eyeballing it (just an estimate), I’ll be taking the HP spread at 32-point differences, largely because I like non-round numbers..  ST will be half of those values.  Thus for HP and ST, we will have spreads of: (422, 454, 486, 518, 550, 582) and (211, 227, 243, 259, 275, 291) respectively.  With these numbers in place, we can finally get hard statistical starting values for each of our PCs.  But before I go ahead and list those, I would like to cover how growth goes for stats across the board.

In RPG Maker VX Ace, and in fact in every RPG Maker, growth is implemented the same three ways: either you decide on a letter ranking and get a random set, set values for levels 1 and 99 before setting a curve type, or manually set each value for every single level.  With scripts, you can also manage equation-based and constant-based statistical growths with little trouble.  If you are angling towards a linear growth setup, then you can easily use the second method with little trouble–the maker will easily handle the setup.  If you want a superlinear patter, however, it would be wise to use a script to your advantage instead–manually handling 8 sets of (designated max level) stats for each PC you make is not only needlessly impractical, but mind-numbing and torturous on par with being reminded to put covers on your TPS reports every so often.


Yeeeah, sorry, but I’ll have to have you come in on Saturday.  …Oh yeah, and you’ll need to come into work on Sunday, too.

With that in mind, in part because linear patterns are easier to think about and easier to handle without extra scripts, I intend to have a linear growth pattern.  And as for magnitude, I’m not exactly certain on how high-level characters will be at endgame, but I feel the base stats quadrupling at level 99 (estimating endgame to be around the 40-60 range) would be functional enough for this game.  It will inform my enemy design in the future should I choose this route and a fast or slow enough experience curve.  But with this, I finally have hard numbers for each of my characters, as seen below, introduced in order of being usable for the first time, with level 1 and level 99 values listed, tweaked to eliminate identical curves as much as possible:


  • HP: 518->2064
  • ST: 243->972
  • POW: 24->96
  • ARM: 24->96
  • PEN: 27->110
  • MNT: 21->84
  • WIL: 23->94
  • FOC: 30->120


  • HP: 550->2200
  • ST: 235->940
  • POW: 23->94
  • ARM: 27->108
  • PEN: 28->112
  • MNT: 20->80
  • WIL: 24->98
  • FOC: 20->80


  • HP: 534->2136
  • ST: 275->1100
  • POW: 22->90
  • ARM: 21->86
  • PEN: 27->108
  • MNT: 23->92
  • WIL: 20->80
  • FOC: 31->124


  • HP: 470->1880
  • ST: 291->1164
  • POW: 21->84
  • ARM: 21->84
  • PEN: 25->100
  • MNT: 27->108
  • WIL: 26->104
  • FOC: 27->108


  • HP: 422->1688
  • ST: 259->1036
  • POW: 23->92
  • ARM: 22->90
  • PEN: 30->120
  • MNT: 24->98
  • WIL: 23->92
  • FOC: 30->122


  • HP: 486->1944
  • ST: 251->1004
  • POW: 26->104
  • ARM: 23->94
  • PEN: 31->124
  • MNT: 21->86
  • WIL: 21->84
  • FOC: 25->102


  • HP: 582->2328
  • ST: 211->844
  • POW: 27->108
  • ARM: 24->98
  • PEN: 26->106
  • MNT: 26->104
  • WIL: 24->96
  • FOC: 25->100
  • Special: MFR 200


  • HP: 454->1816
  • ST: 227->908
  • POW: 21->86
  • ARM: 23->92
  • PEN: 30->122
  • MNT: 24->96
  • WIL: 27->108
  • FOC: 28->112

And with that entire set, I feel this post is complete.  Next time on Gratuitous JRPG, we will be filling out general equipment sets statistically, and discussing equipment stats more now that we have base statistics laid down more.  Until then, this is Epic Alphonse, signing off.


Posted July 9, 2013 by EpicAlphonse in Uncategorized

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: