User:WinterRose/Racial Gains

Racial Gains is was an effort to gauge modifiers behind racial statistics - i.e. which race gains what stats upon leveling, and how they fare compared to human. This modifier was supposed to indicate a fixed relation between racial gains irregardless of class. The experiment has shown there is none, and read analysis below for an insight as to why.

All numbers, even more than those in pages on racial archery, damage or stealth modifiers are entirely unofficial, and can only be considered as very rough. Do not use them for any predictions.

Some numbers are even evidently false, and this has been noted. If you notice more faults, feel free to edit them.

Gains in general

As a vague definition, gains gained upon leveling can be described as a result of:

• racial modifier,
• class modifier,
• ability modifier (con and int for HP and MP),
• unknown factors (code specifics, can be seen in the last part),

that are (maybe) all applied onto a human set of gains.

These modifiers are separate and all factor into the actual gains. Unfortunately we cannot see these without running every race with an identical constitution and intelligence score and then deducing the numbers out of the results, something that is probably too much of an effort.

In the following tables these are fused into a single modifier, that compares the score to a human, and by definition this number cannot be the correct factor used in gain generation. However it is an usable approximation.

• How useful are these modifiers? Do they apply to 999 stats?

As has been shown, the modifiers cannot be reasonably applied on other classes, so unless one builds a table of modifiers per classes, they are likely not very useful.

Gain generation

• All runs were done with mages ran from level 2 to level 50. This is done since mage hp and mana modifiers are not as skewed as warrior or wizard modifiers, and present a usable middle ground.
• All characters were leveled with fully boosted stats with training and leveling gear.
• All characters are leveled as atheists. Any worship would bring additional factors which can hardly be correctly accounted for, especially in these short runs.

Errors

Please read all the disclaimers:

• Reasons for the wide inaccuracies of this data are threefold:
  1. Sample size in most cases is 1¹. Without at least 10 passes there cannot be any semblance of serious statistical analysis, and it is unlikely anyone will make 10 atheist trog mages. This is not actually relevant as purpose of this page are the rough modifiers, not projected stats. For far more accurate information on projected stats, use actual hero 999 or lord 1 data, and try to deduct worship gains.
  2. Rounding affects gains differently at hero than at lowmort where it doesn't occur, and also keep in mind that hero stats are based on boosted attributes, which are then halved. The modifiers should remain even through rounding, but might be a bit different at hero. Irregardless, see 1.
  3. Mana is also much more variable than hp, so it is likely some of the characters had a streak of bad or good gains and not enough time in the 48 levels to make up for the inaccuracies. Multiple runs would account for this but they were not performed. See 1.
• Where modifiers are not directly derived from data, some rounding had occurred. For example, Elf HP modifier is more likely -22%, and Half-Elf about -18%, but these were rounded to -20% and -15% respectively, etc.

(1) Sample size is not 48, since the modifiers are calculated by dividing total gains of two entire runs, which are singular numbers. If are not using specific gains for anything here.

Gain lists

All comments to the talk page please, and feel free to update the tables with your own data.

Legend
Divine
Legendary
Great
Good
Average
Below Average
Bad
Terrible
Ugh

Creatable Races Gains

Race	HP	HP/lvl	HP Modifier	MP	MP/lvl	MP Modifier
Centaur	539	10.6	+20%	473	7.55	+40%
Deep Gnome	460	8.95	0%	412	6.3	+15%?
Draconian	650	12.9	+45%	422	6.5	+20%
Drow	370	7	-20%	398	6	+10%
Duergar	516	10.1	+10%	343	4.85	-10%
Dwarf	499	9.8	+10%	343	4.85	-10%
Elf	366	7	-20%	408	6.2	+10%
Ent
Firedrake	N/A	N/A	N/A	N/A	N/A	N/A
Gargoyle	669	13.3	+50%	361	5.2	-5%
Giant	783	15.7	+75%	271	3.35	-40%
Gnome	459	8.95	0%	410	6.25	+15%?
Goblin	310	5.85	-35%	320	4.4	-20%
Halfling	366	7	-20%	411	6.25	+15%?
Half-Elf	381	7.3	-15%	395	5.95	+10%
Half-Orc	510	10	+10%	321	4.4	-20%
Harpy	460	8.95	0%	391	5.85	+5%
Human	458	8.9	0%	372	5.45	0%
Imp	N/A	N/A	N/A	N/A	N/A	N/A
Kobold	261	4.8	-45%	276	3.45	-40%
Kzinti	567	11.2	+25%	414	6.3	+15%?
Lizardmen	565	11.2	+25%	388	5.8	+5%
Ogres	700	13.95	+55%	331	4.6	-15%
Orc	564	11.1	+25%	329	4.5	-20%
Troglodyte	542	10.7	+20%	342	4.8	-10%

Some comments on above numbers:

• HP are generally a stable statistic and vary much less than mana.
• Giants probably have more mana than Kobolds in the long run. Must have been a poor set of gains.
• Half-Elves have less mana than Elves or Drow, generally, but here they have near-identical mana. Probably a good set of gains.
• Gnome, Deep Gnome, Kzinti and Halfling all received more mana than either Elves or Drow. Being suspicious, Elves and Drow were rerun for virtually the same gains. The elven races are expected to have more mana in the end, naturally, so these 4 races have suspicious results.
• Ent was not run. Run an Ent in your own free time. For the TNL, not for the gains.
• Evolution races make no sense in the above context.

Remort Races Gains

Race	HP	HP/lvl	HP Factor	MP	MP/lvl	MP Factor
Demonseed
Dragon	723	14.4	+60%	489	7.9	+45%
Golem
Griffon
High Elf	398	7.7	-15%	514	8.4	+55%
Hobgoblin
Minotaur
Miraar	N/A	N/A	N/A	N/A	N/A	N/A
Sprite	191	3.35	-60%	504	8.2	+50%
Troll
Tuataur

Some comments:

• HiE and Sprite have in fact identical mana gains (compared 999 wizards , the two of which had identical gains within 30 mana - about 7190), so in this scenario they have a same modifier, it's just unclear if it's 50% or 55%.
• HiE and Sprite at first have suspiciously high gains compared to regular Elves, but since we know the two race are identical, and both Elves and Drow were ran 2 times, this is probably correct.
• Base Miraar gains are irrelevant.

Testing mage modifiers on warriors

In an attempt to see how the mage hp modifiers fare when applied on warriors (applying mana modifiers is meaningless since wariors have no mana gains to speak of), here are results of a few warrior runs. Similarly, all were ran to lvl 50, all had fully boosted stats via leveling gear, and were atheists.

Remort Races Gains

Race	Warrior HP	War HP Factor	Mage HP	Mag HP factor	War HP gains Mag HP gains
Giant	1163	45%	783	75%	1.5
Ogre	1091	35%	700	55%	1.6
Lizardmen	896	10%	565	25%	1.6
Dwarf	851	5%	499	10%	1.75
Human	818	0%	458	0%	1.85
Elf	694	-15%	366	-20%	2
Goblin	639	-25%	310	-35%	2.2
Kobold	561	-40%	261	-45%	2.3

Some obvious questions:

1. Why is the difference in warrior gains so much smaller than in mage gains?

Warriors obviously get boosted stats with much less variance. Since they have same constitution as mages, constitution cannot be the governing reason. It's probably how things are coded, and cannot be explained via mathematics. Without such capping giants at 75% ratio could probably hit 18k HP at hero so it may be a sanity tool.

2. Why does the ratio of gains rise so fast for the low-HP races? Shouldn't warrior class gain modifier be static compared to mage class gain modifier?

A fantastic question, isn't it? If it fell sharply for the Giants and Ogres, it'd be easy to claim it as capping of sorts. It's not constitution meddling in either. Again, it is possibly a shadow of the code hovering over data for similar reasons as 1.

3. Since there is so much difference in the factors, is this whole experiment worthless? Monks and other classes will likely have their own factors...

Probably. It's an insight into relative strengths and weaknesses of races compared to each other, even though it failed to produce a universal rule of thumb according to which racial gains could be calculated.