Character Growth rates (and Data-mine Topic)

[triabolical]

If anyone is curious what happens if multiple stats are tied.

• HP is skipped if 3 or more stats are tied for closest.
• if HP is selected as the first stat, then there's some weird skipping of tied stats to select the second stat.
  • Skips the first stat in any 2 way ties, and any 3+ stat ties if Str and Mag are not both tied. 
  • Ignores Str and Mag if they are both tied in any 3 or more stat ties. 

 

[Dark Holy Elf]

Great work! It's nice to finally get closure on how this works.

I tried writing a short computer program to estimate what this means for characters' actual growth rates in practice. It's only an estimate because there are 10^18 different theoretically possible RN sets for levels (though they might not all be possible in reality due to computer RNs not being truly random) and I couldn't think of a good way to account for them all, but I used 10^9 of them to estimate the effects. It looks like normal characters (e.g. Ferdinand) gain 0-1% on their lower growths, and 1-2% on their higher. Lysithea, who has a lot of low growths and is more likely to get natural 0- or 1-stat levels, benefits slightly more: 1-2% on her lower stats and 2-3% on their higher. Constance's magic is the standout, being somewhere between 4% and 5% higher, i.e. about the only case where someone can reasonably expect to be 1 point higher than their listed averages by endgame. In all cases it's worth noting the effect is muted as you reclass to classes with better growths (which raise your totals and decrease your chance of 1- or 2-stat levels).

 

[triabolical]

4 hours ago, Dark Holy Elf said:

Great work! It's nice to finally get closure on how this works.

I tried writing a short computer program to estimate what this means for characters' actual growth rates in practice. It's only an estimate because there are 10^18 different theoretically possible RN sets for levels (though they might not all be possible in reality due to computer RNs not being truly random) and I couldn't think of a good way to account for them all, but I used 10^9 of them to estimate the effects. It looks like normal characters (e.g. Ferdinand) gain 0-1% on their lower growths, and 1-2% on their higher. Lysithea, who has a lot of low growths and is more likely to get natural 0- or 1-stat levels, benefits slightly more: 1-2% on her lower stats and 2-3% on their higher. Constance's magic is the standout, being somewhere between 4% and 5% higher, i.e. about the only case where someone can reasonably expect to be 1 point higher than their listed averages by endgame. In all cases it's worth noting the effect is muted as you reclass to classes with better growths (which raise your totals and decrease your chance of 1- or 2-stat levels).

 

Yeah I think this is something that you would need to simulate to get an effective growth rate. That’s what I would expect, low growth stats will have little to no change, and a 4-5% increase at the extreme side with Constance’s magic (this what I empirically saw, 64%-65%). 

Does not make much of a difference at the end. 

[asm335]

On 8/9/2022 at 12:39 PM, triabolical said:

I can now 100% confirm that the stats are picked by minimum difference. No rerolling of stats is used, just picks the stats that are closest to level up. 

I used an adjutant level up, which the RNs are fixed for a map, and continuously change the growth rates to get the RNs and play around with the growths to see what stat would appear on the level up. 

Weird things happens with ties. If all of them are tied, you get Str + Mag, skipping HP. Other than that, it goes by order.

 

 

Thank you so much. It was Lysithea's 2 stat levels that got me to really question this because it seemed like her two stat rolls were almost always Magic and Dex/Speed so my suspicious was that it was minimum difference that decided it but considering how low all of her other growths are compared to those 3, I couldn't rule out her just getting 2 of those before the 2-stat minimum took effect. Thanks again.