There are multiple issues with the current speed formula in FFXIV and how the game handles other damage calculations. I will try to attempt to lay the foundation for rigorous testing and allowing other people to replicate my results.
First and foremost, it seems that all direct or instant damage and healing have only two values that are important: the absolute minimum and the absolute maximum. The damage spread is un-important (0.97 to 1.03) as testing using this method yields too much extraneous information, and is difficult to verify the 1.00 value. Testing using damage or healing over time methods gives us 11 unique base values for one single experiment (unless the value given back to us by the game is a double-digit) and allows us to see what value the game calculates at the 1.00 damage variance multiplier.
It is imperative to locate the 1.00 damage variance multiplier as it is much more accurate in assessing the game’s value to which we build all our models off of. Standard procedure for gathering data should be to note the 11 unique values of base damage given by the game, then re-verify the 1.00 value.
For example, I get these 11 unique values for regen with attributes of 105 WD, 408 MND, 379 CRIT, 218 DET, and 581 SPD:
These 11 unique values land on the 0.95, 0.96, 0.97, 0.98, 0.99, 1.00, 1.01, … , 1.05 multipliers. If you take the 1.00 value (488) and multiply it down the list of decimals (floor or truncate the value to an integer) you’ll come up with your values you saw in-game. This is key to knowing that you have the correct 1.00 value.
Do note that critical hit damage does NOT apply the decimal value multipliers. The game calculates your base damage and most likely populates a table of the damage variations, then calculates criticals off of that.
The current (apparently 100% accurate since it was data-mined…) critical hit rate and damage formulae are posted here at
but I have noticed that he does not take into account at what point the game floors the value. I have run multiple tests and with an un-floored formula it fails to predict in-game values I observed (in a later post). The floored formula is
= FLOOR( ( ( CRIT – 354 ) / ( 858 * 5 ) ), 0.001 ) + 1.45.
As it stands, I have no clue how speed functions within a formula except knowing that it is a multiplier of sorts. I have no idea how Dervy arrived at his current speed formula (the denominator value) and it doesn’t work with the data points I have currently observed in-game (in a later post).
The point I am trying to make is that if people are going to be making formulas off a body of data, at least have the data be as accurate as possible. Post your procedures so people can copy and replicate your data and re-confirm / re-verify your conclusions.