From some testing, it seems like the new formula for spell cost has been found. It’s as follows: finalCost = floor(ceil(spellBaseMana * (1 - intPct / 100) + idSpellCostRaw) * (1 + idSpellCostPct / 100)); if (finalCost < 1) finalCost = 1; Which, broken down, means that it’s calculated by: 1. Taking the base mana cost of the spell and then multiplying it by the intelligence modifier 2. Adding the raw spell cost on top of it 3. Rounding that off to the highest nearest whole number 4. Applying the spell cost % multiplier 5. Rounding that down to the nearest whole number 6. Putting the number to 1 if the outcome would be lower than it. From this, we can gather a few things: spell cost % weighs more than intelligence, and raw spell cost can have increasingly more effect whenever spell cost % is also in play. I would like this to change. First off, intelligence and spell cost % working differently is simply confusing. I think that from a player and balancing standpoint, just make them equal would solve the confusion and allow for easier balancing, since you don’t need to worry about raw spell costs floating around; which is very common since the addition of mainly the Order of the Grook rewards and Anima-Infused Cuirass. That part of the formula would then become something like this: (spellBaseMana * (1 – intPct / 100) * (1 + idSpellCostPct / 100)) And as you can see, I left the raw spell costs out of it since I would also like to see a change there: in order to fit nearly all other raw ids, they should be calculated after percentages. It would make spell cost % equal to intelligence in power, and clear up quite some confusion. So the updated formula would be: ((spellBaseMana * (1 – intPct / 100) * (1 + idSpellCostPct / 100)) + idSpellCostRaw) and after that I’d give the intelligence modifier a ceiling (always round upwards), like how it has always had and floor the spell cost % modifier as it gets added from base so you'd have tiers, just like the current int does, to keep the both of them equal. So, the whole formula for spell costs in full: finalCost = (floor(ceil(spellBaseMana * (1 – intPct / 100)) * (1 + idSpellCostPct / 100)) + idSpellCostRaw) + repeatCost) To give an example (one that inspired me to make this suggestion, actually): This is Orange Lily. Spoiler As you can see, it has -3 3rd spell cost, and 200% 4th spell cost. This is Draoi Fair. Spoiler It has -1 1st spell cost and -4 4th spell cost. With the current formula, at 129 intelligence and using 2 Draoi Fairs (because you can, and you should), the spell costs would be 4 = floor(ceil(6 * (1 - 75 / 100) + -2) * (1 + 0 / 100)) 1 = floor(ceil(3 * (1 - 75 / 100) + 0) * (1 + 0 / 100)) 5 = floor(ceil(8 * (1 - 75 / 100) + -3) * (1 + 0 / 100)) 3 = floor(ceil(10 * (1 - 75/ 100) + -2) * (1 + 200/ 100)) Whereas my proposed formula gives: 0 = (floor(ceil(6 * (1 – 75 / 100)) * (1 + 0 / 100)) + -2) 1 = (floor(ceil(3 * (1 – 75 / 100)) * (1 + 0 / 100)) + 0) -1 = (floor(ceil(8 * (1 – 75 / 100)) * (1 + 0 / 100)) + -3) 7 = (floor(ceil(10 * (1 – 75 / 100)) * (1 + 200 / 100)) + -2) Which, well, when looking at the build as whole, seems much more obvious than the first result. tl;dr spell cost % is more effective than intelligence with negative raw spell costs and worse when having positive raw spell costs and affects raw spell cost for some reason, let’s clear that one up with a slightly changed formula that makes it much easier to understand and build around If you have anything to add (or correct), please do!