Seperating Defense from HP

[sutarpratinum]

Could you label the axis data please?

[NoBanana]

Graphs are all titled Y against X.
For example, “Percentage Increase per Point of Agility against Agility” is the percentage increase from each point of agility on the Y axis, and the agility value on the X axis.
Blue points are actual measured data, orange points are predicted points based on a formula (which is usually written on the graph).

Couple of them use slightly outdated data because I couldn’t be asked to open my spreadsheet.

[sutarpratinum]

Yeah I gotcha, it just tears at me when everything isnt fully labelled when it comes to graphs

[NoBanana]

real scientists don’t need to label their axis

[sutarpratinum]

Where I’m from we introduce with x and y first but then again I’ve stopped pursuing math

[Level]

Going by current scaling again, our agility if someone was complete glassy agility(Highest scaling stat out of all % based stats)

We’d have 1,825 agility so the last formula you placed could be accurate.

Oh wait I forgot to specify at level 1000, I’m dumb

[anon94253326]

its been multiple minutes

[VDruid0]

It will be worth it. Please wait a little longer

[NoBanana]

Yeah, though we’re not going to stay at our current scaling for the next ~4 or 5 tiers of items.
5,000 is perhaps an overestimate, but I wouldn’t be surprised if we surpassed it.
Also remember that this formula was written for v1.1 - vetex was planning for a level cap of 5,000 there.

Future predictions aside, I’m pretty confident the power formula isn’t accurate - it doesn’t match the first handful of data points. There’s admittedly a lot of room for error there, but I also have a lot of data for those first few points so I’m fairly confident on those values.
To get a power formula to fit the first dozen points, it needs a way higher index around 0.6 (also note this still doesn’t reach the first datapoint):

But if you do that, it doesn’t come close to fitting the rest of the data.

[Level]

What if it had conditional diminishing returns, aka only causing diminishing returns at a certain point of time.

I’m not good at math so that might not be the name but dnc.

[VDruid0]

Reject the meta, Embrace the club

[image]

Here

[NoBanana]

If the formula is piecewise then we have literally no way of knowing how it works until we hit one of the arbitrary boundaries.

Though if the formula is piecewise I’d expect it to be better right now - vetex would have to be actively trying to make the percentage stats garbage if the level 90 stats were this bad in a piecewise formula.
Also, the only reason for the curve on the first dozen points to be this sharp is because it’s trying to cover several thousand points. We wouldn’t need something so strange if it only had to cover the first couple hundred.

[Level]

What if it’s just straight up 23*log(x)

Idk I might be under-thinking it

[NoBanana]

you’re underthinking it

[Level]

45×log(x)-50?
55×log(x)-70
64×log(x)-90
100×log(x)-170
eventually I’ll get it right if I keep spamming answers

[NoBanana]

while it would work, i don’t think infinite monkey theorem is the best approach here

[bikbok]

ayoooo this would be so interesting. maybe we’d see more glass cannon people with this implemented.

[Inferno]

not necessarily a good thing considering we need more people using other stats like magic speed…

[VDruid0]

Isn’t this a nerf?
therefore a incentive to use other stats?

[Inferno]

yeah pmuch a nerf