D2 Say no more to blurry ring appraisal requests! I present a machine learning based ring appraiser

[rayliv]

Hi guys. After the initial release of alpha version yesterday in a discord channel. Today I am releasing the beta version of this noncommercial website (no ads no anything but the tool, already tested by several other websites). The basic idea is to go through hundreds of recent transactions on various sources, dump the data into a machine learning model and voila! So say no more to blurry-picture-for-ring-appraisal requests!

As a reference, I put some relative scores of runes as of Jan14-ish. It is just to give you a rough idea on how much a ring is worth. The actual price is determined by both seller and buyer of individual pieces. It's not perfect either - still in need of more data, especially in the department of dual leech rings, pure mf rings, etc.

Feel free to take a look and let me know what you guys think!

d2ring.com

 

[drg-123]

Hey, that’s cool. Nice tool.

It said the 10 fcr/17str/71 mana/9MF ring I found (and am keeping) the other day was worth more than an ohm.

 

[Nal]

neat. had a bit of fun with a pipe dream

 

[rayliv]

neat. had a bit of fun with a pipe dream

Thanks! You can put as many mods as you want on it, which is kinda careless of me. But the tool "calculated fields form" is quite limited in functions. May need to find workarounds.

 

[deBuren]

do rings cost more for "perfect" rolls and does the calculator do this? like a 30 res is multiplied compared to going from 28 to 29

 

[rayliv]

do rings cost more for "perfect" rolls and does the calculator do this? like a 30 res is multiplied compared to going from 28 to 29

Good question. 30 is definitely better than 29 but not in a way that 30 would double the value in the current model. With more data maybe I will be able to do that but I'm just not seeing a lot of perfectly rolled rings yet. I recently stumped opon a ring with mods like 10fcr/57life/20str/6mana/3ll selling for about 15 Bers, while my tool only values it at 72.8, a bit above a Sur. It's certainly related - insanely high life and max strength. With more entries like this, I think the model will learn it.