Poe Price Check – Check Single Item and Stash Tab Items with Ease

@Fnts7

Thanks a lot for your generous donation! And your feedback is well received.

For unique items, most of the time the price is the median given the outliers are removed (very high or very low). This can definitely be improved.

For low end uniques (less than an alch), my current thinking is that I should probably use a large sample size instead of default 100. Depending on the roll, some will not even return 100 matches based on the upper/lower threshold. If more than 100 is requested, more response time is needed. Any thought is welcome.

For perfect or (perfect-1) rolls, we are thinking of dealing with them separately. Please see the following post. This function should be online soon.

https://www.patreon.com/posts/dealing-with-max-9914743

You are also welcome to view our other effort trying to improve rare predictions here.

[Removed by Support]

Thanks again for your support!
Last edited by Stacey_GGG on Jul 24, 2018, 9:36:40 PM
I'm glad about your great effort in that tool. However I need to complain more. Currently in standard it works even worse than my last post during which time I mostly played Legacy.

Main problems:

1) Junk uniques are like 80-90% cases overpriced.

2) However it also happens for good uniques that they are underpriced. E.g. I recently sold a middle-roll voidwalker murder boots with a good enchantment for 10 chaos in few hours after posting. The price suggestion for that item was 1 chaos.

3) The tool doesn't even follow the orbs prices in the league. E.g. in standard currently alchemy to chaos orbs is 3 to 1 according to market prices. However uniques priced as 1 alch are displayed 0.48 chaos, while they should 0.33 obviously.

4) Very often I see that in normal search there are lots of similar items (or uniques with almost the same rolls), while in price check it is noted in red that there were too few items to compare, so the price is unreliable.

My random suggestions:

Introduce a setting to the algorithm - can be fluid or maybe 3 state - sell fast, normal waiting, long waiting.

For stuff that is hard to compare to anything. For example there is a rare with 5 affixes, from which similar items have only 4 of the affixes and the 5 one is very good (the one that doesn't take part in comparison). The result is underprice. For such cases you could open a database of fixed prices. Perhaps it could be open for anyone for voting. For example - single suffix of chaos resistance S1 could add up 10 chaos to the price of most items. That is naturally a heuristic, but seeing how it works now, it should be rather for plus.

When it comes to uniques I think you shouldn't just treat perfect rolls specially. It should be always compared to prices of the same uniques based on rolls. For example some unique item may have an affix of very low significance, so it doesn't influence the price much. Other affix might be so important that its perfect or almost perfect roll makes the price like + 1ex, while middle rolled item price is like 1c. The algorithm should take people's lazyness into account also unfortunatelly. So if many sellers just don't bother and put a high rolled item with the same price as middle rolled, than the suggested price for such high rolled unique should also be lower.

I will write more ideas when I have new ones.
Could you please update the system?
The new jewel mods aren't recognized at all and just get completely excluded, the ES regen mods on chests don't work, etc.
"
Could you please update the system?
The new jewel mods aren't recognized at all and just get completely excluded, the ES regen mods on chests don't work, etc.


Mods have been updated. If there are errors, please let me know. Thanks!
@Fnts7, very good suggestions! We are not sure whether we can address all of them at the same time. While we are improving the uniques by adding max/max-1 mods search and also providing mechanism to collect feedback about 'non-sense' max mods, wrong numbers, etc. So with some help from the community, we can improve the address some laziness issues :-) For rares, we are in the process of rolling out predictions by machine learning models as well as the current models. We also want to provide feedback mechanism to let community help us to improve. We will learn through the process over time. At the same time, feel free to write down any of your valuable thought, which is helpful.
Hello again, I have an issue with this item:
Spoiler
Rarity: Rare
Brood Crown
Praetor Crown
--------
Armour: 288 (augmented)
Energy Shield: 109 (augmented)
--------
Requirements:
Level: 68
Str: 42 (augmented)
Int: 61 (augmented)
--------
Sockets: B-R-R
--------
Item Level: 81
--------
+48 to Armour
38% increased Armour and Energy Shield
+55 to maximum Life
+31 to maximum Energy Shield
20% increased Rarity of Items found
+44% to Cold Resistance
32% reduced Attribute Requirements


This item doesn't give me the tiers, this issue's been there for quite some time though.
(I will vendor it by the way but it's an issue with more rare items)

Also Unset Rings don't get their socket values (minimum and maximum) automatically when searched so if I check the box but don't add 2x a 1 for minimum and maximum it won't work.
Can the 1's be added automatically?
Last edited by World_Destructor on Aug 16, 2017, 10:43:37 AM
@World_Destructor

Hey, thanks again for finding and letting me know the bugs.

(1) This item may have multiple new hybrid mods that my code does not check. For now, I made it to still report the prefix or suffix as much as it can, but the number of prefix or suffix may not add up to 3. This will be fixed after I add the code to check the new hybrid mods. May take some time though.

(2) For unset ring, now you can check the number of socket and search for matches. While I didn't make it by default, since some similar matches can worth similar but not having a socket. You can check this criterion manually. Does this work?
"
SawnMendes_2 wrote:
Thanks for letting me know your thread id. I guess the first spoiler was created by Acquisition


I don' t think I get this...
About point 2, for accurately pricing you can't neglect the value of the additional socket if you won't neglect implicits on other rings (think of resists for example, a 120% total resistance Ruby Ring with 50 life isn't even nearly as hard to find as an Unset Ring with the same values making it MUCH more valuable).
If you disregard the value of the implicit you have to do it to other implicits as well, basically.
But that's also not always accurate because of meta builds.
For example, a well rolled Steel Ring and an Unset Ring are similar with their explicits, not only is the white drop of a Steel Ring more valuable, also that they're similar on their explicits doesn't change that the implicit stat is valued much more in combination with the well rolled explicits because it fits the meta better making it much more desirable than an Unset Ring with the same explicits.
If 2 different base rings have the same rolls and stats as explicits then the implicit can variably influence the value of each ring because of the meta.

If you get what I mean, lol.
So I think disregarding all implicits maybe should be optional.
Last edited by World_Destructor on Aug 17, 2017, 10:27:37 AM
Yeah, I think I got you:-) For a really good steel ring, maybe other rings cannot reach the total physical damage it can get. The search is always based on total mods values (e.g. if there are two same mods, one in implicit and one in explicit). Hence, the search may work well.

I am still wondering about whether should I make an exception just for unset ring (by default check the socket attribute). If I make this exception, the unset ring price match will only return unset rings by default. Is this situation better?

Report Forum Post

Report Account:

Report Type

Additional Info