Self found can use a bit of love

"

ScrotieMcB wrote:

"

Nykken wrote:

"

ScrotieMcB wrote:

Potentially useful math:

Spoiler

Let p equal the probability of achieving a particular affix combination or better with a single Chaos Orb (0<p<1)

Let P equal the probability of achieving a perfect affix combination with a single Chaos Orb (0<P<1)

ln(x) = natural logarithm of x

If the vendor offer for each affix combination is equal to ln(P)/p Chaos Orbs, then the average vendor offer after applying a Chaos Orb is equal to just under 1 Chaos Orb (approaching 1 as P approaches zero).

Now if course, I wouldn't suggest that "crafting" with your currency consistently provide a full vendor refund. But simply halving this amount (or whatever multiplier 0<k<1 one chooses) should be sufficient.

In this way, a smart vendor formula system could be implemented with confidence that it could not be abused.

How are "better" and "perfect" defined by the system?

The math holds for any number of arbitrary sub-groupings, providing "is this a member of the group?" is a given and doesn't reduce the probability.

For example, the math holds if you say there's only one perfect roll, and everything else is inferior. It also holds if you divide into three subgroups - say, physical weapons, elemental weapons, and "other" - but don't allow physical weapons to roll elemental affixes when determining probability values. Or four subgroups, or twenty, etc.

The number of subgroupings used is a design decision. I'd argue for many groupings, though, because that embraces build diversity.

I don't understand how any of that means the system can look at a dagger with 4 affixes and say "this is X% towards a perfect dagger" in a reasonable amount of time.

Are you saying there would be a database of every possible affix combination ranked by how good it would be for various builds and you would use a lookup table?

It's not hard to figure out the probability of getting a certain roll using Chaos Orbs. For example, getting a Brood Twine with a Chaos Orb is about a 1 in 236 trillion chance (assuming a single "lookup table").

It's also worth noting that what the math describes is the point where:
1. An item twice as scarce is worth twice as much (aka fair pricing)
2. The average rare vendors for 1 Chaos (this amount can be reduced by dividing all vendor offers by a constant)

In hindsight, the "halving" suggestion was regrettable. Getting half a Chaos on average for vendoring rares is crazy talk. Chaos formula on steroids.

Edit:

"

Nykken wrote:

Are you saying there would be a database of every possible affix combination ranked by how good it would be for various builds and you would use a lookup table?

If one were so inclined, it'd work with the math. But I don't think it's a good idea, because the system isn't limited to particular builds. Any number of independent lookup tables could be used, to include a separate table for every combination of affix categories. Ex: if this item has added fire damage, life leech, hybrid % physical and accuracy, and + Strength, use this table.

"

ScrotieMcB wrote:

It's not hard to figure out the probability of getting a certain roll using Chaos Orbs. For example, getting a Brood Twine with a Chaos Orb is about a 1 in 236 trillion chance (assuming a single "lookup table").

It's also worth noting that what the math describes is the point where:
1. An item twice as scarce is worth twice as much (aka fair pricing)
2. The average rare vendors for 1 Chaos (this amount can be reduced by dividing all vendor offers by a constant)

And a Brood Twine would be worth the same as a bow that rolled the same types of affixes but minimum level values for every one of those affixes unless you also establish a way to quantify value of the item, which is not feasible.

"

ScrotieMcB wrote:

Edit:

"

Nykken wrote:

Are you saying there would be a database of every possible affix combination ranked by how good it would be for various builds and you would use a lookup table?

If one were so inclined, it'd work with the math. But I don't think it's a good idea, because the system isn't limited to particular builds. Any number of independent lookup tables could be used, to include a separate table for every combination of affix categories. Ex: if this item has added fire damage, life leech, hybrid % physical and accuracy, and + Strength, use this table.

if this item has these 5 specific affixes? For a bow there are 220 possible combinations just for the types of prefixes (assuming exactly 3 prefixes), multiply again by the number of possible combinations of suffixes and you have over 5000 combinations. Some person has to go through every combination and say "oh that's the hybrid table, that's the physical table, that's the elemental table, that's the CoC table, that's the physical table, that's just garbage, that's the CoC table..." before we even get to the point where all the bows that fall into that category need to be evaluated for value to the player taking that build with the status of their passive tree and gear left as a complete mystery.

This is obviously a problem with no deterministic solution, impossible to reasonably implement in a computer program.

"

Nykken wrote:

"

ScrotieMcB wrote:

It's not hard to figure out the probability of getting a certain roll using Chaos Orbs. For example, getting a Brood Twine with a Chaos Orb is about a 1 in 236 trillion chance (assuming a single "lookup table").

It's also worth noting that what the math describes is the point where:
1. An item twice as scarce is worth twice as much (aka fair pricing)
2. The average rare vendors for 1 Chaos (this amount can be reduced by dividing all vendor offers by a constant)

And a Brood Twine would be worth the same as a bow that rolled the same types of affixes but minimum level values for every one of those affixes unless you also establish a way to quantify value of the item, which is not feasible.

If you're talking about Divine Orb type of stuff, that's feasible, too.

I already explained this earlier in this thread, but every item's trade value falls within the range of vendor offer (minimum) to perceived average crafting cost (maximum). The thing is, average crafting cost is a value which isn't a matter of opinion but one of fact and numbers. If, say, every item vendors for x% of average cost to craft that or better, that still gives a (100-x)% range for players to consider when collectively assigning market values.

However, I really do need to go back to the drawing board with my suggestion. Having two RANDOM rare items vendor for 1 Chaos is obviously too rewarding, but having an amazing item vendor for considerably less than 50% of its market value also doesn't achieve the objective for self-found. I'd have to find some way to keep x at about a 5% global average while raising x to about 50% at the high end. To do that, I'd need a different formula.

Edit: Actually, I have that backwards. I mean, very few players get a Mirror or a truly amazing item and think "it's too much of a bother to trade this to other players." The high value is a motivator. I need to focus on keeping the possible price range narrow on lower items, while allowing price range to get wider on big-ticket items.

"

ScrotieMcB wrote:

average crafting cost is a value which isn't a matter of opinion but one of fact and numbers

And the average crafting cost for items are much, much closer to each other than you seem to think.
A bow with any 6 random affixes that you get from an alch is about as rare as brood twine, it's just that no one gives a crap about a bow with the lowest roll increased physical damage, chaos resistance, and increased stun duration. There are probably a lot of affix rolls that are more rare than brood twine but also complete garbage for any build.

"

Nykken wrote:

And the average crafting cost for items are much, much closer to each other than you seem to think.
A bow with any 6 random affixes that you get from an alch is about as rare as brood twine

False. Extremely fucking false.

.

Just checking in to say that I've really been enjoying the new self found changes in the recent patch! Just kidding, there weren't any. I think we're now close enough to the next league starting that there's not much time left to actually make some changes, though.

The game kept getting better and better up until Atlas hit and took a gigantic dump all over self found play. Prophecy was a solid league for self found. Please, at the very least, make POE as good as it was two months ago.

Actually, also do the gem thing. Poor gem availability is the worst.

I think map diversity isnt an issue past a certain point if you run the maps...a lot. it evens out eventually. in the beginning, definitely an issue, but it gets much better. plus you can progress with 3 for 1 recipe. I havent bought any maps as I havent in 3.5 years and I was able to progress through most of the atlas.

other issues are fair game, especially unique maps. I dont know what they were thinking with unique maps. still waiting for my first unique map drop, looks like Ill get a mao kun through div cards faster than I will drop a unique map.

shapers orbs issues also mostly stem from those gated by unique maps. all other shaper orbs are relatively easily gotten through mere atlas grinding.

It seems POE is possibly getting a decent-sized influx of players from Diablo 3. Since that game's pretty heavy on the solo self found side, maybe it's time for a little outreach to players who enjoy that style of play. Possibly changes like those proposed in the OP?

Just saying. Could be nice.