We've been using a Zero-Sum DKP system, but decided to re-work our DKP because of (self-inflicted) problems we've encountered (i.e. not reaching a linear loot distribution). Our problem was mainly giving out incentive DKP on loot-free runs (new bosses) and the resulting adjustments.
Now the main goal of a DKP system ist to achieve a linear (i.e. fair) distribution of loot. So I thought instead of giving out DKP for items that dropped as well as incentive DKP that need to be adjusted why not do it differently: You keep track of every person's raid attendance as well as the number of items he/she received. Then you divide the number of raids attended by the number of items received and whoever has the highest quotient is "DKP leader". You can of course weigh the attendance points based on the duration of the raid as well as weigh the items differently, depending on how good they are (i.e. you don't divide by the actual number of items, but value items differently, e.g T3 head ist worth 1 point, T3 bracers/girdle only half a point).


Does anyone use a similar system? Tell me your thoughts!


Edit: Short summary for those to lazy to read it all ;):

- You gain attendance points by attending a raid (e.g. 1 point for weekday ~4 hour raid)
- Items are weighted considering how good they are (e.g. 0.5 points for T3 bracer, 1 point for T3 head)
- You divide raid attendance points by item points; highest quotient has loot priority (less items received per raids attendend)
- The average quotient is actually the gradient of the ideal linear loot distribution (take a look at http://padwen.tripod.com/)

Seems time intensive to me. And, if I understand it, lots of problems for objectivity. What's the point of the DKP leader? Does this only change once a week then? So presumably, this person could show up and loot 5 items in one night due to DKP leader not being upgraded until next week or something.

I like the attendance thing, and working it into DKP, but as a sole factor? I don't think so. You would have to do a lot of work to price items, and raid lenghts/difficulties and whatnot.

That being said, it could be interesting if fleshed out.

Well its not more time intensive than a regular zero-sum DKP... just instead of giving people DKP you give them attendance points and have the database calculate the quotient. Asl o if someone loots an item, their quotient will obviously change on that run, as you factor in that item.

And isn't the whole pupose of DKP to get a linear loot dristribution? That quotient just shows exactly that if someone has an above average quotient, then he has received more than his "fair" share, if someone has a below average quotient he has received less than his "fair" share.

Take a look at this comparison between zero-sum DKP and rolling: http://padwen.tripod.com/ Ideally, everyone would be on a linear graph. The gradient of that ideal graph is exactly the average quotient that I want to use as a DKP replacement. The ideal quotient of course changes all the time, depending on how many items are picked on the average raid.

The biggest problem I'd see with this kind of system is that as attendance goes up, it becomes harder and harder to move your quotient. Your veteran members' quotients won't move around much even after a month of raiding.

However, the new guy will get an insane amount of spike changes because his denominator is really low - any change in either attendance or items looted causes a big swing. So the new guy doesn't loot anything for one run and suddenly he's the dkp leader, whereas the veteran has to not loot anything for 2 months to achieve the same effect.

The classic zero-sum system has one often neglected, but very cruical property - it has no 'memory'. It treats new people and veterans equally - you have exactly the same ability to move your score up or down regardless of how long you've raided, or how many items you've looted.


edit; missed a couple of words.

This is a flawed system because it assumes that every item is of equal value, which is obviously not the case. People will start holding out for really really good items, and you are going to have nexus crystals instead of a better raid group.

Linear loot distribution, in my opinion, is best achieved by making all items much more expensive. If a player can buy an item and stay at the top of the standings, there's a definite problem. If this was happening (especially in zero-sum), then your dkp system was seriously flawed. (This is ignoring high vs. low attendance players which is a definite separate issue.)



What my guild does (and it works pretty well): A non-zero system dkp system (separated per instance), where non-resist items are worth upwards of 15 boss kills worth of dkp. Then on top of this, your (dkp earned - dkp spent) is multiplied by your 60-day attendance percentage when looking at highest dkp in regards to winning an item. It definitely rewards people who are consistent raiders but at the same time keeps items going to different people (since items are so expensive and people easily go negative lots of the time)

Anyway, YMMV. Just my input

That's not a problem at all, as you can weigh the items, as I said before. I.e. T3 head is worth 1 point, while T3 bracers are only worth half a point.

Well, for the first few weeks new members usually only get items in our guild when every old member has passed. But I don't know if new members would have much of an advantage. Assuming you get about 1 point worth of items every 10 raids (e.g. T3 bracers and girdle à 0.5 points or a T3 head à 1 point) a new members would have to go on 10 raids to reach the average quotient. People who receive items should most of the time have a higher raids attended/loot received quotient, so the new member will have to come on quite a few raids before being able to loot anything.

Might I suggest a constrained bid system.

Basically, you award points for whatever -- time spent raiding is a good baseline, bonuses for new bosses, or whatever.

Then, when accepting bids, you allow a finite set of bids from any individual player -- we use "half your current points" and "ten points"; no other bids are allowed. It ends up with some elements like your system -- effectively, you end up with "raids attended since last full bid" as the most weighted attribute in the system.

Then you have two (or more) point stacks -- for example, durability armor pieces in one point stack, everything else in another -- with the goal of keeping some stacks at a relatively high turnover with more valuable items in another.

End result is a dynamic valuation system (which I'm partial to) with much less ability to game the system a la bid DKP (which I'm not partial to). We've been using it for over a year and have been very happy with it, although we're far from an uber guild.

Thanks for suggesting it, but we're not big fans of a bid system due to the possibility of collusion. E.g. warlocks agreeing on who gets the set piece to minimize the points they have to spend and put them ahead of the mages (although this problem isn't that much of a problem anymore with shared set drops).
We have considered a bid system, but it's just not suited for our guild. We're pretty much set on wanting the system to be attendance based, but would like some input on any flaws that we've missed or experiences other have had with a similar system.

Re-read the post (or, the Wikipedia page one of the people in my guild wrote about our loot system -- http://en.wikipedia.org/wiki/Shroud_...stem_%28SLS%29 -- I should point out I had nothing to do with it being set up on Wikipedia...). A constrained bid system combined with multiple stacks for gear eliminates much of the potential collusion.

As I've said, we've done this for over a year, and valuable items go for a half bid (which is the only possible 'valuable' bid), and not-so-valuable items go for ten points.

The system combines a number of features we wanted: Attendance-based point awards (with extra point awards possible without affecting the integrity of the system), anti-collusion mechanics, dynamic valuation, relative inability to game the system to minimize loot cost for valuable items, and anti-inflationary mechanics.

Thanks again for the link, looks good. But we're gonna be using the system as a basis for an officer decision system, so bidding isn't a possibility. It's supposed to be a guideline that shows the people that are reasonably close to each other and should be considered when allocating an item (the player who gets it then should be the one in whose hands that item will bring the largest benefit to the raid).

Easar, i'm a big fan of the variety of loot system you describe. My ideal system would work like this:
Each item has a fixed value, worked out by however. When you get an item the value gets added to your "loot points".
Players can earn "work points" by attendance, boss kills, providing items, anything you want. (My preference is not for boss kills though, just pure attendance).
Then in order to make it zero sum,
1) Get the total loot points L, summing over all players in the system for the entire history
2) Ditto for total work points W.
3) Find the constant C such that LC = W / C. C = sqrt(W / L)
4) Then for each player, divide their individual work points by C and multiply their loot points by L.

Subtract loot points from work points, and you have their final score for priority purposes.

As a prevention to hoarding, i'd make the system nondeterministic, by combining each player's final score with a dice roll when they go for an item. Balancing the power of the roll and the player's score is difficult but this addition is pretty important.

BTW, what exactly do you mean by linear? That your loot should be proportional to time spent?

Actually, Easar, that's exactly what I've been doing the last while. We're officer awarded loot as well, so we use it as a guide. Attendance is recorded by zone every half-hour using the EK mods. I wrote an excel sheet so I just have to copy and paste in the attendance records and the zone # to get the totals, when I have to add new member records, etc... And then a loot awarded column for each person/raid to which I assign an arbitrary "loot value" to the loot they got (up to 2 for a highly desirable, rare item, 0 for a sidegrade, resist gear, would-otherwise-be-disenchanted item). It gives a good heads-up when deciding between people. We haven't been using it for long, but I'm glad this came up, as it brought up the problem of converging quotients after a long time. I'll include a limit to that, either in time, raids or time attended.

As for recruits getting the ups and downs early on, I find that doesn't really become a problem, as new recruits usually have tons of gear to get from earlier zones which are still an upgrade, while we favour long-standing members with high attendance %s for the rare items. But we can still use it for which recruit should be getting that gear.

You need more constraints in your system since someone with no items picked up and one raid attended has a problematic score....

1/0 = ..........

My guilds system right now is zero sum upgrade dkp with incentive given on progress nights and 80% dkp for standby, periodic rezeroing based on 'active' members only but applying to inactives too.

What i would like to do to it is attach 30 day rolling attendance to it in some way. You can pick any time period or even multiple ones i just happen to like 30 days.

Automating the rezeroing via scripts would be nice and reduce the one off new person at zero joining on the 31st may start at -20 if we enter them into system immediately or they make a small amount of raids in one period. While a second new person joining just after rezeroing avoids the rezero calc. Other options are shorter rezero periods and one off protection, or also getting slightly complex dividing the rezero applied by the proportion of the time period new member was in the ststem.

Anyway back to the rolling attendance. To accomplish two things; disincentive hoarding of current dkp and incentivising attendance i want to have current dkp adjusted by recent attendance. To quote a song "What have you done for me lately?". The time period needs to be both long enough to even out unavoidable situations such as power dying we need to give time for the foolish casual to go out and buy a diesel generator and allow someone with 90% lifetime attendance who was in hospital for two months to work their way back into the swing of things in a reasonable timeframe.

From that point there are three paths to go down:

Relative Percentile Brackets. If your in the top x% of peoples attendance you get priority. Can use multiple brackets. Simple with standby raids. Potentially divisive if 30% have perfect attendance and 5% have almost perfect but your system says top 25% attendance last 4 weeks get priority over everyone else then perhaps the 5% or 10% if your systems bad are getting treated unfairly. If spread is large it works fine though.

Flat Percentage Brackets. If your below x% attendance you dont get priority. Can use multiple brackets. Needs exclusion of standby raids (we double list raids giving people inside and outside differant dkp amounts). Another way to do this with relative adjustments is take the highest attendance persons 30 day attendance percentage, say its 80%. Then anyone above 75% of the top persons attendance gets priority.

Dynamic Current Points Adjustment. Rather than having a cutoff everyone current dkp is modded by recent attendance. If you have 100% attendance you get 100% of your points. 0% attendance last 30 days and you are treated as having 0 current dkp, unless you were negative in which case my preference would be to leave current points unchanged. 50% attendance would have 50% of their real points as current points. Alternatively you could do something like halve someones negative current points who has 100% recent attendance but leave someone with 0% recent attendance at their real value but i see no real need.

Also anywhere i said priority in the first two options could instead be replaced with gets fixed adjustment of current dkp based on failing attendance levels so rather than people with low recent attendance always being after those with high recent attendance they could automatically lose say half their current dkp temporarily.

Just reread some of your stuff and since the system is more a tracker of who the officers are giving too much loot for perhaps you can just ignore the 1/0 thing.

And Pater makes a valid point to about skewing more loot to higher attendance people, though just skewing better loot can work as well.

Well, we want the active members to get more loot as well, and since we're only using that system as a guideline, we've got no problem with point hoarding. A low attendance member won't be getting that rare item that everybody wants just by picking up nothing for several weeks. Instead, he'll be getting the items that the "core team" doesn't need very much anymore (i.e T3 Bracers, Belt, Boots and that kinda stuff)

That's exactly what I mean. The average attendance/loot received constant is actually the gradient of that function. Take a look at this page again http://padwen.tripod.com/

But I agree that high attendance members need some kind of bonus, especially when you're not just using the system as a guideline for an officer decision system. You could solve that by having a new system for every new instance and having old, high attendance members start out with a bonus to attendance. That way you also avoid the problem with a person's quotient being pretty much locked in place after a large number of raids (if you've been to 100 raids and picked 10 items picking one more item/attending one more raid won't make as much difference to your quotient as it would for the 20 raids & 2 items guy).
Also, by having everyone start out as if they had looted one item thats worth one point, you avoid the division by zero problem and also give out a bonus to the high attendance members, as that one item will make less of a difference wehn you got 90% raid attendance than wehcn you got just 30%.

We're trying to keep this as simple as possible as it's pretty easy to mess up with a zero-sum DKP because of all the adjustments and fixes you need to make due to loot-free runs etc. And it doesn't get much simpler than this, if your goal is a linear loot distribution.

Have you tried the giving out less than full DKP spent on farm runs and banking it for progression runs yet? Thats one way to reduce adjustments if you want to get rid of them yet stay close to your current system.

Yeah its simple in terms of tracking and so on. Though it looks like you still want to give differant weights to differant items which is always complicated to sort out.

The hardest parts of loot council i found to deal with was balancing constant and irregular attendance of veteran and newcomer members with a desire for loot to be somewhat normalised per class. Your system has a lot more memory though so has a much better shot.

As far as the logistics of your proposed system goes (and just to complicate it since i enjoy complex loot systems :) what i would suggest is declaring each day you raid in something important to you a 'raid day' this could be everything MC and up or whatever you prefer. If you do a 8 hour weekend raid and 4 hours every other day feel free to count that as 2 raid days.

Two options:

1) Enter an attendance and loot list if you like whenever you kill something. Then sum however many entries you have per day and attendance credit is the fraction of events attended on that day.

2) Have time periods designated and track attendance at those points. Be it once per hour or start and end of raid. Credit is fraction attended again.

Insert extra events as desired to increase weight of progress runs over farm runs or just weight the entire progress raid. I prefer a method like that over simple boss kills as although it incentivizes fast clearing there is no real penalty for not showing up when nothing dies. If what you want to do is basically track loots/kill equilize for that and then bias seperately against people who dont help get new content on farm thats viable too although i dislike it unless it is tracked since everyone tends to overstate ones own contribution.

Our guild uses a zero-sum system. The NurfedDKP has a built in 6-week attendance value which must be at 80% or higher to receive any loot from BWL or AQ40. Simple zero-sum with attendance :)

The problem with that though is, if I don't need items from BWL anymore, and there's always enough people showing up for BWL, I still need to be there or I will drop below 80% attendance. Thus I can't get any AQ items anymore, even though I may have a 100% AQ attendance.


We planned on doing that from the very beginning. And it's not more complicated than in any fixed price DKP system. Even in the old system, we never had much discussion with item prices, they seemed to be pretty good.

We were also planning on giving 1 attendance point for the standard ~4 hour weekday raid, and give out more for longer weekend raids.

How about something like this?

Essentially, a raider's current point total can be described as:

score = t - xg

"t" is the total amount of time that a member has spent on guild raids. If they attend two Molten Core runs that each lasted 4 hours, this value is 8.

"x" is the sum of the "item multipliers" of all the items that the raider has purchased. In this system, instead of using set point costs for items, the value of the items -- ie, the amount of contribution needed to merit an item -- changes based on the pace of the guild's progress. I don't have preset item multipliers, but they can easily be extrapolated from a previous DKP system. Essentially, the multiplier of an average item is considered to be one (1), though better items have larger multipliers, and worse items have smaller multipliers.

(To import values from a DKP system, find the average value of all items in that system, and set that value to 1. IE:average * x = 1 Then, you can multiply each item by x to find its adjusted multiplier in my system.)

"g" is the central value for this system, and can basically be defined as the average time contribution that a given member needs to contribute to merit an item. This value is constantly updated after each raid, and is calculated by the formula

g = (40 * cumulative time) / (cumulative items)

As a simple example, let's say that it takes a 40-person raid 5 hours to complete Molten Core, and they receive 30 items on that run. For each member to receive an item, the guild would have to make 1.333 runs (30 * 1.333 = 40) at 5 hours each. Thus, to give each person an item at that rate, each member would have to work for 5 * 1.333 = 6.665 hours; this number is g.

The key to this system is the idea that g would be calculated not for a single raid, but for all the raids by the guild. This allows the system to reward participation in instances that don't yield items initially -- Blackwing Lair, for example -- and require proportional participation to the guild's total effort.

In summary, this system basically measures one raider's participation versus the maximum time he could have participated, and assigns each item a value based on how much effort was needed to acquire it (with the unit being hours of time).

I see nothing wrong with that. Dramatic incentive for members that have for one reason or another, finished getting gear from an instance should still have to go there so long as some main raiders need gear from there. Far too often do people get all the items they need from an instance (prot warriors and AQ for example) or people that have no chance of getting an item that run (people low on DKP, when there's multiple people with higher DKP in their class already there) decide they have better things to do than run an instance wherey *they* can't get loot.

Of course, NurfedDKP has massive problems with their fixed item prices and their 'upgrade' system as well as their priorities (Hunters can't get weapon drops til after all rogues and warriors). With a years worth of data in the system, my old guild found that there was a nearly direct relation between Total Raids Attended and position of most DKP. Someone had full GS, Xbow of Smiting, Prestors, Accuria, Arcimtiros, DFT, Dooms Edge, CHT, while almost every other hunter was in 5 GS with Rhok, 2 Dawns Edge/Bone Slicing, Tarnished Elven Rings, Darkmoon Neck, And the guy with full GS was still point leader after missing a few weeks of raids, and most likely would have gotten constant first choice of gear in AQ. The problem: Orriginal raiders balanced their points between each other with almost every item being full cost, and were thus all balanced and equal between each other, New raiders would purchase items for full cost when they showed up, giving the older raiders an immense boost, but now all the older raiders were purchasing items at upgrade costs (often 1 point when upgrading from a 20 point t1 to a 21 point t2 item) causing new raiders to drop disgustingly into the negatives.