Sunday, June 20, 2010

Goblin GDKP Variant

I missed this when Gevlon first proposed it, but he proposed a variant on the standard Gold DKP loot system.

For those people unaware of GDKP, it's a loot system where raid members bid gold for items. Items are auctioned off in a standard multiple-round, public bidding process. The gold all goes into a pot and the pot is split equally among all the raid members at the end of the run. It's good system for pickup groups because it provides an easy way to compensate people who don't win items. If you don't get any items, you at least get several hundred or several thousand gold. This encourages geared and experienced players to join the PuG, giving it higher chance of success.

Gevlon proposes a rule giving 1/3rd of the winning bid to the second place bidder, in order to encourage higher bid prices and discourage "fake" bids.

I'm not really sure that this rule actually accomplishes anything useful.

Look at it from the point of view of the optimal bidding strategy. In a normal English auction, the optimal bidding strategy is to bid (up to) what you think the item is worth. In a normal GDKP auction, the optimal strategy is to bid you think the item is worth plus a little premium, because you get back a portion of your bid when the pot is divided up at the end of the raid.

Under the Goblin variant, the optimal bid is to bid 150% of the items worth (plus a little bit for the pot-split). To see this, let us ignore the pot-split for now, and imagine that there are two items that a player values at 1000g. The player bids 1500 for both and wins one and loses one (for 1500.01).

For the winning bid, the player nets 1000g (item) - 1500g (bid) = -500g.

For the losing bid, the player nets 500g for being the second place bid.

So when the two bids are taken together, it's a wash. The extra money you get from being second is cancelled out by the extra money you have to pay to win items.

Also, note that the size of the pot doesn't increase. There were two bids of 1500g, whereas under the regular rules there would be two bids of 1000g. But the pot is 2000g in both cases.

So as far as I can see, the only person who benefits from this rule is someone who is very skilled at bidding just enough to get second place. For example, consider a spoiler who only joins in bidding when there are two other people bidding. With three people bidding, the price gets pushed up and up. When one of the original two bidders drops out, the spoiler immediately drops out right after, guaranteeing that she has second place. It is very unlikely that two legitimate bidders will drop out at the exact same price point. There will be a small difference between their maximum amount which they are willing to pay, which the canny spoiler bidder can exploit.

Far from discouraging strategic bidding, I think this system actually creates a legitimate opening for someone who is bidding to push prices up. People who win items do not benefit. Higher end players carrying the group to earn some extra gold and maybe hoping a rare trinket drops do not benefit.

This rule just adds extra complexity for no good reason. It encourages people to bid using a non-optimal manner, just because they want the extra gold from being in second place.

Regular GDKP is a good system. It is close enough to English auctions that everyone understands it, and close enough that the strategy of bidding what you think the item is worth is very close to optimal. It's been my experience that when bidding systems stray from this optimum, they are more fragile, because people don't really understand how to bid well.

Gevlon's rule of "1/3rd to the second bidder" is not an improvement on GDKP, and should be avoided, in my opinion. It will still work, but it is more complicated, has a less intuitive optimal bidding strategy, encourages strategic bidding and does not actually add any positive effects that I can see.

11 comments:

Azzur said...

Gevlon's 1/3 rule is not designed to discourage "fake" bids but rather to discourage "raid optimisation" where a raid leader may take less of certain classes to reduce competition (and hence price).

For me, this is a mere refinement over the existing GDKP system (like there are many refinements over DKP). The most important concept is the GDKP itself. It doesn't really do much, so it doesn't matter much whether the raid uses it or not.

spinksville said...

It's probably a great scheme if you have pots of gold.

It encourages people to bid on items they don't want and can't use (so that they can try to sell it to someone else in the raid).

So the question is whether the raid as a whole profits more if the 1/3 scheme is used. Does it drive bidding up high enough to make up for the fast that there's less to share among the raid (because 1/3 of each winning bid goes straight to the second bidder).

Gevlon said...

Just as Azzur said. The point is to discourage people from excluding possible competition from the raid by not inviting competition, or not accepting invitation if competition present.

Jason said...

I'm not sure I follow the math of the example of 2 bids. Would the pot not be only 1500g using the 1/3rd rule, since there is most likely a second place bidder on the item that was won??

Rohan said...

Jason, a winning bid of 1500g = 1000g to the pot + 500g to the second place bidder.

So two bids of 1500g have a pot of 2000g in Goblin GDKP, just the same as as a normal GDKP system with two bids of 1000g.

Sir Nicholai said...

I agree with your post.
I've been running GDKP on Chromaggus US for 8 months now, and have 3 25 mans every Saturday in ICC. Pots have gotten up to 187k, but I don't think this rule would help bidding or grow the pot bigger. Seems like unnecessary complexity.
Also, I don't think GDKP works at all for 10 mans, I'm not sure if Gevlon was using it for those or not.

Joshua said...

"With three people bidding, the price gets pushed up and up. When one of the original two bidders drops out, the spoiler immediately drops out right after, guaranteeing that she has second place."

That strategy would get awfully tricky once there is more than one spoiler attempting it at the same time.

Anonymous said...

You have a few problems.

I've never seen any of these type of runs pay out every member. Ever.
Why should anyone ever get a cut of the gold when they bought an item??

THe spoilers are easy to spot really. Unless the item being auctioned is a highly sought after trinket or weapon. If its a small upgrade in general, people will bid low because of the small gain. The caveat to this is simply that items that are considered BiS or 2nd best can really go super high.

Gevlon is a tool so of course his method is just wacky.

Strum said...

I'm not sure I see the reason why a bid of %150 of the item's value is optimal. Why should someone bid (and pay) 1500g for an item worth only 1000g? If you're the highest bidder, the transaction has a net -500g in value.

Rohan said...

Strum, it's 150% because you have to consider the case where you get second. In that case, the extra gold you get from being second balances the extra amount you pay if you won.

Strum said...

@Rohan:

I can see that there could be some sort of offset based on trying to increase the probability that you're the 2nd highest bid. I don't see why %150 of the value is optimal though.

Your expected return to a bid is: prob(you're highest bid)*(value-bid)+prob(you're 2nd highest bid)*(1/3*highest bid). It's not clear why those probabilities should be equal though, which is what the post assumes. Given the possible bidding strategies of the 9 or 24 other people in the raid, those probabilities can be anything.