Market question: Sales Queue

I've been halfheartedly using the markets for a few years and recently decided to "expand" my gaming fun by learning to work the markets. So,a couple of questions.

How are sales determined when prices are identical? If there are 3 of X recipes for sale, all for Y pricing, is the oldest placed on the market sold first? That would seem to be logical to me.

Additionally, if there are 3 of X recipes for sale at 5000 and I place X for sale at 4500 AND someone bids 6000, whose recipe sells first? The older recipes (5000) or the newer (4500) for 6000? Again, the cheapest would seem to sell first but blind auctioning might dictate that the "older" recipes would sell first at the higher price point.

Thanks in advance for some answers. Hope this all doesn't seem elementary but I'm a little at sea in economics and free-market forces.

Ties are broken by the allegedly random generator. A lot of us tested this year(s) ago led by peterpeter and he documented the results around here somewhere.

Barring bugs, the lowest for sale item will sell to the price that meets or exceeds its listed amount. In your example of 3 listed at 5000 and yours listed at 4500, yours will execute the sale for the person that afterwards comes and bids 6000.

However if there were bids for 4999 that the 3 for 5000 would not sell to, yours would immediately sell for 4999.

Thanks SwellGuy. Definitely live up to your name!

He's Kilroy too? Who knew?

Turns out I still have my old market columns in text files on my pc. Here's the one on resolving ties.

================================================

Which Bid Wins?

A couple of weeks ago, I decided to tackle a question which has been puzzling marketeers for countless generations. When there are multiple bids for the same item at the same price, which bid is resolved first when an item is listed? If the bids are for different amounts, then the highest bid should always get the item, but when the bids are for the exact same amount, no one really knows what happens.

There are two dominant theories. The first is the FIFO theory: First In, First Out. According to this theory, the first person to place a bid will get an item before the second person who places a bid. The last person to place a bid will be the last person to win an item.

The other theory is that the bids are resolved in random order. According to this theory, there is no way to know which person is going to receive the item when the bid amounts are the same. Some random roll of the dice on the server makes that determination.

In order to put these two theories to the test, I gathered up a few stacks of unpopular recipes: Level 11 and 14 Fear/Range recipes in the Horror set. I asked for volunteers from the Market section of the forums and the global channel TheMarket. I asked each of the volunteers to place a bid for exactly 89 influence on the level 14 recipe. Nine people offered to help. I carefully recorded the order in which they placed their bids. 0th Power placed a bid first, then Miuramir, Macskull, Anti-Gravity Man, Lili, AC Bolt, Blazing Stone, Ironblade1, and finally Grandpa Squeak. Once all nine bids were in place, all for the same item, all for the same amount, I listed one recipe for sale at 1 influence.

The recipe sold instantly to... (drumroll please) ... 0th Power. I listed another recipe. It sold to Miuramir. So far, it was looking like the FIFO theory might be correct. I listed the third recipe and leaned forward into the monitor, waiting to see who had bought it. Would it be Macskull? No! It was AC Bolt! The sixth person to place a bid was the third person to get an item. The next recipe went to Grandpa Squeak, the last person who had placed a bid! FIFO was out the window! The remaining recipes went to Ironblade1, Lili, Macskull, Anti-Gravity Man, and Blazing Stone. First in was definitely not first out.

We could have stopped there, and just assumed that the random theory must be the correct one. After all, the results looked pretty random after the first couple of names. That wasn't good enough! We decided to run the exact same experiment over again: the same people placing the same bids in the same order on the same recipe. If the purchase order was determined randomly, then we would see a new and different mix-up of names. That's not what happened.

I listed the first recipe, and once again it went to 0th Power. I listed the second recipe, and once again it went to Miuramir. Who would get the third recipe? Would it be AC Bolt again? Yes! Yes it was. The exact same people received the recipes in the exact same order. The random theory was sporting a black eye!

Now we were eager to learn more. We tried the same experiment with level 11 recipes instead of level 14. It didn't make the slightest difference. The buy order seemed to be fixed in stone. What magical power gave 0th Power a lock on first place? What dark curse kept Blazing Stone stuck in last place?

It didn't look alphabetical by global name or character name. We tried comparing the levels of the characters, and the number of vet badges. Nothing made any sense. What about other badges? Market only badges? Influence on hand? Server? Age of the character? Time zone? People were throwing out information and data points as fast as they could type, and I was trying to compare everything. Suddenly, I noticed one point of correlation: origin. The first two characters to get buys were both Natural origin. Next was the one Magical character, then all three Science characters, followed by the lone Mutant. Did the market favor Natural characters and punish Mutants?

I double checked my notes, and realized that one origin was missing. For the correlation to be perfect, Grandpa Squeak needed to be either Science or Magic. I typed out the question, and waited for a response. Grandpa Squeak's character was... a Mutant. The series was broken. Humans are very good at spotting patterns. Our brains are wired for it. Sometimes we're so good that we spot patterns which aren't even there. I decided the partial correlation could be chalked up to coincidence. We still didn't know what factor determined the buy order, and we had an endless stream of possibilities.

Luckily, our next experiment pointed us in the right direction and away from the vast swamp of character details. We ran the same experiment as the first time, except we reversed the order in which people placed their bids. Would 0th Power still come out on top, even when he was the last person to put in a bid? Would Blazing Stone still come in last, even when he put in the third bid?

Nope. Grandpa Squeak put in the first bid, and Grandpa Squeak got the first item. Ironblade1 put in the second bid, and got the second item. Blazing Stone put in the third bid, and the third item went to... Anti-Gravity Man, who had put in the sixth bid. Even though we had changed the people around, the bids were still resolving in the same order relative to when they were placed. The last person to put in a bid became the fourth person to receive an item, just like last time. Which person actually placed each bid didn't matter!

Once more unto the breeches! For the next round, I assigned a random order, starting with Grandpa Squeak and ending with Miuramir. It made no difference who placed which bid. The first two bids always received the first two items. The third bidder was always the seventh buyer. The last bidder was always the fourth buyer.

Obviously there were many potential variables which we did not test. What if each person placed two bids? What if they all bid something other than the 89 influence we had used over and over? What if a few bids were placed, then a couple of them were canceled, then a few more placed, then two items were sold, and then another bid came in? Sadly, we only had time for one more test. What if there were only eight bids instead of nine?

Quat and Hecktender stepped in to replenish the dwindling ranks of volunteers as we ran our last experiment. Once again, the first bidder got the first sale. Once again, the second bidder got the second sale. The first surprise came when the fourth item sold went to the eighth bidder instead of the ninth. Of course, we only had eight bidders, so it wasn't that big of a surprise. There was a ripple effect, though: the fifth item went to the fifth bidder instead of the eighth, the sixth item went to the third bidder instead of the fifth, the seventh item went to the fourth bidder instead of the third, and the eighth and final item went to the seventh bidder instead of the fourth.

The numbers were different, but clearly the same pattern was still present. The conversation quickly dissolved into a complicated analysis of random number generator seeds and number sequences patterns. In the end, we could only resolve to come back and puzzle away at it again on some future date. We certainly accomplished what we set out to do: to thoroughly test the two main theories regarding bid resolution order. The system is clearly not FIFO. The system is also not truly random. The bidders may have been jumbled up a bit, but they were jumbled up in the exact same way every time. There may have been a random sequence generated at some time, but during our testing window we were clearly dealing with a fixed sequence.

I will leave you with a few rows of numbers to think about. The first row is the ID of the bidder, from bidder #1 to bidder #9. The second row shows the order in which each bidder received an item. Bidder 1 received item 1, bidder 2 received item 2, bidder 3 received item 7, and so on. The final row shows the order in which the bidders received items during the final experiment, when there were only eight bidders and eight items instead of nine.

1 2 3 4 5 6 7 8 9
1 2 7 8 6 3 9 5 4
1 2 6 7 5 3 8 4

TL;DR version:
Unless you're coordinating with all the buyers, and placing bids in a systematic manner, it's random enough.