Talk:Merchant's house/@comment-4111213-20120221192840/@comment-4102115-20120304102609

Thanks Mancini for sharing your experiences. If more people shared their accurate documented experiences we could would out more accurately what the coding is to more effectively predict a blueprint to be offered / swapped out.

It certainly seems that Leonid will never offer more than one blueprint at any one time. And as I've noted above on the page if you buy 1, 2, or 3 items Leonid will replcae those 1, 2, or 3 items only, so if there is a blueprint piece and you buy 1, 2 or 3 scrolls every cycle, that blueprint piece will never be swapped out.

So the only way to force a blueprint piece or a scroll to be swapped out if you don't want to buy it, is to buy nothing for one or more cycles. As also noted above and in the comments below, with v1.2.1 and above, if nothing is bought in a cycle, then one shelf is randomly replaced. I have seen a blueprint piece be offered one cycle, and the very next cycle it was the shelf that was swapped out, yet as also shown below, one time I had a blueprint piece stay there nearly 4 days, with buying some shelves some times and when buying nothing, it just took many cycles before that shelf was swapped out. With a cycle time of 90mis plus 60 mins having to check in on the game twice each cycle, the laws of randomness says either extreme is not unusual.

It looks like we should nearly have enough data to narrow down the algorithm.

eg: It seems that not only that: (a) 2 pieces will not be simultaneously offered, but (b) it seems if there was a blueprint on offer one cycle, one will not be offered the next cycle - whether that piece was bought or randomly swapped out. But also (c) there seems to be something of a maximum amount of replacement items between each blueprint.

So, with this information, Ezcry's method of a complete buyout every 60mins brings a blueprint piece every second or third cycle. I miagine Ezcry would have seen each 60mins after buying all 4 items:

1 blueprint + 3 scrolls

4 scrolls

1 blueorint + 3 scrolls

4 scrolls

4 scrolls

1 blueprint + 3 scrolls

ie. Never 2 blueprints offered 2 cycles back to back, and once the blueprint is bought/swapped, there will be only about X number of scrolls before the next blueprint is offered.

Note: X could be as low as 6, which would actually fit in with Mancini and my findings of getting blueprints at a higher blueprint/scroll ratio than Ezcry - when I was buying every blueprint I saw, but not every scroll, I was getting one blueprint about every 6th scroll. It just took longer than 2 or 3 x 60mins as it was 2 or 3 x 60+90mins, but was cheaper as I bought half as many scrolls and no expensive destroy or sell or large time scrolls.

Summary: the hypothesis of a blueprint piece offered about 6 scrolls apart, so long as no blueprint piece two cycles back to back, seems to hold for all 3 of us.

Note: X can't be exactly 6, otherwise Ezcry would get a piece every second cycle, but it seems to be something like the Repair algorithm - but weighted the other way, where it is either the 6th, 7th or 8th offer.

Or something like that...