Customizing offers for sales of combinations of virtual items

A method comprising:

• performing operations, using one or more processors of a networked system, for gravitating toward an optimal variant of a plurality of variants of weighted offers for presentation to a plurality of users, the operations comprising:
• randomly generating the plurality of variants of weighted offers for storing in a database of the networked system, each of the plurality of variants of weighted offers including a plurality of discount percentages and a plurality of probability percentages corresponding to the plurality of discount percentages;
• determining that a first variant of the plurality of variants is outperforming a second variant of the plurality of variants over a first time period with respect to revenues obtained from the plurality of users over the first time period;
• repeatedly randomly modifying at least one percentage of the plurality of percentages until the second variant matches or exceeds a performance of the first variant with respect to revenues obtained from the plurality of users over a second time period;
• and
• selecting the second variant as the optimal variant for the presentation to the plurality of users over a third time period.

A non-transitory machine-readable storage medium storing a set of instructions that, when executed by at least one processor, causes the at least one processor to perform operations for gravitating toward an optimal variant of a plurality of variants of weighted offers for presentation to a plurality of users, the operations comprising:

• randomly generating the plurality of variants of weighted offers for storing in a database of the networked system, each of the plurality of variants of weighted offers including a plurality of discount percentages and a plurality of probability percentages corresponding to the plurality of discount percentages;
• determining that a first variant of the plurality of variants is outperforming a second variant of the plurality of variants over a first time period with respect to revenues obtained from the plurality of users over the first time period;
• repeatedly randomly modifying at least one percentage of the plurality of percentages until the second variant matches or exceeds a performance of the first variant with respect to revenues obtained from the plurality of users over a second time period;
• and
• selecting the second variant as the optimal variant for the presentation to the plurality of users over a third time period.