What this guide establishes
F07 is the name used by this independent information guide. Its browser demo is Leximin Assignment. The playable demo does not establish an outside company, official affiliation, licence, app publisher or financial service. This guide describes its own virtual-coin rules; it does not certify another product.
How the six assignments work
Leximin Assignment draws nine independent values, each equally likely to be 0, 1, 2 or 3, in a 3 × 3 board. An assignment selects one cell from each row and uses each column once. There are six assignments. Sort each selected triple from smallest to largest, then compare the triples: prefer the larger first value, then the larger second value if the first ties, then the larger third value if both tie. The middle value of the winning triple is the raw reward s, from 0 to 3. This compares assignments across rows, rather than sorting each row separately.
Two simple scoring examples
A board containing only zeros gives six triples [0, 0, 0], so the raw reward is 0. A board containing only threes gives six triples [3, 3, 3], so the reward is 3. For a mixed board, first compare the smallest entries of the sorted assignment triples. A triple with a higher middle entry does not automatically win if its smallest entry is lower.
For the board with rows [0, 2, 1], [2, 1, 0] and [1, 0, 0], the winning sorted triple is [1, 1, 1], giving reward 1. An available triple [0, 2, 2] has middle value 2 but loses because its first value is smaller.
Reward powers and gross return
Reward power 1 uses s, power 2 uses s², and power 3 uses s³. Each weight has its own exact mean μ over all 262,144 equally likely boards. The gross multiplier is 0.96 × sᵖ ÷ μ for the selected power p; multiply it by the captured stake for the raw gross payout. Each mode therefore has a 96% expected gross return in the mathematical model. This is an average over the full distribution, with no guarantee for a board or session. Gross return includes any stake recovered; it is not net profit.
The exact mean weights for powers 1, 2 and 3 respectively are 606681/262144, 1520669/262144 and 4021275/262144. Each is the average of its own powered raw rewards, before multiplying by 0.96.
Restoring the virtual wallet
The first wallet starts at 1,000 virtual coins. When no draw is active and the balance is strictly below 10, “Restore balance” sets it to exactly 1,000. It does not add 1,000, refund earlier stakes or start a draw. The fractional remainder, selected stake and selected power stay unchanged. A balance of exactly 10 is not eligible. Restoration is optional and never automatic.
Local storage and outside requests
The demo uses browser localStorage for the wallet balance and fractional remainder. Clearing downloaded files from the cache can leave that storage intact; site-data controls are separate. The pages include Google Fonts stylesheets and Statinside counter 44 in the published site. These requests and local wallet storage do not establish anonymity, describe all possible processing or determine how long outside services retain data.
Frequently asked questions
Does the demo prove F07 is an official operator?
No. A working browser demonstration does not establish an outside operator, company, licence or affiliation.
When can I restore the balance?
Only while no draw is active and the balance is below 10. “Restore balance” sets it to 1,000 and preserves the fractional remainder, current stake and power. It does not add 1,000 or start play.
Does 96% mean I recover 96% on every draw?
No. It is the mathematical expected gross return across all equally likely boards in each separately normalized mode. Individual boards and sessions can return less or more.
