A three-minute spending diagnosis. Twelve scenarios cover four kinds of leak — working out which way of buying is cheaper, limited-time prices that are not lower, subscriptions that keep renewing, and the size of an emergency-fund gap — and the result names the one this round leaked most on. Nothing is self-reported: every question has one answer that the arithmetic settles, and every price is invented for the question and printed inside it.
The goal
Money Personality deals twelve spending scenarios, three from each of four categories, interleaved so you never know which one you are being asked about. Every question shows four ways of paying for the same thing and asks which lands lowest on one stated basis: unit price paid, total paid, spend across a year, or the gap still to cover. Exactly one option wins, and it wins by at least a yuan and at least one percent, so nothing turns on small change. The four categories are the four leaks the result can name: deals, snap buys, renewals and buffer. A round takes about three minutes.
Controls and play
Tap an option, or press the number beside it — 1 to 4. A right answer flashes and moves on by itself after a moment. A wrong answer, or a question the clock runs out on, stops and shows the correct option with one line of why, and only your press moves it on from there: the explanation is what the mistake bought you, so nothing takes it away on a timer. Each question has its own fifteen seconds, shown as a ring that turns urgent under five.
Points and streaks
A right answer scores 100 points times a streak multiplier, and a harder question is worth 30% more. The multiplier climbs one rung every three answers in a row and tops out at 2.5 — a twelve-question round cannot reach a higher rung, so no higher one is advertised. A wrong answer and a timeout both score nothing and break the streak; neither takes points away. Because a perfect round scores differently depending on where the harder questions land in the streak, the result page also shows a comparison index from 0 to 1000, which is weighted accuracy and does not move with question order.
Finishing and results
The round ends after the twelfth question. The result names where this round leaked most, and shows each category as misses over questions actually answered, plus how many timed out. Two rules decide the label: a timeout is not a leak, because what it carries is reading speed rather than a spending judgement, and no label is issued at all unless all four categories were answered at least twice. If one category was left under-measured the result says so and offers another round, rather than guessing from the other three. At most one miss in twelve reads as no real leak.
Replay and progress
With no deal link in the address bar every round is dealt fresh: eight candidate deals are laid out and the one repeating the fewest of the questions you just answered is the one you play. That makes a repeat unlikely rather than impossible, and when every candidate overlaps, the smallest overlap still wins. Your best score stays in this browser and nothing is uploaded — this game collects no financial figures about you at all, and the share text says only what the result card already showed you: the label, the grade, the score, the index, the four leak counts, how many questions ran out of time, and the deal link. Opening a shared deal link plays exactly the same twelve questions in the same order with the same option order, including in the other language.
Language and context
English and Chinese are written separately rather than translated, and the two banks line up question for question: the same twelve ids, the same difficulty, the same correct slot, and the same underlying arithmetic. Only the wording differs, and it differs where the languages differ — a discount written as 8 折 in Chinese is written as 20% off in English, which is the same fact in both. Every price is invented for its own question and stated inside it, so no question depends on one country's prices, taxes or credit rules.
