A round of about two and a half minutes against a board of US-listed companies ordered by market capitalisation. Three stages of five; one company is already placed with its figure showing, and the other four arrive one at a time with the figure hidden. You pick the rank each card claims, and the answer comes back immediately with the real figure and one sentence about why the guess went the way it did. Off by one earns partial credit and keeps the streak alive, a wrong rank costs three seconds off a 45-second stage, and the gap the board guarantees between neighbours tightens every stage — twofold, then one and a half, then 1.3. The lesson is what survives the round: the companies you can name are not the large ones, revenue is not market cap, and a board of this shape needs a logarithmic axis because the top name is more than ten times the fifth.
Three stages, four placements each, twelve in a round. Every stage opens with an anchor already on the board — the true largest or the true smallest of the five — with its market cap visible, so you always have one reference point. The other four arrive one at a time with their figures hidden. A stage lasts 45 seconds, and the round takes about two and a half minutes in total. There are no lives: the clock and the streak are the whole cost of a mistake.
The board is a column of placed cards with a rank target beside each boundary. Tap one and the card lands there, claiming that rank. On a keyboard, the up and down arrows move the cursor between targets and Enter places the card; during a reveal, Enter moves to the next card. Nothing is dragged — a single tap is the whole interaction, every target is at least 44 pixels, and there is no hover and no long press.
An exact placement scores 100 points before the multiplier; off by one scores 40 and keeps the streak; a wrong rank scores nothing, breaks the streak and takes three seconds off the clock. The multiplier rises half a step every three correct placements up to 4×. Each stage banks a time bonus of five points a whole second left, capped at 150. A perfect round with 30 seconds left on each stage is 2,550.
The round ends after the twelfth placement. The result card shows the grade, the score, and three separate figures: accuracy, which is the weighted score across all twelve placements and is what the grade reads; exact placements out of twelve; and off-by-one placements out of twelve. They are separate on purpose, because a round of twelve off-by-ones would read as twelve correct answers on a hit count and is in fact 40%. Below that sit the best streak, a comparison with the site average when one is available, the share link and a review of every placement with its real figure.
Every round is dealt from one seed plus the board's version, and the share link carries both as ?run=seed.version. Opening that link in the same language deals the same three stages in the same order. When the board has been refreshed since the link was made, the link still opens and says so: the round is dealt from the current board, with both snapshot dates named when they are known and only the version number when they are not. Leaving mid-round keeps the stage, the score, the streak and the seconds left on your own device; a save made against an older board is discarded rather than re-ranked against new figures.
The two languages are different boards, not translations. This one is companies with a primary common-stock listing on NYSE, Nasdaq or NYSE American, in dollars; the Chinese one is A-shares in yuan, where four of the five largest are banks and three of the companies people name first are not listed on that market at all. Because the boards differ, the same calendar day gives the two languages different daily rounds, and the knowledge cards and FAQ are written separately rather than translated.
It measures what the equity is priced at in total, not what the company owns, earns or employs. Two firms with the same revenue can be an order of magnitude apart on this measure, which is why the board can be sorted on it alone and still surprise you.
Source: SEC Office of Investor Education, "Market capitalization"
Enterprise value adds debt and preferred claims and subtracts cash, so a company carrying heavy debt can be far larger on that measure than on this one. The board uses market cap because it is the figure the public sources publish consistently; the distinction matters the moment anyone compares two firms with different balance sheets.
Source: CFA Institute curriculum, equity valuation, enterprise value
A retailer moving several hundred billion dollars of goods a year can be worth less than a chipmaker moving a fraction of that, because one runs on thin margins and the other does not. On this board, Walmart sits below Micron.
Source: Frozen snapshot, companiesmarketcap.com, 2026-10-16
The companies most people can name serve consumers directly, and most of the largest serve other companies instead. Every stage on this board is built to contain at least one pair where the well-known name is the smaller one, because that pair is the whole point of the game.
Source: Dataset rule: every dealt stage carries a fame reversal
The largest company on a five-name board here is routinely more than ten times the fifth, which is why the reveal uses a logarithmic axis. A straight axis would draw the bottom four as one indistinguishable line and hide exactly the comparison the round just asked you to make.
Source: Reveal chart: bar length is log-ladder distance
Caps move daily and ranks swap. This board is frozen on one date, shown on every reveal, and neighbours in a stage are never closer than 1.3× — far enough apart that ordinary monthly drift does not make the answer wrong before the board is refreshed.
Source: Dataset rule: minimum neighbour ratio 1.3×
This one is primary common-stock listings on NYSE, Nasdaq or NYSE American, checked line by line against the exchange symbol directory. That rule is why depositary receipts and second listings are absent, and why two famous names are missing: the directory records one as capital stock rather than common stock and prints no share class at all for the other.
Source: Nasdaq Trader symbol directory, 2026-10-16
No — this is a size-reading exercise and nothing on the page is advice. The game shows a frozen public snapshot of company names and market capitalisations so that one misconception can be tested: that the companies you know best are the big ones. It makes no claim about any company's prospects, offers no forecast and takes no view on any share price, and there is no account to open.
Because landing next to the right slot shows you read the scale correctly and missed one comparison. It is worth 40% of a placement and it keeps the streak, which is also why the result card refuses to count it as a hit: a round of twelve off-by-ones reads 40% accuracy and a D, not 100% and an S. Accuracy, exact placements and off-by-one placements are printed as three separate figures for exactly that reason.
From one public source on one date, frozen in the repository with its snapshot date printed on every reveal. The figures are stored as whole units — millions of dollars — so no rounding differs between the reveal, the chart and the result card, and every row is re-derived from the frozen page by a check that runs before the game can ship. Caps drift, so the board is reviewed at least once a year and its version number moves whenever a figure does.
Because the board's rule is a listing rule, applied line by line. A company is included only if the exchange symbol directory records it as a primary common-stock listing on NYSE, Nasdaq or NYSE American, which leaves out depositary receipts, second listings and any share class the directory labels as something other than common stock. The rule is deliberately mechanical: deciding membership by reputation is how a board starts reflecting the author's taste.
No — they are different boards, so the same seed and the same calendar day deal different rounds. The English board is US-listed common stock in dollars; the Chinese board is A-shares in yuan, where banks and energy take most of the top ten and several of the best-known Chinese consumer brands do not appear because they are not listed on that market. Only the shell, the scoring and the one idea are shared.
Yes — the stage, the score, the streak and the seconds left are kept on your own device, and reopening the game resumes that stage. What is kept is the list of placements rather than a score, so a resumed round cannot disagree with the answers that produced it. A save made against an older board is thrown away and a fresh round starts, because re-ranking yesterday's answers against today's figures would mark them against a board you never saw.
Company names and market caps are a dated public snapshot, shown for knowledge only. Not investment advice, no forecasts, and no view on any share price. No account can be opened here.
