$10,000. Two secret years of real market history. One button: sell or buy back in. Cash earns the real T-bill rate while you're out. Your opponent — the Desk — is plain buy & hold: one position on day one, never touched again. Beat it, prove it wasn't luck, and sign the leaderboard.
The markets. Stocks: daily S&P 500 total returns (price + reinvested dividends), January 1928 → present — 24,700+ trading days, extended past 2019 with the S&P 500 Total Return index. Crypto: daily BTC/USD closes, August 2011 → present. Each game deals a uniformly random two-year window. In daily mode the window is seeded from the UTC date, so everyone gets the same hand.
The tilt report. Trade enough in a run and we grade what the pain made you do: exits within 3 days of a sharp drop, re-entries within 3 days of an exit, and how much faster you clicked while losing. A+ goes to people the market couldn't rattle.
Leverage. Optional 1×, 2× or 4×. A leveraged day is your leverage times the market's move, minus financing at the T-bill rate on the borrowed part. Fall to 5% of your starting stake and you're margin-called: position closed, you ride out the window in cash. The Desk never uses leverage. It doesn't need to.
The leaderboard. Beat the Desk and you can sign the board — display name public, email private. Ranked two ways: dollars taken off the Desk, and percent finished above it. Top 10 hang in the lobby.
The Desk. The Desk is just buy & hold: $10,000 into one position on day one, then lunch. It does not watch the chart. It does not have feelings about wicks. It wins anyway, which is the entire point.
Cash. While you're out, your money earns the historical 3-month T-bill rate for those exact dates. Hiding pays a little. It rarely pays enough.
The monkey audit. Beating the Desk once is easy to fluke. Every win is replayed by 1,000 monkeys using your exact schedule — same number of exits, same out-of-market block lengths — at random dates in your window. Clear the 90th percentile or your win is stamped LUCKY.
Part II. The game is a simulation; the audit isn't. Drop a Hyperliquid wallet (public, read-only) or an exchange/broker CSV and Arlo runs the same statistics on your real trades — monkey percentile, you-vs-doing-nothing, hold-time asymmetry, fee drag, tilt profile — and emails you the full assessment. Free.
Known limitations. No fees, taxes, or slippage in the game — all of which hurt you, not the Desk. Survivorship bias is doing the Desk a favor. The dataset and the daily window are visible in the page source; you can cheat, if beating a lunch break matters that much to you.