Trader publishes 6 years of BTC orders
A crypto trader posted six years of Bitcoin order data—more than 40,000 orders—to GitHub, presenting it as high‑quality context for AI agents to study full bull and bear cycles. The dataset was shared publicly on social media to invite AI‑driven analysis. (x.com)
A crypto trader has published a public GitHub archive of a real Bitcoin-heavy account, exposing nearly six years of orders, executions and wallet history for anyone to inspect. (github.com) The repository says the dataset runs from May 1, 2020, to April 17, 2026, and includes 43,214 order rows, 173,058 execution rows and 17,099 wallet-history rows. The account history is presented as a daily versioned archive with stable filenames and Git tags in the format `data-YYYY-MM-DD`. (github.com) The files are mapped to BitMEX application programming interface endpoints, including `/api/v1/order`, `/api/v1/execution/tradeHistory` and `/api/v1/user/walletHistory`. BitMEX’s own documentation describes those endpoints as order history, balance-affecting trade history and wallet transaction history. (github.com) (docs.bitmex.com 1) (docs.bitmex.com 2) In market structure terms, orders show intent, executions show what actually traded, and wallet records show deposits, withdrawals, fees and profit-and-loss movements. Putting all three in one archive gives outside researchers a way to compare trading decisions with fills and account balance changes over a full bull-and-bear stretch. (github.com) (docs.bitmex.com 1) (docs.bitmex.com 2) The repository says Bitcoin-related trades account for about 84.0% of executed notional across the full archive, rising to about 93.7% since 2022, 96.1% since 2023 and 99.0% since 2024. That makes the release less a general crypto ledger than a concentrated record of one trader’s Bitcoin positioning over time. (github.com) The README frames the project as an “open-intelligence experiment” for the artificial intelligence era and argues that most public trading content is “narrative without ledger truth.” It says the goal is to let others inspect a real execution ledger and terminal snapshots instead of screenshots or selective summaries. (github.com) The same README says the archive includes a derived equity curve and a cumulative-performance chart built from Bitcoin and Tether balances. It lists a baseline of 1.83953943 XBT on May 1, 2020, and a latest adjusted wallet-equivalent wealth of 96.38685218 XBT, or about 52.40 times that baseline by the April 17, 2026 build. (github.com) Public crypto datasets already exist, but they usually cover market-wide prices, order books or short windows of exchange data rather than one account’s multi-year trading ledger. Examples include public blockchain datasets, commercial order-book feeds and short-horizon academic or Kaggle-style market data. (github.com) (tardis.dev) (kaggle.com) The release also leaves limits. It is one trader, one venue family and one account history, so it cannot show how a broader market of traders behaved or whether the same decisions would have filled elsewhere. BitMEX’s documentation also notes that raw execution data can be noisy, which is why the repository relies on both order and trade-history tables. (github.com) (docs.bitmex.com) For now, the immediate result is simple: a trader who usually would have shown screenshots has published the ledger instead. That turns six years of Bitcoin trading from a claim into a file set other people can test, replay and challenge. (github.com)