Open source
Code is the kind of evidence a reader can actually check — you can read it, run it, and find the bugs. Everything below is public, with the reasoning and a video where one exists.
Start here
buou_trail ★ 302 Tiered trailing take-profit. Author's words: "materially improves win rate — mostly by keeping your hands off the keyboard. Discipline, maxed out."
jiezhen ★ 170 Needle-catching strategy: use EMA to judge short-term trend, then rest a limit order below to catch the wick as it spikes down. Runs on a per-second loop.
ds ★ 846 Using DeepSeek to trade. Author's note: "the whole point is falsification testing" — explicitly not a commercial product.
Trading tools
4-
buou_gridAS high-frequency grid. Two-sided OKX WebSocket quoting with position cap and hedge threshold. Note the example config uses 50x leverage. -
shijianheyueEvent-contract auto-order client (reverse-engineered). ★ This is the Windows client I actually traded with. -
pylighterPython SDK wrapper for the Lighter.xyz exchange. -
pacifica-sdkHigh-frequency trading SDK for Pacifica.
Research & experiments
4-
dongliangMomentum strategy experiment. Author's warning: "P&L and risk come from the same source — start with tiny leverage; short-side risk is large, so live I would not exceed 0.3x." -
KronosK-line model prediction experiments (local use of an open-source financial candlestick model). -
asBackBacktest implementation of the core AS strategy logic. -
ak47_backpack_market_makerBackpack market-making quote strategy. Author claims minimal adverse selection, tested across 300–400M in volume.
Learning material
2A note on the numbers
Star counts are shown for context, not as a quality signal. Several of these are research experiments with explicit warnings in their own READMEs — leverage limits, "just for fun, mind the risk", and so on. Read those warnings before running anything.
These tools place real orders and manage real positions. Nothing here is investment advice, and running someone else’s trading code on your own account is entirely your own risk.