What it is not
read this before the numbers.
OUTRIDER is a set of research and monitoring tools with a paper trading desk and an optional live mode. It does not promise results, live trading can lose money, and several of its ideas have already been measured and found wanting. This page says which.
01Paper by default, live only if you turn it on
- The desk starts in
papermode. Real orders happen only inlivemode, with keys you provide, the confirmation sentence typed exactly, and a venue switched on (Live trading). The keys stay on your machine or in the console's encrypted store; OUTRIDER holds no funds for you. - Live trading can lose all the money in the trading wallet or account. Memecoins can go to zero in minutes; slippage, failed transactions, RPC outages and exchange errors happen. The caps limit each order, not the market.
- Paper results include fees, slippage, priority tips, latency and failed transactions, but they are still simulations. Real fills, real queues and real market impact can be worse.
- The paper bankroll, paper equity and paper PnL in the dashboard, console and Telegram are not money and not earnings.
02What the measurements say
These results are written in the code and its comments by the author. They are observations on past samples, not forecasts.
| Idea | Result | Where |
|---|---|---|
| Liquidation-map entries (trade toward a cluster, fade a cascade) | No entry edge after costs, walk-forward and against random timing. Use the map as a risk tool. | src/lab/liqbt.mjs; console description of the Liquidation map |
| Trend following | Protects in falling years and lags in rising years; it can trail simply holding BTC for long stretches. | Console description of Trend & leverage; trend-backtest |
| Memecoin lanes (moonshot, graduation, revival, smart) | Unproven. None has a demonstrated edge; each exists to be measured on its own trades. | Console description of Memecoin lanes; src/config.mjs |
| pump.fun entry filters | None of 288 filter combinations tested was positive after costs out of sample; the filters make the paper book lose less. | Comment on pump.minUniqueBuyers in src/config.mjs |
| Jev forecaster | Early public tests reported AUC around 0.5 and losses after costs; measure it on your own questions. The desk keeps it in shadow mode until its own promotion rule is met. | jev-report |
| Leverage | Higher profiles had deeper drawdowns: 24 %, 32 %, 44 % for safe, growth, aggressive on Hyperliquid daily candles 2021-01-07 → 2026-09-27 (BTC hold: 77 %). | research run of 2026-09-28 (Trend vs BTC); leverage-backtest |
Run the tools yourself on your own data. If a result disagrees with this table, trust your run over this page.
03How to judge any result
- Fewer than 30 trades is not evidence. The report says so.
- Look at the result without the luckiest trades. If it vanishes, it was luck.
- Prefer out-of-sample numbers:
sweep,wallets,fitandevolveall judge on data they were not chosen on. - Count your tries. The more configurations you test, the more of them pass by chance;
evolvereports the probability of backtest overfitting for this reason.
04Not financial advice
Nothing in OUTRIDER, its output or these docs is investment, financial, legal or tax advice. Memecoins, leveraged perpetuals and prediction markets can lose most or all of the money put into them, quickly. Some of these markets and data sources are restricted in some countries; you are responsible for using the software lawfully where you are.
05Before you trade live
Live mode (Live trading) sends real orders with your keys. Start on paper, then a dry run, then the smallest caps. Changing live settings or the mode applies after a restart; close live positions before you switch back to paper. Live trading can lose all the money in the trading wallet or account. Nothing here is investment advice.
06Data you see
Wallet addresses, positions and trades shown by the desk come from public blockchains and public APIs. The whales.wallets and hlcopy.watch lists start empty; add wallets you choose yourself. Self-reported leaderboard PnL is not verified by anyone.