TradingView alert in, managed position out — on hardware you control, with credentials that never leave it. It runs whatever rules you give it, so it does not need our strategy to be right, or any strategy at all.
Bracket verification, position reconciliation, ghost detection, circuit breakers, a kill switch. The things you only think to build after they have cost you money.
Why trust it: we test our own ideas against matched controls and publish the ones that fail. Liquidity sweeps, timeframe continuity, bar-pattern combos, opening-range breaks — all killed, in public, with the numbers. The whole graveyard is here.
Get the radar ✦ Read the open trade ledger — freeA short-horizon oversold reading, taken only while the longer trend is still up. The oversold trigger is what carries the edge; the trend condition is a risk filter and is not claimed to add return. Confirmed on the closed bar, entered at the next open — never the same bar.
Thirty names can fire on one close and a small account funds three. Which three you take is measurable, not a matter of taste — the ranking we use beats a random pick from the same day — and its inverse trails random by the same margin, on 399 names it was never fitted to. The effect is real but modest: about a fifth of what an unranked pick from the same signals earns.
Sell on a short-term strength signal, or on a hard time limit — whichever comes first. A 1R stop was tested on identical entries and made it worse — it binds 12% of the time and costs a third of the edge. The only stop is a catastrophe backstop 8R below entry, which binds on ~0.03% of trades and halves the worst historical loss at ~1% of expectancy.
The risk unit is measured from each name's own recent volatility. Shares = your risk budget ÷ that unit, which means the calm names take the biggest positions — not the volatile ones.
Every close, 228 names scanned and the full board published — ranked, with the losers shown too.
Every signal is recorded at the close that produced it, before anyone knows the outcome. Results are computed afterwards and never revised. Losers included. No token, no signup — read it yourself.
Open the live ledger →20 tested · 1 kept
Liquidity sweeps, timeframe continuity, bar-pattern combos, opening-range
breaks — each measured against a matched control and each refuted. The
numbers that killed them are published alongside the one that survived.
Depth: 64 years of daily bars back to 1962 — 1987, 2000, 2008, 2020 and 2022 all inside the sample. 150,000 observations on timeframe continuity alone, 112,000 on sweeps, 31,478 intraday on opening ranges, 24,862 signals on the one that lived.
See what we killed →The bot is the product. It executes whatever rules you give it, on your own machine and your own broker, and it is priced against what building it would cost you rather than against what any signal earns. The radar is an optional add-on: the one strategy that survived our testing, for people who would rather not bring their own. Buy either alone — the radar does not require the bot, and the bot does not require the radar. The research and the forward ledger are free regardless.
The bot is a tool, not a promise — it executes rules, it does not know whether they work. MR Core has a 64-year backtest and zero forward trades. The open ledger is how you check it, and it is free to read without an account.