Every signal's DNA, vetoes and context, in the open.
Sentinel Aleph reads Smart Money structure across 4-hour crypto markets, passes every setup through nine risk gates, and publishes what happened next in an open ledger. Not a promise of profit. A record you can check.
The desk that runs on your machine and the kit that connects AI agents are public. Audit the vault, the order gate and every tool before you trust them with anything.
The desktop trading desk: Sentinel signal bots, paper DCA and Grid bots, backtests and a local key vault.
The agent kit for Sentinel's MCP server: connect Claude Code or any MCP client to signals, the ledger and backtests.
Every MCP tool with its arguments, usage units and limits: ribqa.com/mcp. The Aleph Edge screen is from the app's sample-data preview; the MCP screen is real server output.
The counters below restart from zero every day at 00:00 Swiss time (Zurich, CET/CEST), so each day is judged on its own. Nothing older is hidden: the full history stays in the ledger, labelled as history. Results decided today are shown as they stand; winners settle faster than losers, so an early read can run higher than the day will finish.
Updated 8 Oct 2026, 09:48 UTC. Every figure above is read from the same public endpoint the ledger page uses, at most a minute old. The net is a sum of percentage points over every entered position not closed by the BTC veto, not a return on capital; the per-trade average beside it is the honest scale. Positions still inside the 96-hour guard are counted as they stand today and can still change. Bitcoin today is shown beside the book because a green day in a rising market is mostly the market: read the two together.
Three stages, all of them visible. No stage is a black box, and no stage is allowed to publish its own report card.
23 sources read the same tape: order blocks, fair-value gaps, market-structure breaks, liquidity pools, sweeps, indicator families and math models. Each one scores independently; agreement is measured, not assumed.
A setup is refused, not published, when Bitcoin is crashing, when dominance turns against altcoins, when the source's own live record is weak, or when the book already holds too much. The refusals are logged too.
Every published setup is replayed against exchange candles by an evaluator that was checked against raw Binance data. Wins, losses, scratches and expiries are all counted; the cost of trading is charged on every one.
A result like this depends on where the target and the stop sit. Ours is published next to its geometry, its sample size and its confidence interval, never alone.
Where the book's edge is not established, the signals page says so in plain words, above the signals. That notice comes down only when a pre-registered test says it should.
We regress every trade on Bitcoin's move over the same window and publish the split. Most crypto signal products would rather you didn't ask.
Counters can be reset for a new run; rows cannot be removed. The verified ledger exposes both the current run and all-time history, side by side.
Aleph Edge is the desktop companion to Sentinel. It follows the published signals, runs paper DCA and Grid bots on public Binance prices, and backtests them on historical candles. Nothing leaves your computer except the requests it makes for prices and signals.
Every bot starts on paper with real Binance prices and fees. Real orders need a separate live build and a typed opt-in per bot.
Exchange keys are encrypted on your device (Argon2id, AES-256-GCM). Only trade-only keys are accepted; keys that can withdraw are refused.
12 DCA and Grid templates, each tested once on two years of data. 2 passed every check; the other 10 are labelled with the check they failed.
Screens from the app's sample-data preview. Every figure in them is sample data, not a trading record.
Read the ledger first. If the record earns it, join the beta and follow the same signals with the same disclosures.
No results recorded yet
