Compare
Chart Library vs Tickeron — a signal versus the evidence behind it.
Tickeron packages AI “robots,” pattern-recognition scanners, and trade signals: it tells you what to do — enter here, exit there, this pattern is forming. It is an opinion engine.
Chart Library produces no signals and no directions. For any (symbol, date, timeframe) it returns the cohort of real historical analogs and their calibrated forward-return distribution with an audit receipt — the evidence a signal would be built on, served to an agent over API/MCP so it can decide and explain for itself.
Signals vs calibrated evidence
- Tickeron = directives. Buy/sell signals and bot trades. You either follow them or you don’t; the historical hit-rate behind a given signal is rarely something you can audit yourself.
- Chart Library = auditable base rates. The analog cohort’s distribution per horizon plus the receipt (nominal 80% band held 80.8% across 300K+ audited cases, PIT-flat). Descriptive history, never a call.
- An agent should reason, not obey. A capable trading agent wants the evidence (calibrated precedent) to form and defend its own decision — not a black-box signal to relay.
- Composable. If you do use signals, Chart Library is the sanity check: pull the calibrated base rates for the same setup and see whether the signal’s implied odds match the history.
Frequently asked questions
- Does Chart Library generate trade signals or bots like Tickeron?
- No. It returns the calibrated forward-return distribution of a setup's historical analogs plus a coverage receipt — evidence, not a buy/sell signal. Your agent (or you) makes the call.
- Is it a Tickeron alternative?
- It's the layer underneath a signal: the auditable historical base rates a signal should be justified by. If you want evidence instead of directives — especially for an AI agent to reason on — Chart Library is the fit.
- Does it predict the move?
- Never. Similarity-only by design: it surfaces what analogs did next as a calibrated distribution with a receipt, no directional forecast.
Try it
Run a cohort_analyze call.
Free Sandbox tier — 1,000 calls/day, no authentication. MCP install for Claude or Cursor takes 30 seconds.