Research note

TimesFM Risk Range: display-only, uncalibrated

Zero-shot median and P10–P90 paths are display-only research. No trading rule or empirical calibration result is inventoried.

2 min read

Backtest status: Not applicable — display-only forecast. TimesFM output never votes or trades; empirical calibration and trading efficacy are unknown.

Open this research tool in Backtesting

Open live page · Forecast — TimesFM Risk Range

The rule

The original page displays a TimesFM zero-shot median path and P10–P90 range. Its opening configuration is SPY, 512 daily closes and a 21-session forecast horizon. These inputs are settings, not performance observations. The Library reference opens no inference; use the original live page to inspect its research UI.

Entry, exit, long/short/cash exposure, stops, targets, fill timing, sizing, costs and slippage are not applicable because no trading rule is specified.

Methodology

Evidence as of: the INVENTORY review dated November 5, 2026; this is the consolidation planning baseline, not a new observation window. Publication does not extend any study's sample.

The honesty boundary is explicit: P10–P90 is not calibrated confidence. The inventory has no empirical coverage test, frozen forecast vintage, historical benchmark comparison or trading study. Input window length must not be described as a successful track record. A model can output a range without establishing how often future prices fall inside it.

Check the returned source and price basis for a particular forecast; neither a dated validation sample nor a tested total-return strategy basis is recorded here. No inference was requested for this article.

Results

Study / variantInventory evidenceWindow / basisStrategy return / CAGRBenchmarkMax drawdownClosed trades / observation NSignificance / ICNative verdict
TimesFM median / quantile displayD04zero-shot; validation window not recordednot applicablenot applicablenot applicablenot applicablenot applicableNot applicable — display-only

The linked workspace retains the complete native output and notes when available. Missing evidence is not a zero return, a successful empty result or a failed strategy.

Verdict and limitations

Not applicable — this page is not a backtest. Calibration remains unknown, and no trading edge follows from a forecast chart. The display-only restriction is preserved, with no votes or trade signals. A future empirical claim would need a dated, independently evaluated forecast artifact rather than a chart screenshot.

Sources and live research

INVENTORY D04. Sources: frontend/app/forecast/page.tsx, the HONESTY text in frontend/lib/timesfm-view.mjs, frontend/lib/timesfm-page.ts, frontend/components/timesfm/TimesfmBoards.tsx and backend/app/scoring/timesfm_forecast.py. Source-to-check: an empirical calibration/backtest artifact; none is established in the inventory.

Open this research tool in Backtesting

Not financial advice

Built for education. Backtests are historical; they do not predict future returns and are not a recommendation to buy or sell any security.

connecting…