t / TURQUET TRADING
CROSS-ASSET OBSERVATORY

The market,
in perspective.

Explore price extremes, inspect historical evidence, and understand the assumptions.

Market radar

InstrumentLast / changeDetrended σATR / dayEvidence

Select an instrument to inspect its targets, daily risk and historical outcomes.

How to read this research

Extremity is not probability

Fit a linear trend to log prices from the preceding 150 sessions, excluding today’s observation. Center residuals at their median. Divide today’s out-of-sample residual by residual standard deviation (N−2). ±2σ is extended; ±3σ is extreme. No Gaussian tail probability is assumed.

Robust σ uses 1.4826 × median absolute deviation. The percentile compares today’s residual to the fitted historical residual distribution; it is descriptive, not a calibrated p-value.

Risk and confirmation

ATR(14) uses Wilder smoothing. Daily σ is the standard deviation of 150 daily log returns; price × σ is an approximate one-day price move. Neither measure is a guaranteed bound. RSI(14), one-day return shock and 20-day volatility provide context. “Turning” requires a close toward the center and declining absolute detrended σ.

What counts as a past reversal?

Match the same asset, extension side, 2–3σ or ≥3σ severity, and trend alignment. Use non-overlapping 20-session outcomes with next-open entries. The median target is frozen at the event; stop is 2 ATR from entry. Skip entries already beyond the target. If both barriers occur in one bar, count the stop first; adverse opening gaps fill at the open.

Target probability means reaching that target before the stop within 20 sessions. The interval is a 95% Wilson interval. Hit timing describes successful events only; failures and timeouts are separate.

Expected value and limitations

EV is mean directional return of the target/stop/20-day-timeout strategy, after the displayed round-trip cost assumption. 5/10/20-session returns instead hold to that horizon, without target or stop. Their 10th–90th percentiles are historical dispersion, not confidence intervals.

These are exploratory analogs, not held-out forecasts. Sparse samples, changing regimes and scanning many correlated markets limit reliability. “Historical support” requires ≥30 events, a target-rate lower bound above 50%, positive EV and a current turn; it still does not imply an imminent reversal. Financing, borrow and futures roll costs are excluded. Indexes are not directly tradable; futures quotes can contain roll discontinuities. FX is quote currency per base unit; NOK is covered through currency pairs.