Tessera Phase 8 — Oracle Strategy | 2023-01-01 → 2023-12-31 | BTC-USDT-PERP | v0.8 (Compounded)
3 Jan, 2023 - 31 Dec, 2023 (matched dates)

Benchmark: BTC BUY-AND-HOLD • Periods/Year: 252 • RF: 0.0% | Generated by QuantStats (v. 0.0.81)


2026-05-18T07:24:25.141953 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.188674 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.234537 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.276707 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.332361 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.369017 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.414133 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.458684 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.501087 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.539868 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.620458 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.657113 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.708417 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/
2026-05-18T07:24:25.763058 image/svg+xml Matplotlib v3.10.3, https://matplotlib.org/

📊 Tessera Evaluation Metrics

Statistical Significance

MetricValueDescription
Point-estimate SR (annualised) 1.147 √T-annualised μ/σ
Bootstrap 95% CI for SR [0.880, 1.419] Stationary block bootstrap (Politis & Romano 1994)
Probabilistic SR (PSR) 0.9158 P(true SR > 0 | data), accounts for skew/kurt
Deflated SR (DSR) 0.3968 PSR adjusted for 180 independent trials
Trial count (N) 180 Optuna trials + manual notebook experiments
SR std across trials (σ_SR) 0.5000 Cross-trial SR dispersion used in DSR
Return skewness / excess kurt 0.002 / -1.964 Used in PSR/DSR variance correction
Observations (T) 364 Return bars used for SR estimation

Interpretation: WEAK: DSR < 0.95 — insufficient evidence of skill after {trial_count} trials. Likely noise. PSR answers "is this SR genuine over this sample?"; DSR additionally penalises for the 180 configurations tried before picking this model. Bootstrap CI shows sampling uncertainty in the SR estimator. DSR < 0.75 = discard. DSR 0.75–0.95 = needs more data. DSR ≥ 0.95 = deploy-ready.

Stress-Window Performance

EventWindowIS/OOSData Total ReturnMax DDAnnualised SR
COVID Crash 2020-02-20 → 2020-03-13 IS none n/a n/a n/a
China Mining Ban 2021-05-12 → 2021-05-20 IS none n/a n/a n/a
LUNA Collapse 2022-05-08 → 2022-05-15 IS none n/a n/a n/a
FTX Collapse 2022-11-06 → 2022-11-12 IS none n/a n/a n/a
USDC Depeg 2023-03-10 → 2023-03-13 OOS full 0.29% -0.97% 1.01
Yen Carry Unwind 2024-08-02 → 2024-08-07 OOS none n/a n/a n/a

IS = in-sample (less meaningful). OOS = out-of-sample (credibility evidence). "none" coverage means the strategy backtest window does not overlap this event. OOS losses are expected during tail events and are not a reason to reject the strategy — they confirm the evaluation is honest.