Alpha Clio — Backtest Results Dashboard

Key Performance Metrics

Out-of-Sample Sharpe

0.95

2023–2026 holdout · 16 events

OOS Total Return

+10%

Locked 20% holdout period

Signal Hit Rate

49.5%

107 events · full backtest

Events Analyzed

365+

NFP 136 · CPI 216 · FOMC 13

OOS Max Drawdown

−2.2%

vs ≤ 15% Go/No-Go threshold

Cost Safety Margin

12×

Model: 10 bps · ZT actual: 0.79 bps

Equity Curve & Train / Test Split

Simulated Equity Curve

TLT × CPI events only — Phase 1 baseline · 5-min intraday window

Train vs Test Performance

80/20 locked holdout split

Central Methodological Finding

🔬 The 30-Hour Timing Lag Problem

The initial backtest measured returns at next-day market close, introducing a 30-hour gap between the 8:30 ET release and the return measurement point. During that window, unrelated market movements swamped the event signal — producing a near-zero Sharpe ratio (0.089). Redesigning to a 5-minute intraday window eliminated the confound and produced a methodologically sound signal framework.

❌ Before Fix

Signal: 08:30 ET release
Measure: next-day 16:00 ET
Gap: 30 hours of noise
Sharpe: 0.089

✅ After Fix

Signal: 08:30 ET release
Measure: 08:35 ET (5-min)
Gap: pure event window
OOS Sharpe: 0.95

Event & Signal Analysis

Sprint 1 — Event Coverage

Event	Tier	Releases	Windows
Non-Farm Payrolls	Tier 1	136	1d / 2d / 5d
CPI	Tier 1	216	1d / 2d / 5d
FOMC Rate Decision	Tier 1	13	1d / 2d / 5d
PCE Price Index	Tier 1	~144	1d / 2d / 5d
ISM PMI (Mfg + Svc)	Tier 2	~288	1d / 2d / 5d
Initial Jobless Claims	Tier 2	~884	1d / 2d / 5d
Housing / Durable / GDP	Tier 3	~108	1d / 2d / 5d

Train / Test Split Metrics

Metric	Train (80%)	Test (20%)
Period	2009–2022	2023–2026
Events	91	16
Sharpe Ratio	−0.085	0.95
Total Return	−5%	+10%
Max Drawdown	−19.2%	−2.2%
Status	In-Sample	OOS ✓

5-Instrument Portfolio

TLT

20-Year Treasury ETF

📌 Nominal rates / duration

~120 trades/yrPANIC: holds 100%

EUR/USD

Euro / Dollar FX

📌 Dollar strength signal

~184 trades/yrPANIC: 50% size

USD/JPY

Dollar / Yen FX

📌 Dollar + JPY carry hedge

~184 trades/yrSTRESS: 150%

Crude Oil Futures

📌 Energy / global growth

~40 trades/yrPANIC: paused

Gold Futures

📌 Real rates / inflation hedge

~44 trades/yrPANIC: holds 100%

VIX Regime Overlay

Position Size Multipliers by Regime

Regime	VIX Range	TLT	EUR/USD	USD/JPY	CL	GC	Logic
NORMAL	VIX < 20	1.0×	1.0×	1.0×	1.0×	1.0×	All instruments at standard size
ELEVATED	VIX 20–25	1.0×	1.0×	1.0×	1.0×	1.0×	Monitor only, no size change
STRESS	VIX 25–35	1.0×	1.0×	1.5×	1.0×	1.0×	JPY carry amplification
PANIC	VIX > 35	1.0×	0.5×	0×	0×	1.0×	Safe-havens hold; risk assets paused

Execution Cost Model

Normal Regime

10 bps

VIX < 20 · round-trip

12× safety margin vs ZT actual (0.79 bps)

Stress Regime

15 bps

VIX 25–35 · wider spreads

Conservative risk-off assumption

Panic Regime

20 bps

VIX > 35 · liquidity-adjusted

Stress-tested against flash crashes

Go / No-Go Validation Gate

Five criteria · all must pass before paper trading begins

Criterion	Threshold	OOS Result	Status
Newey-West Adjusted Sharpe (holdout)	≥ 0.50	0.95	✓ PASS
Maximum Drawdown (holdout)	≤ 15%	−2.2%	✓ PASS
Stress-Period Expectancy	≥ 0 bps	Pending Phase 5	⏳ PENDING
Single-Event P&L Concentration	≤ 40%	Pending Phase 5	⏳ PENDING
Signal Log Completeness	≥ 95% of expected events	Pending Phase 5	⏳ PENDING

Project Roadmap

Phases 0–7

✓ Phase 0 Event calendar · signal construction · consensus data collection
✓ Phase 1 Daily backtest · timing lag diagnosis · cost model validation (ZT tick data)
✓ Phase 2 Intraday backtest engine · asymmetric 3-rule exit system · regime classifier
✓ Phase 3 5-instrument portfolio · analytics module · factor concentration (PCA)
○ Phase 4 Data collection — Alpaca / OANDA fetch · Bloomberg import
○ Phase 5 Full intraday backtest across all 5 instruments · Go/No-Go gate
○ Phase 6 Paper trading on IBKR simulated environment
○ Phase 7 Live deployment — post paper-trading validation only