Methodology & Performance Disclosure
JEDI AI is a product of Alethiq Labs LLC, a Washington limited liability company.
How JEDI algorithms are built, tested, and evaluated. Full transparency on backtesting assumptions, synthetic data modeling, execution mechanics, and risk disclosures.
This methodology underpins all signal-based allocations and performance reporting across the platform.
1. Backtesting Framework
All JEDI algorithms are evaluated using a walk-forward backtesting engine that processes historical market data day by day, simulating the exact decision logic used in live trading. Backtests are generated using the same signal engine that produces live algorithm signals — there is no separate backtest-only model. Position sizing, regime classification, and strategy voting all operate identically in simulation and production.
2. Synthetic ETF Modeling
Leveraged ETFs such as TQQQ and SQQQ were launched in 2010. To evaluate algorithm behavior across full market cycles — including the dot-com crash (2000–2002) and the global financial crisis (2008) — JEDI reconstructs historical leveraged ETF behavior for pre-2010 periods using underlying index data and standard leverage assumptions. Synthetic prices are modeled as daily-rebalanced leveraged exposure to the underlying index, incorporating ETF expense ratios and daily compounding effects. Post-2010 backtests use actual historical ETF prices.
3. Execution Assumptions
All backtests assume the following execution model:
- Order type: Market-on-Close (MOC) orders submitted before the closing auction
- Fill price: Official closing price, matching the price used in backtesting
- Slippage: 0.05% per trade applied to all simulated executions
- Commissions: Commission-free execution assumed (consistent with supported brokerage integrations)
- Timing: Signals are generated and orders submitted at 3:45 PM ET daily
Actual execution may differ due to market conditions, liquidity, and auction dynamics.
4. Performance Metrics
JEDI reports the following metrics for algorithm evaluation:
Primary Metrics
- CAGR — Compound Annual Growth Rate over the full backtest period
- Sharpe Ratio — Risk-adjusted return (annualized, using risk-free rate)
- Max Drawdown — Largest peak-to-trough decline during the backtest
Supplementary Metrics
- Calmar Ratio — CAGR divided by maximum drawdown
- Win Rate — Percentage of trades that closed with a positive return
- Profit Factor — Gross profits divided by gross losses
- Alpha — Excess return over the underlying index (in percentage points)
5. Risk Disclosures
Backtested results are not actual trading results. All performance figures are hypothetical and provided for informational purposes only. They are derived from historical simulations and are subject to the limitations inherent in any backtest. Backtests benefit from hindsight and may not account for all real-world factors such as liquidity constraints, market impact, or extreme events without historical precedent. Leveraged ETFs are subject to volatility decay, tracking error, and compounding effects that may result in returns significantly different from the leveraged return of the underlying index over periods longer than one day. Past performance does not guarantee future results.
6. Algorithm-Specific Documentation
Each algorithm has its own detail page covering strategy architecture, regime classification, risk controls, and backtest results: