## 📊 Strategy Overview * **Asset / Instrument:** S&P 500 ETF (e.g., SPY or IVV) * **Timeframe:** Daily ($1\text{-day}$ close bars) * **Strategy Type:** Systemic, trend-following, long-only regime model * **Execution Frequency:** Low (infrequent rebalancing upon signal trigger) ## 🧠 Core Trading Hypothesis An asset's long-term upward trend creates sufficient price momentum to compensate for signal lag and false-breakout whipsaws. By capturing major bull markets and exiting during severe bear markets, the strategy aims to achieve higher risk-adjusted returns (or lower maximum drawdowns) relative to a buy-and-hold benchmark on the S&P 500. ## 📥 Entry Rules A **Long Entry** is generated when the fast moving average crosses above the slow moving average on the daily close. * **Trigger Condition:** $\text{MA}_{\text{fast}}[t-1] \le \text{MA}_{\text{slow}}[t-1]$ AND $\text{MA}_{\text{fast}}[t] > \text{MA}_{\text{slow}}[t]$ * **Execution Time:** Market Open on bar $t+1$ (assuming next-day execution to avoid lookahead bias). * **Execution Price:** $\text{Open}[t+1]$ ## 📤 Exit Rules A **Long Exit** is generated when the fast moving average crosses back below the slow moving average on the daily close. * **Trigger Condition:** $\text{MA}_{\text{fast}}[t-1] \ge \text{MA}_{\text{slow}}[t-1]$ AND $\text{MA}_{\text{fast}}[t] < \text{MA}_{\text{slow}}[t]$ * **Execution Time:** Market Open on bar $t+1$. * **Execution Price:** $\text{Open}[t+1]$ ## ⚙️ Variables & Parameters | Variable Name | Type | Baseline Value | Parameter Sweep / Alternative Options | | --- | --- | --- | --- | | `Fast_MA_Period` | Integer | $50$ | $10, 20, 50, 100$ | | `Slow_MA_Period` | Integer | $200$ | $100, 150, 200$ | | `MA_Type` | Categorical | Simple Moving Average (SMA) | Exponential (EMA), Weighted (WMA) | | `Transaction_Fee` | Float | $0.00$ USD/share | Commission rates per broker ($0.0005\text{--}0.005$/share) | | `Slippage_Pcnt` | Float | $0.02\%$ per trade | $0.00\%\text{--}0.10\%$ | | `Max_Portfolio_Risk` | Float | $2\%$ | Configured via sizing or hard stops | ## 🛡️ Risk & Position Management * **Position Limits:** Maximum 1 open position at a time (no stacking or scaling). * **Leverage:** $1.0\times$ maximum (no margin, long-only). * **Position Sizing:** Allocate portfolio equity based on specified max risk parameters. * *Option A (All-In Trend Filter):* Allocate $100\%$ of unencumbered cash upon signal. The $2\%$ account risk rule requires a defined stop distance $S_{\text{dist}}$ (e.g., $2\%$ loss limit from entry price or trailing stop). * *Option B (Fixed Volatility Sizing):* Position Size $= (\text{Equity} \times 0.02) / (\text{Vol}_{\text{ATR}})$, capped at $100\%$ equity. * **Stop-Loss / Take-Profit:** None explicitly supplied in core concept. Pure crossover rules rely on the opposing cross as the stop. *(Assumed for testing: Evaluate baseline without hard stops vs. adding an $X\%$ trailing stop or ATR stop).* ## 🧪 Backtest Scenarios ``` ┌───────────────────────────────────┐ │ Strategy Optimization │ └─────────────────┬─────────────────┘ │ ┌───────────────────────────────┼───────────────────────────────┐ ▼ ▼ ▼ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │ Baseline Model │ │ Parameter Sweep │ │ Friction & │ │ (50/200 SMA) │ │ (20/100, 10/200)│ │ Stress Testing │ └─────────────────┘ └─────────────────┘ └─────────────────┘ ``` 1. **Baseline Scenario:** $50\text{-day}$ SMA crossing $200\text{-day}$ SMA. $100\%$ equity allocation, zero slippage, zero commission. 2. **Conservative Scenario:** Include $0.05\%$ slippage + commission. Implement a $2\%$ hard stop-loss per position or ATR-based exit to conform strictly to the risk constraint. 3. **Aggressive Scenario:** Test Exponential Moving Averages (EMA) with faster periods ($20\text{-day} / 100\text{-day}$) to reduce signal lag during market turns. 4. **Stress-Test Scenario:** High-friction modeling ($0.15\%$ slippage per trade) combined with extreme gap-down execution simulation. ## 🌍 Market Regime Tests To validate robustness across varying market structural states, evaluate performance separately within the following isolated date windows: * **Bull Regimes:** Sustained macro upswings (e.g., 2012–2019, 2020–2021). * **Bear Regimes:** Severe market drawdowns (e.g., 2000–2002 Dot-Com Crash, 2007–2008 GFC, 2022 Fed tightening). * **Sideways / Consolidation Regimes:** Volatilities without clear directional momentum (e.g., 2015–2016 range-bound market, 2023 choppy periods). ## 📈 Evaluation Metrics The backtest module must output the following non-fabricated performance parameters: ``` Performance Dashboard ┌───────────────────────┬───────────────────────┬────────────────────────┐ │ Absolute Return Metrics │ Risk & Drawdown │ Trade & Expectancy │ ├───────────────────────┼───────────────────────┼────────────────────────┤ │ • CAGR │ • Max Drawdown (%) │ • Total Trades Count │ │ • Total Net Return (%)│ • Max DD Duration │ • Win Rate (%) │ │ • Benchmark Relative │ • Sharpe Ratio │ • Profit Factor │ │ Alpha / Beta │ • Sortino Ratio │ • Expectancy ($ / %) │ └───────────────────────┴───────────────────────┴────────────────────────┘ ``` * **Absolute & Adjusted Return:** Compound Annual Growth Rate (CAGR), Total Net Return (%). * **Risk & Drawdown:** Maximum Drawdown (MDD %), Maximum Drawdown Duration (days), Annualized Volatility. * **Risk-Adjusted Ratios:** Sharpe Ratio (using 0% or prevailing risk-free rate), Sortino Ratio, Calmar Ratio. * **Trade Dynamics:** Total Trade Count, Win Rate (%), Profit Factor ($\text{Gross Profit} / \text{Gross Loss}$), Average Trade Duration, Expectancy per Trade. ## ⚠️ Biases, Risks & Assumptions * **Data Leakage & Lookahead Bias:** Calculations must strictly use close data from bar $t$ to execute at open of bar $t+1$. * **Survivorship Bias:** Must run testing using historical index constituents or survivor-bias-free ETF historical series (accounting for dividends and stock splits via total-return series). * **Slippage & Gap Risk:** Moving-average systems suffer severe execution losses during overnight gap-down exits (e.g., exiting below the 200 SMA on a flash crash open). * **Whipsaw Risk:** In sideways markets, moving-average crossovers produce repeated false signals, leading to consecutive losses (capital erosion). * **Unstated Assumptions To Validate:** * Cash reinvestment yield during out-of-market periods (assumed $0\%$ risk-free return). * Treatment of ETF dividend yield (assumed re-invested total return index). ## ✅ Backtest Checklist * [ ] Verify total-return data is adjusted for dividends and split events. * [ ] Ensure point-in-time signal generation ($t$) executes on next-bar open ($t+1$). * [ ] Apply fixed transaction costs and variable slippage models. * [ ] Partition dataset into In-Sample (training/parameter search) and Out-of-Sample (validation) periods. * [ ] Run parameter sensitivity sweeps across variable grids to map performance stability plateaus. * [ ] Validate risk controls (ensure maximum portfolio risk stays within the $2\%$ threshold). ## 🎯 Decision Criteria Abandon or reject the strategy if any of the following occur during testing: * **In-Sample vs. Out-of-Sample Decay:** Out-of-sample Sharpe ratio degrades by $> 50\%$ compared to the in-sample period. * **Benchmark Underperformance:** CAGR fails to outperform a basic S&P 500 Buy-and-Hold strategy while displaying higher maximum drawdown. * **Whipsaw Vulnerability:** Consecutive losing trades in range-bound regimes exhaust $> 15\%$ of total portfolio equity. * **Low Trade Sample Size:** Fewer than 30 total trades across a multi-decade backtest, rendering metrics statistically insignificant.