Executive Summary

ProjectAlpha is a systematic, rule-based long-only strategy designed as a quantitative alternative to passive buy-and-hold on the S&P 500. The objective is to strictly preserve upside exposure while mitigating downside impact by stepping aside during statistically significant risk windows.

Methodology

Core Posture: Default long exposure to S&P 500. Cash/Flat positioning only during specific risk triggers.
Signal Generation: Multi-factor assessment based on fixed statistical thresholds (Price, Log-Returns, Realized Volatility).
Execution: End-of-day rebalancing. No intraday execution. Low turnover frequency.

Key Characteristics

Transparent Logic: Rules are pre-set, stable, and deterministic. No "black box" elements.
Robustness: Fixed parameter set designed to be all-weather, avoiding curve-fitting.
Quant alternative: Enhances risk-adjusted returns vs. benchmark.

Constraints & Scope

No Macro/Fundamental Inputs
No Short-Selling
No Leverage
No Derivatives
No Predictive Modeling / ML

Structural Advantages

Drawdown Mitigation: Systematically avoids high-volatility crash periods.
Consistency: Minimal model drift due to fixed parameter architecture.
Efficiency: Low operational drag; easily replicable via standard ETFs.

Risk Factors

Rebound Lag: Potential for delayed re-entry during V-shaped recoveries.
Risk Premiums: Dependent on statistical persistence of volatility clusters.
Investment Horizon: Optimized for long-term compounding cycles.

Cumulative Return

Strategy --

Buy & Hold --

CAGR

Strategy --

Buy & Hold --

Annualized Volatility

Strategy --

Buy & Hold --

Days Out of Market

Strategy --

Buy & Hold --

Annualized Return

Strategy --

Buy & Hold --

Sharpe Ratio

Strategy --

Buy & Hold --

Max Drawdown

Strategy --

Buy & Hold --

Win Rate

Initializing Visualization...

Month-by-Month Returns

Year	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec	Mean	Total	Benchmark	Alpha

Efficiency Analysis (Out-of-Market Performance)

Total Days Out

Avoided Crash Days

Avg Avoided Return --

Missed Rally Days

Avg Missed Return --

Defensive Ratio

(Avoided / Total Days)

Recent Trades

Entry Date	Exit Date	Entry Price	Exit Price	Return

Frequently Asked Questions

How does the multi-layer approach work? +

The strategy applies a multi-timeframe, three-layer tree logic.

Layer 1 — Regime Context: A slower-timeframe read classifies broad market conditions (e.g., risk-on, neutral, risk-aware). This gates how strict the downstream checks must be.
Layer 2 — Risk Filters: On a daily basis, a subset of indicators evaluates statistical risk conditions. Thresholds are fixed and interpreted conditional on the Layer-1 regime (i.e., tightened or relaxed depending on the backdrop).
Layer 3 — Execution: Entry/exit is triggered only when the required combination of Layer-2 conditions is met. Some nodes act as hard filters, others contribute to a composite risk score. The system is long-biased by design and steps aside temporarily when risk is statistically elevated.

This structure ensures that signals are contextualized, reduces false positives in benign regimes, and prioritizes protection when stress builds.

Short Answer: Three-layer, multi-timeframe tree: regime context gates strictness → fixed-threshold risk filters → execution that briefly goes flat from a long bias when risk is elevated.

What data do you use and how frequently do you trade? +

Simplicity and repeatability require transparent inputs and modest turnover.

Inputs: index closing prices, derived log-returns, and realized volatility (all public EOD).
Process: signals computed after close; no intraday data or timing.
Cadence: ~2–3 trades/month on average; flat windows are brief unless risk persists.
Replication: no leverage/derivatives; implement with standard S&P 500 trackers (vehicle-agnostic).

Edge cases / Limitations: Holidays and data revisions can shift timestamps; ETF/index tracking differences and slippage apply.

Short Answer: Daily end-of-day public data; typical cadence ~2–3 trades per month; exit windows usually last a few days.

Why these thresholds/levels and weights? +

Context: Thresholds must be stable across regimes and practical under trading frictions.

Regime-robustness: validated across bears, volatility shocks, and benign markets.
Distributional anchoring: thresholds tied to long-horizon percentiles / vol-normalized features of log-returns and realized volatility.
Trade-off discipline: prioritized path quality (Max DD, time underwater, Calmar) over headline CAGR alone.
Parsimony & correlation-aware selection: minimize redundancy and overfitting.
Friction-aware: tested with realistic costs/slippage at ~2–3 trades/month.
Governance: once set, parameters are locked; any change is versioned and prospective only.

Edge cases / Limitations: Not optimized for any single crisis; can underperform in long, “clean” bull runs.

Short Answer: Set via distribution-aware, regime-robust testing to balance drawdown mitigation vs whipsaws, anchored to long-horizon percentiles/vol-normalized features and costs; then fixed for consistency.

What types of indicators are used, and how many? +

We blend classical indicators with proprietary transforms, all computed on daily end-of-day public data:

Trend/Momentum & Breadth proxies (generic, period-agnostic formulations).
Volatility & dispersion measures based on realized volatility and distributional properties of returns.
Mean-reversion risk flags and price-based anomaly checks.
Proprietary signal transforms derived from log-returns and volatility, including normalization, rolling z-scores/percentile ranks, and regime-conditioned thresholds.

No machine learning, no predictive modeling, no intraday inputs.

Short Answer: ~8 total across trend/breadth, volatility/dispersion, mean-reversion/anomaly, plus proprietary transforms on log-returns & realized volatility; daily EOD, no ML/intraday.

How many indicators are there and how are combined? +

Approximately eight indicators are maintained in the library, but they are not always used simultaneously:

Each layer employs a different subset; importance is unequal across indicators.
Some indicators are binary gates (must-pass conditions); others feed a composite risk score with fixed weights.
The combination aims to minimize redundancy (correlation-aware) and balance Type-I vs. Type-II errors (avoiding whipsaws vs. catching stress early).

All weights, roles, and thresholds were locked after extensive robustness testing and are not re-optimized in live use.

Short Answer: ~8 indicators; different subsets per layer, unequal importance—some binary gates, others feed a fixed-weight composite; locked post-robustness tests (no live re-optimization).

What should I expect vs. buy-and-hold over the long term? +

Intended effect: reduce exposure during statistically risky windows, then re-enter promptly.
Potential benefit: shallower drawdowns, improved Calmar, more consistent rolling risk-adjusted stats.
Best case: avoids portions of major drawdowns → potential outperformance vs static B&H over long horizons.
Worst case: missed upside if rebounds occur while risk signals remain active.
Neutral/benign markets: may lag during persistent “clean” bull runs.
Horizon: designed for long-term use; no guarantee of drawdown reduction in every episode.

Edge cases / Limitations: Backtests are hypothetical; live results depend on costs, tracking, and execution.

Short Answer: Seeks shallower, shorter drawdowns and a smoother path; long-term outperformance is possible if major stress periods are partly avoided; worst case, missed portions of rapid rebounds.