From research paper to real strategy results
We continuously research, build, and validate systematic trading ideas with robust testing, then share transparent results so everyone can access evidence-based quant insights, not just institutions. We backtest quantitative trading strategies across global equity indices, options analysis on US index ETFs, portfolio optimization, sentiment tracking, and market regime detection - all with live data updated daily.
What's Here
Quantitative strategies, options analysis, portfolio optimization, and market research more| Strategy Library | 70+ technical indicator-based strategies backtested across 21 global equity indices with 5 years of daily data. Each strategy has a dedicated page with equity curves, performance metrics, parameter optimization, and Python implementation code. |
| Options Strategies | 18 options strategies — from cash-secured puts and covered calls to iron condors, jade lizards, and synthetic longs — backtested on SPY, QQQ, IWM, and DIA using live IV surfaces and Black-Scholes pricing. Each strategy shows per-symbol performance, equity curves, and implementation code. |
| Portfolio Optimization | 30 systematic portfolio strategies built on S&P 500 stocks using factor signals (momentum, volatility, quality, trend). Skfolio-powered optimization comparing Max Sharpe, Min Variance, Risk Parity, HRP, and more across US equity and global indices. |
| Parameter Optimization | Grid search over each strategy's key parameters, optimizing for Sharpe ratio, Sortino, Calmar, win rate, and profit factor. Results visualized as line charts (1-parameter) or scatter heatmaps (2-parameter) showing the full performance surface. |
| Market Analysis | Relative Rotation Graphs for US equity and sector ETFs vs S&P 500, correlation matrix across all strategies, stock sentiment tracker using VADER on financial news headlines, and market regime detection using GMM and hidden Markov models. |
| Project Research | Adaptive portfolio strategies (OLPS algorithms), K-means clustering portfolio construction, OLPS comparison, and market regime detection — each with interactive visualizations, methodology documentation, and source code. |
Top Performers
Best risk-adjusted returns across all strategies (ranked by Sharpe)Data as of 2026-05-25
Abbreviations
- Sh
- Sharpe ratio (risk-adjusted return)
- Ret
- Average total return
- MDD
- Maximum drawdown
- Win
- Win rate
- Vol
- Annualized volatility (portfolios)
- Sort
- Sortino ratio (portfolios)
- Calm
- Calmar ratio (portfolios)
Yield Curve Intelligence
This panel summarises US Treasury term-structure dynamics: inversion diagnostics on benchmark spreads, Nelson–Siegel level, slope, and curvature, and a macro risk regime that combines curve shape with equity drawdown context. It is a condensed view of the live monitor in our Yield Curve Intelligence study, where the full factor history, regime tables, and methodology are documented.
The figures below update from the latest available curve snapshot. The headline cards report the current 10Y–3M spread, the classified regime, and how often the curve has been inverted over the sample; the chart traces the same spreads through the most recent sessions so you can judge whether flattening is isolated or persistent.
Read spreads against zero: negative values mean short rates exceed long rates for that pair. A brief dip below zero is not, by itself, a recession signal; sustained inversion—especially when it aligns with risk-off equity conditions—has more often preceded macro slowdowns, though lead times and severity still vary by cycle.
Fundamental Stock Analysis
Dropdown-driven US large-cap fundamentals snapshotSelect a metric to compare the available US universe using the latest generated dataset.
View MoreRisk reports & stop / take-profit analysis
moreHow diversified equity and options backtests look in risk–return space, plus a controlled experiment on stops and profit targets. You are viewing the still-image summaries. An interactive stop / take-profit comparison may be added later; the figures above are complete on their own for learning risk–return and exit mechanics.
Equity index strategies
Each scatter dot is one strategy averaged across all index backtests (left). On the right, pick any strategy from the dropdown to see that same cross-index mean as an equity and drawdown path.
Each dot is one strategy, with Sharpe and max drawdown averaged across all 20 index backtests. Compare rules on a level playing field, not a single lucky market.
Use the dropdown to pick any strategy. Each path averages normalized equity and drawdown across every index backtest (same basis as the scatter).
Options on US index ETFs
54 options strategies on SPY, QQQ, IWM, and DIA (metrics averaged per strategy). Stacked vertical bar charts show win rate and avg return per trade separately (see chart caption). Bar and scatter colors use strategy category (Income, Directional, Spread, Volatility, Neutral, Hedge). All four charts use the same simulated exits on the underlying ETF (stop, target, or 45-DTE expiry).
One dot per strategy: mean Sharpe and drawdown across SPY, QQQ, IWM, and DIA. Metrics come from trades that can exit on underlying stop-loss / take-profit or at 45-DTE expiry (not hold-to-expiry only when risk controls are on).
Reading the chart
- Sample period: Aug 2024 – Feb 2026 (~18 months), 19 rolling 45-DTE cycles on SPY, QQQ, IWM, and DIA.
- Same axes as the equity scatter: drawdown (%) vs Sharpe, but each point averages four ETF backtests.
- Each dot’s Sharpe and drawdown reflect realized P&L after simulated SL/TP on the ETF path (see risk note on the chart when present).
Vertical bar charts for all 54 strategies (metrics averaged across SPY, QQQ, IWM, and DIA). Win rate and avg return per trade use the same risk-managed exits as the scatter (underlying SL/TP when enabled, else 45-DTE expiry).
Reading the chart
- Sample period: Aug 2024 – Feb 2026 (~18 months) with 19 rolling 45-DTE entries, rolled every ~21 trading days on SPY, QQQ, IWM, and DIA (see report JSON for exact dates).
- Two stacked charts (not one shared axis): top = win rate (%), bottom = average return per trade (%). Summing every trade’s PnL % can exceed 1,000% and is not plotted; tooltip shows that sum on request.
- Bars are not “expiry-only” returns when risk controls are active — early stops and targets change both win rate and average return per trade.
Counts trades by simulated exit rule: stop-loss and take-profit on the underlying ETF path, max-hold time, or hold to 45-DTE expiry. Requires options analysis run with risk controls enabled.
Reading the chart
- A single bar means every trade in the stored backtest exited at expiry — the JSON was built without underlying stop-loss / take-profit (re-run with --stop-loss-pct and --take-profit-pct).
- Multiple bars: compare how often stops cut losers vs targets bank gains vs holding through expiry.
- These are model exits on the index price path, not broker fill types (roll, assignment, etc.).
MAE = worst unrealized loss versus entry while the trade was open. MFE = best unrealized profit before exit. Each point is one completed trade, exited via the same underlying SL/TP / expiry rules as the other options charts.
Reading the chart
- Horizontal: MAE (%), vertical: MFE (%). Points toward the upper-right had room to move both ways intratrade.
- Color shows realized P&L (%) at the simulated exit bar (stop, target, or 45-DTE expiry) — not P&L as if every trade ran full term.
- Many reds with very negative MAE can mean stops were wide or the structure bled before exit rules fired.
Stop / target experiment (static figures)
These match the interactive charts below: one symbol, one horizon, many strategies, several stop and take-profit configurations. Use the PNGs for slides or print; use the live charts to hover exact numbers.
Compares several risk settings applied to the same basket of simple strategies on one index. Bars summarize averages across that basket.
Reading the chart
- This PNG may rescale series so all three fit visually; use the interactive chart on this page for raw numbers on one scale.
- Read across: baseline vs stop-only vs target-only vs combined tells you which rule family moves the average the most.
Only the stop under entry moves; profit targets are turned off for this experiment. The left end is “no fixed stop” in this setup.
Reading the chart
- Upward slope: the average strategy Sharpe rises when the stop is wider (often fewer whipsaws).
- Downward slope: tighter stops hurt the average Sharpe in this sample — cutting loss also cuts good trades.
Only the profit target above entry moves; stops are off for this sweep. The left end is “no fixed target” here.
Reading the chart
- Larger targets may help trend rules ride longer or hurt mean-reversion rules that mean-revert quickly.
- A flat line means this basket is insensitive to target choice in the tested range; a steep line means target choice matters.
For each stop setting, height is the average of each strategy’s worst drawdown (%). Lower is shallower on average.
Reading the chart
- Tighter stops do not automatically produce smaller average drawdown — churn can create new drawdown episodes.
- Compare to the Sharpe line chart for the same sweep: you want joint insight, not one number alone.
Same drawdown idea as the stop-only chart, but only profit targets change. Baseline is the first bar.
Reading the chart
- Compare bar heights across 5%, 10%, 15% targets vs baseline — sometimes wider targets smooth the path.
- Pair with the take-profit vs Sharpe line to reason about risk and return together.
Interactive stop / target lab
The still images above are enough for most readers. A live version of the sensitivity experiment (hover tooltips, exact averages) appears when that dataset ships with the site.
Site maintainers
Generate sl-tp-sensitivity.json with npm run data:sl-tp-sensitivity, or enable SKIP_SL_TP_SENSITIVITY=1 only when you need to skip it. Expect a long run; it is optional for publishing.
Loading strategies...
Ranked snapshot of strategy performance metrics.
No fundamental strategy dataset found yet. Run npm run data:fundamental-strategies.
Loading options strategies…
Loading portfolio strategies…
| 1. | Signal Generation Framework | Modular framework for generating trading signals from multiple data sources. Supports technical indicators, fundamental analysis, machine learning models, and custom signal generators. Includes signal validation, filtering, and combination logic to create robust entry and exit signals. |
| 2. | Execution Engine | High-performance order execution framework with support for multiple broker APIs and exchange connections. Features include order routing, slippage management, fill simulation, and real-time position tracking. Designed for both backtesting and live trading environments. |
| 3. | Risk Management System | Comprehensive risk management framework with position sizing, portfolio-level risk limits, drawdown controls, and real-time risk monitoring. Implements VaR calculations, position concentration limits, and dynamic risk adjustment based on market conditions and strategy performance. |
| 4. | Portfolio Management | Advanced portfolio construction and optimization framework. Supports multiple optimization objectives including Sharpe ratio maximization, risk parity, minimum variance, and custom utility functions. Includes rebalancing logic, transaction cost modeling, and constraint handling. |
| 5. | Data Pipeline & Processing | Scalable data ingestion and processing framework for market data, fundamental data, and alternative data sources. Features real-time data streaming, historical data management, data quality checks, and normalization. Supports multiple data formats and timeframes. |
| 6. | Backtesting Infrastructure | Robust backtesting framework with realistic market simulation, including bid-ask spreads, market impact, and partial fills. Supports walk-forward analysis, Monte Carlo simulation, and out-of-sample testing. Provides detailed performance metrics and attribution analysis. |
| 7. | Strategy Orchestration | Framework for managing multiple strategies simultaneously with resource allocation, priority queuing, and conflict resolution. Includes strategy lifecycle management, performance monitoring, and automated strategy deployment and retirement mechanisms. |
| 8. | Monitoring & Alerting | Real-time monitoring framework for system health, strategy performance, and market conditions. Features customizable alerts, performance dashboards, and automated reporting. Includes anomaly detection and automated response mechanisms for critical events. |
Clustering Portfolio Analytics
moreS&P 500 names are grouped with K-means on scaled risk/return features—annual return, volatility, Sharpe ratio, beta, correlation, and related inputs—so each cluster captures a distinct cross-sectional risk profile. Portfolios are built by equal-weighting the highest-Sharpe names from every cluster, with a separate top-20 Sharpe basket and the S&P 500 index as benchmarks.
Features are estimated on the training window and cluster assignments are frozen before validation. The elbow method on within-cluster sum of squares guides the choice of K; the live run uses K = 4 clusters across roughly 200 liquid constituents, then scores out-of-sample cumulative wealth on the held-out validation dates below.
The panels summarize universe setup, per-cluster training statistics, a return–volatility map (hover any point for the ticker), mean Sharpe by cluster, and validation paths for the three portfolio definitions. Full symbol lists, feature-correlation detail, and methodology are on the Diversified Stock Portfolio Clustering project page.
Adaptive Portfolios Analytics
moreOnline portfolio selection (OLPS) updates weights each period from past prices alone—no forward-looking labels—making it a natural test bed for adaptive allocation under regime change. This study benchmarks fourteen published and baseline rules (momentum, mean-reversion, and pattern-learning families) on a diversified six-ETF sleeve spanning US equity, international equity, emerging markets, Treasuries, inflation-linked bonds, and REITs (VTI, EFA, EEM, TLT, TIP, VNQ).
Each algorithm is estimated on 2015–2022 daily closes and evaluated out-of-sample on 2023–2024 with daily rebalancing, zero look-ahead, and wealth indices reset to 1.0 at the test start. SPY and an equal-weight universe portfolio (UFR) anchor absolute performance; a 0.1% per-trade fee variant is tracked in the full project for turnover-sensitive strategies.
The panels below summarize universe setup, leading test-period strategies, equity paths for the top in-sample Sharpe selectors, a cross-sectional risk–return map, average Sharpe by algorithm family, and the full metrics table. For methodology, fee stress tests, and return distributions, see the Adaptive Portfolio Strategies project page.
| 1. | Diversified Stock Portfolio Using Clustering Analysis S&P 500 portfolio construction using K-means clustering on risk/return features (correlation, beta, returns, volatility, Sharpe ratio). Backtested vs index. | Active | 2024 | Python, K-means, backtesting |
| 2. | Relative Rotation Graph (RRG) — US Equity Dynamic RRG for US stocks vs S&P 500. JdK RS-Ratio and RS-Momentum with animation. | Active | 2025 | Python, yfinance, Recharts |
| 3. | Adaptive Portfolio Strategies: Sequential Allocation Methods Comprehensive analysis of 14 sequential portfolio allocation strategies on diversified ETF portfolio. Includes momentum-based, reversion-based, and pattern-learning approaches with transaction cost analysis. | Active | 2025 | Python, sequential optimization, backtesting |
| 4. | Optimal Execution with RL Agent (DQN) Deep Q-Learning execution agent for slicing large orders under microstructure-style market impact. Compared against TWAP, passive, aggressive, and random baselines. | Active | 2026 | Python, Gymnasium, Stable-Baselines3, execution simulation |
| 5. | Fundamental Stock Analysis US large-cap fundamental analytics dashboard with deep cross-sectional ranking, distribution diagnostics, sector structure, risk/return mapping, and multi-factor composite interpretation. | Active | 2026 | Python, fundamental screening, interactive charts |
| 6. | Market Regime Detection Using Gaussian Models Comprehensive market regime identification across 21 global indices using Gaussian Mixture Models (GMM) and Greedy Gaussian Segmentation (GSS). Detects bull, bear, and transition regimes for adaptive portfolio management. | Active | 2025 | Python, GMM, GSS, regime detection |
| 7. | Stock Sentiment Tracker US equity sentiment tracker using VADER lexical analysis on financial news headlines. Tracks prices, sentiment scores, and correlations for S&P 500 stocks. Updated daily. | Active | 2025 | Python, VADER, NLP, financial news APIs |
| 8. | Yield Curve Intelligence Yield-curve research project built from yfinance market data: Nelson-Siegel fitting, slope/inversion diagnostics, and regime-aware risk context. | Active | 2026 | Python, yfinance, Nelson-Siegel, macro risk signals |
| 9. | Fama–French Factor Lab Unified factor dashboard: US FF5 and Carhart momentum, multi-asset ETF panel, international sleeve comparison, volatility regimes, risk budgeting, sleeve attribution, SPY research grid, and q-factor extension notes. Refreshed from Ken French + yfinance. | Active | 2026 | Python, pandas, linear factor models, Ken French data, yfinance |
| 10. | Smart Beta: Factor-Tilted Portfolio Construction One-year rolling study on a sector-balanced U.S. equity panel: equal weight, min variance, value tilt, and value–momentum blend with full risk and factor analytics. | Active | 2026 | Factor models, portfolio construction, risk analytics |
| 11. | Cross-section shrinkage lab (ETF panel) Yahoo Finance ETF panel: PCA spectrum, cross-sectional R² vs K, pseudo-OOS folds, ridge diagnostics. JSON from npm run data:shrinking-cross-section. | Active | 2026 | Python, Recharts, PCA, shrinkage, cross-sectional R² |
| 1. | Adaptive Portfolio Optimization Under Non-Stationary Market Conditions | Journal of Quantitative Finance | 2024 | Novel approach to dynamic asset allocation using regime-switching models and reinforcement learning. |
| 2. | High-Frequency Data Processing: Computational Frameworks | Computational Economics | 2024 | Scalable architectures for real-time processing of tick-level financial data. |
| 3. | Robust Risk Measures in Fat-Tailed Distributions | Risk Analysis | 2023 | Comparative study of tail risk metrics under various distributional assumptions. |
| 1. | Statistical Modeling | Advanced regression analysis, time series forecasting, and multivariate statistical methods for complex data patterns. |
| 2. | Strategy Comparator | Compare up to 4 strategies side-by-side using Sharpe, return, drawdown, win rate, and trade count. |
| 3. | Stock Sentiment Tracker | US equity only: live prices and news sentiment (VADER) for major US stocks. See Sentiment page. |
| 4. | Risk Analytics | Comprehensive risk assessment frameworks including VaR, CVaR, and stress testing methodologies. |
| 5. | Data Pipeline | High-performance data processing infrastructure designed for large-scale quantitative analysis. |
| 6. | HFT Latency Budget Calculator | Plan per-stage p99.9 tick-to-trade budgets and detect bottlenecks before live deployment. |
| 7. | Portfolio Optimization | Modern portfolio theory implementation with multi-objective optimization and constraint handling. |
