Skip to main content
Philosophy Background

Methodology

Our
Approach

Quantitative investment models built on academic research and empirical evidence. We combine scientific rigor with practical applicability for systematic, evidence-based investing.

What We Do

We translate peer-reviewed financial research into working investment tools. Each model starts as a published finding in economics or finance, gets stress-tested against decades of market data, and only reaches production if it survives a validation process designed to eliminate overfitting. The result is a growing library of systematic models that cover risk assessment, macro regime detection, portfolio construction, and tactical allocation.

Most of our quantitative models are open source on TradingView. The portfolio models run as automated daily pipelines and publish their current allocations on live dashboards. We charge only for strategies that have passed extended walk-forward validation and demonstrate consistent out-of-sample performance.

Process

How We Build Models

Every model follows the same development pipeline. The process typically takes 4 to 12 weeks per model, and most research ideas do not survive it.

Phase 1

Literature Review

We start with a specific question about market behavior and search SSRN, NBER, and major finance journals for existing empirical evidence. If no credible research supports the hypothesis, we stop here. We do not invent signals without theoretical grounding.

Phase 2

Signal Construction

The published concept gets translated into a computable signal using publicly available data (FRED, TradingView, CFTC, EIA). We deliberately limit degrees of freedom: fewer parameters means less room for the model to memorize historical patterns instead of learning genuine relationships.

Phase 3

Initial Screening

Before any optimization, we run the signal with default parameters across the full available history. If the raw signal shows no edge at this stage, we discard it. This prevents the common trap of optimizing a fundamentally broken idea until it looks good on paper.

Phase 4

Robustness Testing

The model must pass permutation tests (randomized signal timing to establish statistical significance), parameter sensitivity analysis (performance should not collapse with small parameter changes), and noise injection (adding random perturbation to inputs to test fragility).

Phase 5

Walk-Forward Validation

We split the data into sequential training and test windows. Parameters are fitted on one segment and tested on the next, simulating the experience of a live investor. Combinatorial purged cross-validation (CPCV) extends this across many non-overlapping splits to reduce dependence on any single boundary.

Phase 6

Live Monitoring

Models that pass all tests go into production with daily automated updates. We track live performance against backtest expectations and flag deviations. If a model degrades beyond defined thresholds or its underlying data source changes materially, it gets retired or rebuilt.

Validation

What a Model Must Survive

Minimum requirements. Models that fail any single gate get discarded regardless of backtest performance.

1

Permutation
p < 0.05

Signal timing must beat 95% of randomized permutations.

2

Parameter
Stability

Performance plateau across neighboring parameters, not a spike.

3

CPCV
Sharpe > 0

Positive risk-adjusted return across synthetic OOS paths.

4

Noise
Injection

Graceful degradation at 10% input variance, no collapse.

5

Regime
Consistency

Works in expansion, contraction, high-vol, and low-vol.

Fail any one and the model is discarded

Values

Core Principles

Six fundamental principles guiding our model development and investment philosophy.

Evidence-Based

Every model is grounded in peer-reviewed academic research from leading financial economists. We implement only theories validated through rigorous empirical testing.

Risk-Focused

Risk management stands at the center of our approach. We prioritize downside protection and capital preservation over maximum returns.

Transparent

Free models are open source with full documentation. Professional models are proprietary with proven, backtested performance as our core value proposition.

Long-Term Oriented

We optimize for sustainable investment success rather than short-term gains. Our models identify structural trends, not fleeting price movements.

Accessible

Sophisticated quantitative analysis should not be exclusive to institutions. We democratize professional tools through fair pricing.

Realistic

Models are decision-support tools, not prediction machines. We build robustness and acknowledge limitations rather than promise false precision.

Foundation

Research & Data

Every model traces back to published empirical work and publicly verifiable data. No proprietary feeds, no black boxes.

Research Domains

  • Asset Pricing & Portfolio Theory
  • Behavioral Finance & Market Psychology
  • Monetary Economics & Central Banking
  • Market Microstructure & Fund Flows
  • International Finance & Exchange Rates
  • Commodity Economics & Supply-Demand
  • Volatility Modeling & Risk Premia
  • Credit Markets & Financial Conditions
  • Momentum & Factor Investing
  • Systematic Trading & Quantitative Strategies

Data Sources

Differentiation

What Sets Us Apart

Evidence Over Emotion

Markets are driven by psychology, but successful investing requires discipline. Our models provide objective, systematic signals based on data, helping you maintain a rational approach during market turbulence.

Education Over Marketing

We reject the exploitation of investors through overpriced seminars and false promises. Our models come with comprehensive documentation explaining theory, usage, and realistic expectations.

Accessibility Over Exclusivity

Professional quantitative tools should not cost thousands per month. We provide institutional-quality models at fair prices through the TradingView platform.

Get Started

Discover
Our Models

25+ free quantitative models on TradingView. 7 portfolio strategies with daily-updated dashboards.