Experiment

Smart Wealth Investment Process

The 4 Steps to Smarter Investment Decisions

For over two decades, Smart Wealth has developed a sophisticated investment process that combines mathematical forecasting models with artificial intelligence to optimize investment portfolios.

1. Security Selection
~5,000 → ~100 securities

→

2. Factor Selection
~2,000 → ~20 factors

→

3. Forecasting
Mathematical models

→

4. Optimization
Genetic algorithms

                        Key Innovation: Each investment instrument uses its own separate forecasting model, enabling exceptional results in highly dynamic markets.
                    

🧠 AI-Powered

Neural networks and machine learning algorithms

📊 Multi-Model Approach

Linear and non-linear mathematical models

🔄 Fully Automated

From analysis to execution and reporting

⚡ Real-Time

Continuous market monitoring and adjustment

Security Selection Process

We start with approximately 5,000 securities from global exchanges and systematically narrow them down to about 100 optimal investments through a rigorous 4-stage filtering process.

5,000

Starting Universe

2,500

After Economic Criteria

1,000

After Quantitative Analysis

100

Final Optimal Universe

1a. Economic Criteria: Select securities matching client specifications (region, asset class, sector, market cap, currency)

1b. Quantitative Criteria: Data availability check, correlation analysis, and outlier removal

1c. Statistical Analysis: Cluster analysis to group similar securities, then Principal Component Analysis to select representatives

1d. Liquidity Check: Ensure bid-ask spreads are below 0.25% threshold

                        Quality Assurance: Only securities listed on regulated exchanges for 5+ years and electronically tradeable are considered.
                    

Factor Selection Process

Financial markets are highly interconnected. We analyze thousands of market factors to identify the optimal set that drives price movements, reducing 2,000+ factors to approximately 20 key inputs.

                        Intermarket Analysis: Understanding how stocks, bonds, commodities, and currencies influence each other - because markets never move in isolation.
                    

2,000

Raw Financial Time Series

800

After Expert Knowledge

350

After Data Transformation

Optimal Factor Set

2a. Expert Knowledge: Apply economic theory and decades of experience to pre-select relevant factors

2b. Data Transformation: Clean trends, noise, and outliers using moving averages, oscillators, and volatility filters

2c. Statistical Testing: Rigorous causality, stationarity, and correlation analyses to avoid spurious relationships

2d. Genetic Algorithms: Evolutionary selection process to identify factors with the best forecasting properties

📈 Macroeconomic

Inflation, unemployment, consumer confidence

💼 Fundamental

P/E ratios, share prices, trading volume

🌍 Cross-Market

Currencies, commodities, bond prices

🧬 Evolutionary

Genetic algorithms for optimal factor combinations

Mathematical Forecasting

We deploy both linear and non-linear mathematical models to generate precise forecasts, with rigorous testing to ensure only the highest-quality predictions reach our portfolios.

📊 Linear Models

VARX, GARCH, SARIMA, Factor Models

🧠 Non-Linear Models

Multi-layer Perceptron (Neural Networks)

🔬 Rigorous Testing

In-sample/Out-sample validation

📈 Performance Metrics

Hit rate, Sharpe ratio, Maximum drawdown

3a. Model Creation: Deploy multiple mathematical models processing both linear and non-linear dependencies

3b. Quality Evaluation: Comprehensive statistical testing including hit rates, Sharpe ratios, and stability analysis

                        Model Lifecycle: Every model goes through In-Sample training → Out-of-Sample testing → Forward testing → Live deployment. Only models passing all phases are used for client portfolios.
                    

In-Sample
Model Training

→

Out-of-Sample
Unknown Data Testing

→

Forward
Live Testing

→

Live Use
Client Portfolios

Genetic Portfolio Optimization

Using evolutionary algorithms inspired by natural selection, we create efficient portfolios that maximize returns for given risk levels, with fully automated execution and continuous monitoring.

4a. Fitness Function: Genetic algorithms optimize portfolio weights to maximize return for target risk levels

4b. Order Execution: Fully automated trading system with real-time compliance checking and monitoring

4c. Risk Management: Daily Risk-ON/Risk-OFF signals for dynamic portfolio adjustment

4d. Reporting: Automated daily reporting for all managed accounts and products

                        "Survival of the Fittest": Our genetic algorithms continuously evolve portfolio weights, ensuring only the most efficient combinations survive and reproduce.
                    

🧬 Genetic Evolution

Natural selection principles for optimal weights

🤖 Full Automation

From optimization to execution to reporting

⚖️ Risk Controls

Maximum weights, sector limits, trading costs

📊 Real-Time

Continuous monitoring and adjustment

Optimization
Genetic algorithms find optimal weights

→

Execution
Automated order placement

→

Monitoring
Risk management signals

→

Reporting
Daily client updates