Research

volarixs - applied AI & ML to finance

Explore our latest posts on machine learning, market dynamics, strategy architecture and design

Feature Engineering
Feb 10, 2026

Shrinking the Feature Space: PCA & Autoencoders

Many features are redundant or noisy. High dimensionality = harder to generalize.

PCA
Autoencoders
Features
9 min read
Strategy
Feb 1, 2026

How Asset Managers Can Implement AI & Machine Learning

Part 2: Infrastructure, Governance & Roadmap. What it takes to implement AI in asset management.

AI Implementation
Governance
Roadmap
18 min read
Deep Learning
Jan 28, 2026

Neural Networks for Market Data: MLPs, CNNs & LSTMs

We are selective with deep learning. Expensive to train, easy to overfit, harder to debug.

Neural Networks
MLP
LSTM
12 min read
Research
Jan 22, 2026

Signal Half-Life and Decay: How Long Do ML Edges Really Last?

If you discover a signal today, how long will it work?

Signal Decay
Half-Life
Edge Persistence
13 min read
Strategy
Jan 15, 2026

How Asset Managers Can Use AI & Machine Learning in Investment Decisions

Part 1: Use Cases & Value. Real-world use cases: idea generation, regime analysis, risk management.

Asset Management
AI & ML
Use Cases
15 min read
Volatility
Jan 5, 2026

Modeling Market Turbulence: GARCH, EGARCH & HAR

Volatility ≠ returns: heavy tails, clustering, mean reversion. Dedicated volatility models are essential.

GARCH
EGARCH
HAR
10 min read
Time Series
Dec 18, 2025

ARIMA, SARIMAX & VAR: When Classical Time-Series Still Win

Explicitly model temporal dependence with transparent structure.

ARIMA
SARIMAX
VAR
9 min read
Benchmarks
Dec 9, 2025

Volatility Forecasting Benchmarks: GARCH, HAR, and ML

Compare GARCH, HAR, and ML models for volatility forecasting.

Volatility
GARCH
HAR
11 min read
Machine Learning
Dec 2, 2025

How Market Regimes Break ML Models

Financial machine learning rarely fails because the model is 'bad'. It fails because the market regime changed.

Regimes
ML
Backtesting
8 min read
Models
Nov 25, 2025

Boosted Trees for Alpha: XGBoost & LightGBM

Gradient boosting dominates tabular ML. Learn how XGBoost and LightGBM deliver strong performance.

XGBoost
LightGBM
Boosting
11 min read
Features
Nov 18, 2025

The 19 Most Important Features for Equity Return Forecasting

Most ML performance in finance doesn't come from the model — it comes from the features.

Features
Alpha
Equities
12 min read
Methodology
Nov 7, 2025

Rolling Windows for Financial ML: A Complete Guide

If you use financial data and your model does not use a rolling window, the backtest is wrong.

Rolling Windows
Time Series
Backtesting
10 min read
Evaluation
Oct 27, 2025

Beyond Sharpe: A Research Framework for Evaluating ML Trading Strategies

Sharpe ratio is dangerously incomplete for ML strategies.

Evaluation
Metrics
Sharpe
15 min read
Models
Oct 8, 2025

Random Forests in Finance: Nonlinear Signals Without the Drama

Tree-based ensembles capture nonlinearities and interactions in market data.

Random Forest
Extra Trees
Trees
10 min read
Models
Sep 15, 2025

From Linear Regression to Lasso: Fast, Interpretable Baselines

Linear and regularized regressions still do serious work in finance.

Linear Regression
Ridge
Lasso
12 min read
Regimes
Aug 22, 2025

Market Regimes, Clusters & HMMs: Teaching Models to Respect the Environment

Episodes where statistical properties are stable enough: high vol vs low vol, risk-on vs risk-off.

K-Means
GMM
HMM
11 min read
Architecture
Aug 3, 2025

Building a Universe-Wide Prediction Grid

An alpha factory needs predictions for every asset at multiple horizons from multiple models.

Prediction Grid
Scaling
Alpha Factory
14 min read
Evaluation
Jun 18, 2025

Regime-Conditioned Performance: Measuring ML Robustness

Most backtests report a single Sharpe. But ML models fail by regime.

Regimes
Robustness
Performance
12 min read
Features
November 18, 2025
12 min read

The 19 Most Important Features for Equity Return Forecasting

Most ML performance in finance doesn't come from the model — it comes from the features.

This guide covers the essential features that actually predict returns, why some features work only in certain regimes, which features are overused and unreliable, and how volarixs automates feature pipelines for equities, FX, commodities, crypto.

1. Price-Based Features (Core Predictors)

1. Returns (log, pct, normalized)

  • Rolling 1D, 5D, 21D
  • z-scored returns
  • vol-adjusted returns

These capture momentum and short-term reversal patterns.

2. Moving Averages

Windows: 5, 10, 21, 50, 100, 200.

Especially useful:

  • Slope of moving average
  • Distance to long-term trend
  • Golden/death cross flags

3. Trend Strength (ADX, R² of rolling regression)

Trend strength often predicts short-term follow-through only in calm regimes.

4. Volatility (realised, Parkinson, Garman-Klass)

Key for normalizing other features.

5. Skew and Kurtosis (rolling)

Useful for detecting impending breakouts or regime changes.

2. Volume & Liquidity Features

6. Volume z-score

Detects abnormal participation → strong predictive edge around earnings and news.

7. Volume/Volatility ratio

Liquidity-adjusted momentum.

8. Turnover

Stable cross-sectional predictor.

3. Microstructure Features

9. Bid–ask spread (if intraday available)

Predicts near-term returns via order flow imbalance.

10. Order book imbalance

Directional but regime-dependent.

11. Signed volume

Best intraday momentum predictor.

4. Cross-Asset Features

12. Market beta (rolling)

Beta drift contains information about crowding + factor exposures.

13. Sector/industry momentum

Very strong short-term cross-sectional factor.

14. Index return lead-lag

Useful for single-stock prediction.

5. Volatility & Risk Metrics

15. Realised volatility ratio (short vs long window)

Detects transitions between calm → elevated volatility states.

16. Volatility-of-volatility

Useful around macro events and earnings.

17. Correlation regimes

Cross-asset correlation shifts often precede structural changes.

6. Event-Driven Features

18. Earnings date proximity

Pre-earnings drift is well-documented.

19. Overnight returns

One of the strongest predictors in equities (microstructure-driven).

The 19 Essential Features

Returns (log, pct, normalized)
Rolling 1D, 5D, 21D, z-scored, vol-adjusted
Moving Averages
Windows: 5, 10, 21, 50, 100, 200. Slope and distance to trend
Regime-dependent
Trend Strength (ADX, R²)
Rolling regression R², ADX indicator
Regime-dependent
Volatility (realised, Parkinson, Garman-Klass)
Key for normalizing other features
Regime-dependent
Skew and Kurtosis
Rolling skew/kurtosis for regime detection
Regime-dependent
Volume z-score
Detects abnormal participation around earnings/news
Volume/Volatility ratio
Liquidity-adjusted momentum
Regime-dependent
Turnover
Stable cross-sectional predictor
Bid-ask spread
Predicts near-term returns via order flow imbalance
Regime-dependent
Order book imbalance
Directional but regime-dependent
Regime-dependent
Signed volume
Best intraday momentum predictor
Market beta (rolling)
Beta drift contains information about crowding
Regime-dependent
Sector/industry momentum
Very strong short-term cross-sectional factor
Index return lead-lag
Useful for single-stock prediction
Regime-dependent
Realised volatility ratio
Short vs long window detects regime transitions
Regime-dependent
Volatility-of-volatility
Useful around macro events and earnings
Regime-dependent
Correlation regimes
Cross-asset correlation shifts precede structural changes
Regime-dependent
Earnings date proximity
Pre-earnings drift is well-documented
Overnight returns
One of the strongest predictors in equities

Click on each category to explore the features. Features marked as "Regime-dependent" may behave differently across market regimes.

Feature Importance Calculator

Feature Importance Scores
Top Features
Returns (1D): 85%
Moving Average (21D): 72%
Volatility (realised): 68%
Volume z-score: 65%
Market beta: 58%

Feature importance varies by market regime and forecast horizon. Adjust parameters to see how different conditions affect which features matter most.

How volarixs Uses These Features

volarixs offers:

  • Pre-built feature pipelines
  • Automatic leakage detection
  • Rolling window transformations
  • Regime-conditional features
  • Cross-asset enrichment (FX, commodities, indices)
  • Normalisation + scaling logic baked into the BaseModel class

You can plug these features into:

  • Ridge
  • ElasticNet
  • Random Forest
  • XGBoost
  • LSTM
  • any custom template (coming with LLM v2)

Conclusion

In financial ML, 80% of performance comes from the right features.

This list covers the features that consistently survive out-of-sample tests and provide real predictive edge across market regimes.

Features
Alpha
Equities
Data

Ready to build feature-rich models?

Start experimenting with these features in volarixs.