Time Series Momentum Strategies Using Deep Neural Networks

Hi! Here's Iván with this week's exciting newsletter, brimming with insights and discoveries on building robust investment strategies and risk models using Machine Learning.

In this edition, I am presenting the following sections:

• 🕹️ AI-Finance Insights: I summarize three must-read academic papers that mix cutting-edge ML/DL with Asset Pricing & Quant Finance:

  • Time Series Momentum Strategies Using Deep Neural Networks

  • The High-Frequency Factor Zoo

• 💊 AI Essentials: Today, I'm thrilled to share a user-friendly complete guide on how to start with the VectorBT python backtester.

• 🥐 Asset Pricing Insights: In this edition, I recommend the study "Spending Less After (Seemingly) Bad News," which provides top insights into how sensitive household spending is to bad macroeconomic news.

Today’s Sponsor

Your Brilliant Business Idea Just Got a New Best Friend

Got a business idea? Any idea? We're not picky. Big, small, "I thought of this in the shower" type stuff–we want it all. Whether you're dreaming of building an empire or just figuring out how to stop shuffling spreadsheets, we're here for it.

Our AI Ideas Generator asks you 3 questions and emails you a custom-built report of AI-powered solutions unique to your business.

Imagine having a hyper-intelligent, never-sleeps, doesn't-need-coffee AI solutions machine at your beck and call. That's our AI Ideas Generator. It takes your business conundrum, shakes it up with some LLM magic and–voila!--emails you a bespoke report of AI-powered solutions.

Outsmart, Outpace, Outdo: Whether you're aiming to leapfrog the competition or just be best-in-class in your industry, our custom AI solutions have you covered.

Ready to turn your business into the talk of the town (or at least the water cooler)? Let's get cracking! (And yes, it’s free!)

AI-Finance Insights

“Time Series Momentum Strategies Using Deep Neural Networks”

The paper introduces Deep Momentum Networks (DMNs), a novel method for improving time series momentum strategies.

DMNs enhance traditional momentum approaches by integrating deep learning-based trading rules within a volatility scaling framework.

The main contributions and findings are as follows:

👉 The study introduces DMNs, which leverage deep learning to simultaneously learn trend estimation and position sizing directly from data. This hybrid approach eliminates the need for manually defining these components, optimizing the Sharpe ratio of trading signals for better risk-adjusted performance.

👉 Backtesting on a portfolio of 88 continuous futures contracts demonstrates the superiority of the Sharpe-optimized LSTM model. The LSTM model more than doubled the performance of traditional time series momentum methods in the absence of transaction costs and continued to outperform even with transaction costs up to 2-3 basis points.

👉 To address the challenges posed by illiquid assets, the study proposes a turnover regularization term that factors in transaction costs during training. This addition ensures the model's effectiveness in high-cost trading environments, further enhancing its practical applicability.

👉 The Sharpe-optimized LSTM achieved the highest Sharpe ratio among tested models, indicating superior risk-adjusted returns. It also exhibited the lowest maximum drawdown and maintained controlled volatility, highlighting its robustness across different market conditions.

“The High-Frequency Factor Zoo”

The paper explores the unique roles of continuous and jump risk premia in asset pricing using a novel dataset of high-frequency returns for a comprehensive set of factors.

The main contributions and findings are as follows:

👉 Distinct Risk Sources: Continuous and jump returns represent distinct sources of risk, each with significant pricing implications. This distinction is evident across various factors, not just the market portfolio.

👉 Significant Risk Factors: Only a few factors exhibit statistically significant risk premia. Notably, four cluster-based risk factors stand out:

💊 Accruals Cluster: Derived from negative jumps with a risk premia of 1.64% per annum.

💊Skewness Cluster: Derived from positive jumps with a risk premia of -2.22% per annum.

💊Investment Cluster: Derived from positive jumps with a risk premia of -1.78% per annum.

💊Profitability Cluster: Derived from continuous returns with a risk premia of 3.01% per annum.

👉 Role of Jump Risk: Jump risk premia are crucial in explaining the cross-sectional variation in expected returns. The majority of the market portfolio's excess return is attributed to jump risk, particularly from negative jumps.

👉 Cluster Portfolio Analysis: The analysis of cluster portfolios reveals significant differences in continuous and jump risk premia among various portfolios. This underscores the necessity of differentiating between continuous and jump components for a comprehensive understanding of systematic risk.

👉 Cross-Sectional Variation: The study finds that jump risk premia explain most of the cross-sectional variation in expected returns, with continuous risk premia often being statistically insignificant.

Sponsor 👇

Learn from investing legends

Warren Buffett reads for 8 hours a day. What if you only have 5 minutes a day? Then, read Value Investor Daily. We scour the portfolios of top value investors and bring you all their best ideas.

Subscribe with one click.

AI-Essentials

📹 A user-friendly, complete guide on how to start with VectorBT backtester. It's not the easiest library to begin with, but it is probably the most complete one available right now. 👇

Asset Pricing Insights

“Spending Less After (Seemingly) Bad News”

How sensitive is household spending to bad macro news? The study "Spending Less After (Seemingly) Bad News" provides top insights. 👇

The study leverages local unemployment announcements at the CBSA level as a natural experiment to assess how salient macroeconomic news affects household consumption. Unemployment announcements that reach a 12-month maximum are particularly impactful, receiving significant media coverage and prompting increased internet searches for "unemployment."

In affected areas, discretionary spending drops by 2% and credit card repayments by 3.6%, compared to similar areas with identical fundamentals that don't experience such an announcement.

This decrease in spending, particularly prominent among low-income households, persists for two to four months, reflecting the heightened sensitivity of household consumption to salient, yet sometimes inaccurate, economic news.

If you're enjoying our newsletter and want to support us, please recommend it to anyone you know who's interested in AI and Finance. Your referrals are the biggest compliment and help us grow! 🌟🤖💼