Hi! Here's Iván with this week's exciting newsletter, brimming with insights and discoveries on building robust risk models and trading strategies using Machine Learning. This edition stands out for its focus on LLMs applications to sentiment analysis to improve investment strategies.

• 🕹️ 2 Academic Articles: Dive into groundbreaking research featuring actionable ideas that are reshaping our understanding of how to apply ML/DL in creating successful investment and trading strategies.

• 💊 Video Insight : Great introduction to OpenAI's Assistant API using Python.

• ✔️ Market Insight: Best market reflection shared on my LinkedIn/Twitter during the last week: the financial market is always first!

• 🥐 Tail-risk protection using ML: More advanced risk management application to deepen your understanding of how to design robust ML/DL risk models.

Academic Insights

“ChatGPT-based Trading Strategy using Headlines Sentiment ” 

Although the paper covers a relatively short sample period and the results might not be as robust in other periods, I still think that the idea is worth reading!

Summary of the paper in only 4 ideas: 👇

1. The authors classify headlines as positive, negative, or neutral for each stock using ChatGPT. Their sample period begins in October 2021 and ends in December 2022.

2. They form portfolios using the headlines' sentiment classification.

3. Without considering transaction costs, a strategy that buys the stocks with a positive ChatGPT score and sells stocks with a negative ChatGPT score after the news announcement earns a cumulative return of over 400% from October 2021 to December 2022.

4. They also observe that predictability is higher in small stocks and after negative news, which is consistent with classical theories of frictions and limits to arbitrage.

“Commodity Trading Using Deep Reinforcement Learning.” 

Summary of the paper in 4 simple steps 👇

1. The paper uses Deep Reinforcement Learning to trade front-month Natural Gas futures. It employs trade-level data from the sample period of 2011 to 2022, encompassing over 20,000 samples.

2. The proposed system employs a novel time-discretization scheme that adapts to market volatility, thereby enhancing the statistical properties of subsampled financial time series.

3. They utilize CNNs and LSTMs as parametric function approximators to map historical price observations to market positions.

4. Backtesting on front-month natural gas futures demonstrates that DRL models increase the Sharpe ratio by 83% compared to the buy-and-hold baseline.

AI-Essentials: Step-by-Step Tutorial

🚀 In today's newsletter, we're excited to share a video that provides a great introduction to OpenAI's Assistant API using Python. Begin your journey with LLMs models.

The post: Market Ideas

Sentiment readings are currently just middling around. No extremes in either direction…Sentiment is currently NOT a headwind. If anything, you can argue that it’s still a tailwind. 👇

ML Risk: Insight

“Tail-risk protection: Machine Learning meets modern Econometrics” 

When the market faces big losses, spreading your investments (diversification) doesn't always work. This is because, during these tough times, different types of investments can all lose value together. Normally, diversification works because different investments usually don't move the same way at the same time.

During big market drops, everything can start to move together, causing big losses even in well-diversified portfolios. When this happens, it's good to have other ways to protect your investments. One method is 'tail-risk protection,' which helps safeguard against extreme market drops.

A big challenge in finance is to create models that can protect investments during these extreme events. One strategy is to design your investments to gain when the market goes up but protect you from big losses. This is known as “tail risk protection strategy.” It's usually more efficient and flexible than using options.

But predicting how your portfolio will behave is hard. Markets change, and so do the factors that affect them. To help with this, some experts have come up with a new strategy. They use a method called “dynamic tail risk protection.” This method uses a special type of risk measurement, called VaR (Value-at-Risk), to adjust and protect the portfolio.

They tested different methods and found that combining them (an ensemble method) works best. This approach improved both their predictions and their trading results.

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! 🌟🤖💼