Why Generative AI Still Can’t Trade | David Wright on How Quant Alpha Actually Is Done With Machine Learning, Decision Trees, and Gradient Boosting

May 10, 2026

AI Summary

5 min read

David Wright, co-head of Pictet Asset Management's Quantitative Investments Franchise, which oversees over $30 billion in its quant group, explains that the generative AI revolution reshaping many industries has largely passed quantitative investing by. While tools like ChatGPT are useful for drafting client materials or summarizing meeting notes, they are fundamentally unsuited for making return forecasts. The core problem is that generative AI models are designed to generate plausible text, not to produce testable, interpretable predictions from structured financial data. As Wright puts it, "trading off something like that would really scare me as an investor."

Why Decision Trees Beat Neural Nets for Stock Forecasting

The machine learning techniques that Pictet actually uses are based on decision trees and gradient boosting, not the deep neural networks that power generative AI. Wright walks through the hierarchy: artificial intelligence means a computer doing a task that was historically human; machine learning means an algorithm trained on data to perform that task in a repeatable way. Generative AI, for all its power with text and images, fails on three counts for quantitative work: it cannot define a specific amount of information, it cannot be tested effectively, and its output is not interpretable.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of Monetary Matters with Jack Farley

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:00) **Introduction and Defining the Tools** - David Wright is introduced as co-head of PicTay's Quantitative Investments Franchise; Jack asks him to define the quant tools they use vs. what they don't.
2 (02:49) **Why Not Generative AI?** - Wright explains why generative AI is avoided for alpha generation, focusing on hallucination and lack of testability.
3 (04:38) **Rapid Advancements in Quant Finance** - Wright discusses how the field has advanced over the past three years, separating generative AI from machine learning adoption.
4 (07:14) **Support Functionality of Generative AI** - Wright elaborates on how generative AI is used in support roles, not for alpha generation.
5 (09:21) **The Look-Ahead Bias Problem** - Wright explains the critical challenge of look-ahead bias when testing generative AI models for trading.
6 (11:27) **Broader Economic Views on Generative AI** - Wright gives his personal and firm-level perspective on the massive spending on generative AI infrastructure.
7 (14:13) **The Core Process: Inputs and Outputs** - Wright describes the fundamental goal and data structure of their AI-enhanced investment process.

+ Full timestamped outline available in the app

Show Notes

To learn more about Pictet AI Enhanced US Equity ETF ($PQUS), click here: https://etf.am.pictet.com/pqus/

This interview is brought to you by Pictet Asset Management. To learn more about Pictet AI-Enhanced International Equity ETF ($PQNT), click here: https://etf.am.pictet.com/pqnt/

Jack Farley sits down with David Wright, co-head of Quantitative Investments at Pictet Asset Management, to discuss the machine learning techniques his team uses in their $30 billion quant franchise, and the degree to which AI has impacted serious quantitative investing. Wright explains why he prefers to utilize many decision trees and use gradient boosting rather than Generative AI to generate return forecasts, citing the need to avoid "hallucinations" and ensure models remain interpretable. The conversation explores their sophisticated investment process, which analyzes over 400 features, including accounting data, market trends, and analyst sentiment, to predict relative stock performance over 20-day horizons. These strategies, which now are included in new ETFs $PQNT (Pictet AI Enhanced International Equity ETF) and $PQUS (Pictet AI Enhanced US Equity ETF) are designed as "passive replacements," aiming to maintain a Beta of 1.0 while aiming to deliver an additional 1–2% annual outperformance over the relevant benchmarks, S&P 500 and MSCI EAFE indices. Finally, Wright addresses the common "black box" misconception of quantitative finance, advocating instead for a "crystal box" approach that provides full transparency into the economic rationale behind every trade. Recorded April 21, 2026.

For important information about the fund, please click: https://etf.am.pictet.com/”

Important Information

Before investing, carefully consider the fund’s investment objectives, risks, charges, and expenses. This and other information can be found in the fund’s prospectus or, if available, the summary prospectus, which may be obtained by calling (855) 994-4778 or visiting www.pictet.com/etf. Read it carefully before investing. (In Italic or Bold)

Investing in Exchange Traded Funds (ETFs) involves risk, including possible loss of principal. The fund's principal investment risks include Artificial Intelligence Models and Data Risk, Non-Diversification Risk, Convertible Securities Risk, Rights and Warrants Risk, Real Estate Investment Trusts (REITs) Risk and Sustainability & ESG Data Risk. For additional information about these and other fund risks, please refer to the "Principal Investment Risks" section of the prospectus.

ETFs are subject to additional risks that do not apply to conventional mutual funds, including the risks that the market price of an ETF's shares may trade at a premium or discount to its net asset value, an active secondary trading market may not develop or be maintained, or trading may b