# Antonio Marrazzo: 'You cannot become rich from just prompting'

URL: https://start.joinautopilot.com/journal/antonio-marrazzo-interview
Updated: 2026-11-16

Antonio Marrazzo grew up in a small town in Patagonia, Argentina, then moved to Buenos Aires at 18 to study economics and actuarial science at the same time. Antonio worked as a financial data analyst at JPMorgan and Accenture before moving into quantitative research, then came to the United States as a Fulbright scholar; the latest step is a master's in data science.

[Open it in the Autopilot app.](https://link.joinautopilot.com/eXKr?c=antonio-marrazzo-interview&af_sub1=intro) https://link.joinautopilot.com/eXKr?c=antonio-marrazzo-interview&af_sub1=intro

Antonio now works at an alternative data provider that processes financial text, such as earnings calls, news and social media, from sources around the world. On Autopilot, Antonio publishes Argentina's Ascent, a Portfolio built on Antonio's own read of Argentina's economy from the ground.

In the interview, Antonio was blunt about the limits of asking a chatbot to pick stocks. We talked about the difference between weak data and weak models, how Antonio checks a dataset before trusting it, and what regime-based investing actually means.

This interview comes from the transcript of our conversation with Antonio Marrazzo in June 2026. It has been edited and condensed.

### Who are you, and how did you get here?

My name is Antonio Marrazzo. I come from Argentina, the south of Argentina. I grew up in a small town in Patagonia. Then I moved to Buenos Aires, which is the capital city of Argentina, when I was 18 to study, and I moved here last year as a Fulbright scholar to the U.S. So that's kind of my story, how I moved from place to place.

### What did you study, and how did that shape your approach to markets?

I studied economics and actuarial sciences both at the same time. Studying economics, I was really into social sciences, I wanted to understand better society, and I thought from an economic standpoint that would be the best way to understand things. I've been always driven by curiosity. And from actuarial science, what I like was the applied statistics it has: you're always modeling, doing valuation, pricing, reserve calculations.

### What was your career path into quant research?

I had really both, and that kind of shaped my approach to markets, because that's how I got into quant finance. As a financial data analyst, my first job, JPMorgan, Accenture, it was more data analysis, financial data analysis, and I got into more quant research after that. My first roles were around financial data analysis, I analyzed companies, I learned a lot about financial statements, trying to understand growth trends and industries, but never from a trading standpoint. I like competitive games, and I think one way to prove your skill set as a professional is the trading floor, actually having skin in the game. That's why I got into quant finance. I got into quantitative research at a quant fund in Argentina.

And then when I moved here, I started working on a data provider that is my current company. We process financial text from more than 30,000 sources worldwide, earnings calls, news, social media, and extract actionable insights out of it. We do what's called in the industry alpha mining, from financial text.

### What are you actually looking for in all that financial text?

The data I'm focused on is all publicly disclosed financial text. It's actually pretty valuable, but it's unstructured, so there's work in cleaning it, filtering it, normalizing it, and then extracting an insight that could be actually tradable. The good thing about this data is that it's publicly disclosed, it's on the web, you can scrape it, it's in public filings from companies. But the hard part is actually knowing what works. If there's a manager saying, okay, we're raising our sales target for next quarter, that's improved forward-looking sentiment, that's a signal. It's not everything, you cannot only trade on that, but it's something that's actually valuable for making a decision.

### How do you think about having an edge in the market?

I think about this as a two-layer game. The first layer is you need to have an information edge. One of those edges could come from alternative data, from having better data than your competitor, or data that's actually predictive of something.

One category of that information edge is expertise: you have big asset managers such as Warren Buffett who say, I don't invest in companies I don't understand. There are businesses out there that are really hard to understand, and if you're an expert who actually understands the business, you might have an edge, because you could see that a business is poised to grow, since you actually understand that business. Another category is the data layer: what is my data set, what is my information set, and what position am I in against the rest of the people in this game regarding data.

The second layer is the modeling edge. Suppose you have the same information set as all the rest of the players, how can you differentiate? It comes from how you model. If you model better, if you forecast better than the median consensus, you have a trade edge.

### Do most quantitative failures come from bad models or bad data?

I think most quantitative failures come from weak information rather than weak models. You can have a weak model, but really good features and feature engineering, and the output won't be as bad as if you already have weak information. There's a phrase in data science, garbage in, garbage out: you won't be able to predict accurately if your information set is weak. So paying attention to the information is the most important thing. There are papers that show this: if you compare a set of different machine learning models, logistic regression, random forest, XGBoost, within the same information set, performance doesn't vary that much for a classification model, whereas if you have the wrong feature engineering, that would fail.

### How do you make sure your data is actually reliable?

The way you actually know your data is accurate is you analyze it prior to doing any type of work. You check for anomalies. One thing I always check is checking for biases: you don't want survivorship bias in your data. If you're making a strategy with S&P 500 constituents, you would like to have the delisted ones in your past data too, because if not you will only be working with the ones that have survived. You always want to make sure the data you're using was available at that point in time, so point-in-time data is really important. And checking for anomalies: you shouldn't use restatements from financials, you should use the financials released at that moment, because they are all restated after, you know, they are checked by the auditing firms.

### How does AI actually power your Portfolio?

To understand how AI actually powers my portfolio, you should define at least two types of AI. One is predictive AI, and that relates to machine learning models mainly. The second one is generative AI. If you're having an agent that is processing earnings calls, you will have an agent that analyzes and scores. That score is what you actually ingest into a predictive AI system, a machine learning model. So you have these two layers. Generative AI is just for actually generating text, or images, which is not the case here. The generative AI layer is used for explainability and for processing, rather than just for actually making allocation decisions. For those allocation decisions, you need to use AI models that are more quantitative, and that's where machine learning comes from, more on the predictive side, and on the allocation, the optimization side.

### So you're not just asking ChatGPT which stocks to buy?

No, that would be a really bad decision. You should build an agent that is built on top of good practices of quant investing, with the appropriate tools, with the appropriate math behind it, and not just prompting. You cannot become rich from just prompting.

### What is regime-based investing, in simple terms?

The first thing to understand is markets have cycles. There will be moments where all stocks are rising, there will be moments where they're not. So I would explain regime-based investing as understanding, identifying, which cycle of the market you're at. Markets could be rising, where all stocks are really correlated, rising, or you might have the opposite, which is stocks moving not at the same time. Regime-based investing is understanding what is the current state of the overall market, because probabilities for your individual stock-picking performance will change based on that.

I don't believe in this, just buy and forget it, because if you were buying in 2008, you may be recovering the money in 2013. So I think regime-based investing, it's about timing and being aware in the worst moments of the market, which I think is actually key.

### Why did you build a Portfolio focused on Argentina?

I think retail investors in the U.S. need proper access to Argentinian stocks. That's why I thought I needed to build this, using my own models to give exposure to Argentina for retail investors in the U.S. The good thing is you're able to invest in these stocks because they are all listed in the New York Stock Exchange, they all have ADRs, what's called for international stocks. The good thing is you're able to invest in these stocks because many trade in the U.S. as ADRs. Editor's note: eligible Argentine holdings can trade as ADRs on more than one U.S. exchange. Nasdaq identifies GGAL as Nasdaq-listed ADSs. This note does not reconstruct the Portfolio's June holdings.

> "You cannot become rich from just prompting."

## Follow Antonio Marrazzo's Portfolios on Autopilot

[Antonio Marrazzo publishes Argentina's Ascent on Autopilot, an investing app that lets you follow Portfolios from investors, AI models, and public filings in your own brokerage account.](https://link.joinautopilot.com/eXKr?c=antonio-marrazzo-interview&af_sub1=body) Visit [Antonio Marrazzo's Pilot page](https://start.joinautopilot.com/strategies/pilots/antonio-marrazzo-4159). https://link.joinautopilot.com/eXKr?c=antonio-marrazzo-interview&af_sub1=body

Following a Portfolio on Autopilot works like this:

- You follow in the brokerage account you already have, and the money stays there.
- On Premium Tier, when the Portfolio changes, Autopilot Advisers sends the orders to your broker; on Basic Tier you get an alert.
- Review the Portfolio's method, holdings and any available performance record before following it. New Portfolios show return figures after 30 days.
- The Pilot sets the subscription price, there's a 0.00% AUM fee, and you allocate at least $500.

You're never trading at the same moment as Antonio, and your account won't match the Portfolio exactly.

[Open it in the Autopilot app.](https://link.joinautopilot.com/eXKr?c=antonio-marrazzo-interview&af_sub1=end) https://link.joinautopilot.com/eXKr?c=antonio-marrazzo-interview&af_sub1=end

"Autopilot" refers to Autopilot Holdings Corp and Autopilot Advisers, LLC, an SEC-registered investment adviser. Argentina's Ascent is managed by Antonio Marrazzo, an independent Third-Party Pilot unaffiliated with Autopilot. Autopilot doesn't manage, validate, or guarantee this Pilot's portfolio. This interview reflects Antonio's own views as of June 2026, edited and condensed, and isn't a recommendation to buy or sell any security or to follow any Portfolio. Concentrated and thematic strategies carry heightened risk of loss. Investing carries risk-including possible loss of principal-and past performance is not indicative of future results. Figures current as of 10/09/26. View our Customer Relationship Summary and other important info at [joinautopilot.com/disclaimer](https://joinautopilot.com/disclaimer).

## Portfolios and pilots mentioned in this article:
Antonio Marrazzo: https://start.joinautopilot.com/strategies/pilots/antonio-marrazzo-4159
1 public portfolio.
Argentina's Ascent (Antonio Marrazzo): https://start.joinautopilot.com/strategies/argentina-s-ascent-968349
Returns as of November 16, 2026. Month -23.7%, 3 months -23.8%, YTD n/a, 1 year n/a, Since start -22.4%.

The performance data presented reflects the aggregated results of actual client portfolios managed by the Registered Investment Adviser (RIA), based on the trading activity in the securities contained in the relevant portfolio over the specified period. If fewer than 10 clients are invested in the portfolio, the data represents the aggregate performance of all clients following this strategy. Performance figures are shown net of advisory fees, calculated using the applicable subscription fee and an assumed $50,000 account size. Because subscription fees are flat rather than asset-based, accounts smaller than $50,000 would experience a higher effective fee percentage and lower net performance, while larger accounts would experience a lower effective fee percentage. Actual investor outcomes can vary due to factors like brokerage support for fractional shares, account size, trade timing, fees, and taxes, and past performance should not be interpreted as a guarantee of future results or typical of all portfolios.
