NYU Research | Independent Study

A 15% edge.
With your risk in the equation.

An independent study by NYU Stern doctoral researcher Valerie Baldinger followed Mezzi’s advice into real investment decisions. Investors acted on personalized recommendations, bought more diversified funds and chose funds with lower fees.

By   ·   Oct 2, 2026   ·   4 min read

+15%Higher historical
return-to-risk score.

What else the
investor bought

71.9 / 100
Market benchmark = 100

What Mezzi
recommended

82.6 / 100
Market benchmark = 100
Historical constructed portfolios, not realized member returns. Squares rounded; labels show reported averages. Study details [1]

What this means for you

  • More personal advice. More follow-through.

    25% → 44%People acted on about 1 in 4 of the least personalized investment answers, compared with nearly 1 in 2 of the most personalized.
  • Less dependence on individual stocks.

    36 morediversified fund purchases per 100 purchases, compared with the same investors’ other purchases.
  • More money stays invested.

    $900 a yearless in fund costs on $1 million invested at the lower fee. An illustration using the study’s average fee difference.

From the research paper · Conclusion

“I find high rates of advice implementation. This is notable given extensive evidence that changing household financial behavior is difficult”

Valerie Baldinger

Valerie Baldinger
Doctoral researcher in finance, NYU Stern
From Prompt to Portfolio, conclusion, p. 21. [7]

More personal advice. More people acting on it.

When Mezzi’s answers took more of an investor’s circumstances into account, people were more likely to follow the recommendations.

From about 1 in 4 to nearly 1 in 2.

Answers followed by a recommended purchase within ten trading days

Least personalized advice~25%
Most personalized advice~44%

Across all buy-advice answers, the follow-through rate was 36.2%.

Approximate rates from Figure 7: personalization scores 1–2 versus 9–10. Rates count answers, not individual customers. Source & details [5]

Your financial picture gives that advice context. Lee Nicholson describes the value of sharing more than his stock holdings:

A member’s perspective

Your financial life extends beyond your portfolio.

“You will be rewarded greatly for the time you invest to load it up with information about your financial life. Even if you have lots of non-stock (like real estate) investments, it handles these great if you tell Mezzi what you have.”

Lee Nicholson
Lee NicholsonMezzi member

More diversified funds. Fewer individual stocks.

The investments people chose matter, too. Purchases following Mezzi’s advice were more likely to be funds that spread money across many investments, and less likely to be shares in a single company.

Owning a few companies

More depends on how each one performs.

Owning a broad stock fund

Your money is spread across many companies.

Illustration of diversification, not actual portfolio holdings.

The estimated difference for every 100 purchases:

+36

Diversified-fund purchases

−41

Individual-stock purchases

Compared with the same investors’ other purchases after receiving advice. Rounded estimates. Source & details [6]

This also showed up in their portfolios. Broad index funds made up a larger share of investors’ holdings after they started using Mezzi. [6]

Smaller swings. Shallower declines.

The 15% finding looks at return alongside risk. In the same historical comparison, portfolios built from Mezzi’s recommendations also had smaller price swings and less severe declines from their peaks.

For an investor, the point is to consider what you earn alongside the ups and downs you take to earn it. [1]

This is a historical comparison of constructed portfolios, not returns earned by members. How the comparison works ↓

Keep more. Pay less in fund fees.

Fund fees come out of your investment, year after year. Paying less leaves more of your money invested.

The ETFs and mutual funds people bought on Mezzi’s recommendation charged lower fees than other funds those same people bought after receiving advice. The average difference was about $9 a year for every $10,000 invested. [6]

The long-term value of lower fees

Keep the savings. Let them grow.

The savings can earn returns of their own. Here is how that could add up over 20 years.

$900less in fund costs in year one
$64,600more in your portfolio after 20 years

Additional portfolio value from lower fund fees

How lower fund fees can add up over 20 yearsAt an assumed 7% annual return, a 0.09 percentage point annual fee reduction on an initial $1 million produces about $64,600 more portfolio value after 20 years.$0$20k$40k$60k$80kToday5 years10 years15 years20 years

At 7% assumed annual growth: about $16,500 more after 10 years and $64,600 more after 20 years.

Hypothetical illustration · $1 million starting balance · 0.09 percentage points lower annual fund fees. Savings stay invested. Calculation & disclosures ↓

Holding a constant $1 million in funds at that lower fee would mean about $900 less in annual fund costs, more than twice the $399 annual price of Mezzi Advisory. The dollars depend on how much you invest at the lower fee and which funds you choose.

Understand more of what you own.

Financial advice should help you understand the recommendation, too. The study found signs that investors picked up financial language from Mezzi and brought it into later conversations, including terms such as diversification and allocation. [8]

From the research paper · Learning

“I find suggestive evidence that investors thus learn from interacting with generative AI: financial terms used in AI answers are more likely to appear in investors’ subsequent prompts, including in later conversations.”

Valerie Baldinger, From Prompt to Portfolio, section 6.3, p. 18. [8]
Explore the researchStudy design, comparison charts and the calculation behind the 15%

What makes this study different?

It follows actual purchases and holdings using Mezzi records. Mezzi shared de-identified records, with personal identifiers removed. The study reports only aggregate results. Gallup–Edward Jones surveyed attitudes toward financial advice. Stanford and MIT researchers simulated the effects of following AI advice. These studies answer different questions.

Mezzi investors chose whether to act. The AI did not execute their trades. [4]

How advice became purchases

Over the study, the share of buy-advice answers followed by a suggested purchase rose from about 6% to about 35%, using approximate endpoints from the paper’s chart. The trend measures follow-through, not trust directly. [5]

By the final observed portfolio date, about 3 in 10 investors held an investment Mezzi had introduced, one they had not named in their question or traded in earlier records. [2]

The numbers behind the 15%

How the portfolios compared

Share of the market’s Sharpe ratio. Source: p. 12, Table D.6.

The study uses the Sharpe ratio, which measures return above a cash benchmark for each unit of risk. Recommended portfolios retained 82.6% of the market’s ratio, versus 71.9% for the alternatives. Dividing 82.6 by 71.9 and subtracting one gives the roughly 15% relative difference. [1]

Two portfolios. The same investor.

PORTFOLIO A

What Mezzi recommended

A portfolio built from the investments Mezzi recommended to that person.

PORTFOLIO B

What else they bought

A portfolio built from other investments that same person bought after receiving the recommendation.

The test: compare both over the same five-year history.

The researcher compared their returns, how much they fluctuated and their largest declines.

Historical portfolio comparison. Source: pp. 12–13, Table D.6.

To Mezzi, you are the only client.

You'll never be prioritized based on net worth. There's no scheduling or wait times. You'll never hear "I have a hard stop" or "remind me about that again." With Mezzi, it's all, and only, about you.