Artificial intelligence is replacing the standard 24-month window with 10 years of historical data in credit analysis.Valora Investimentos / InfoMoney
How Is Valora Using Artificial Intelligence in Credit?
Valora Investimentos is feeding artificial intelligence tools with its proprietary database to track credit risk across series spanning up to a decade, rather than focusing solely on the recent 18-to-24-month cycle. The strategy was outlined by Daniel Pegorini, CEO and founder of the asset manager, in an interview with InfoMoney during a meeting with Clara Sodré, fund analyst at XP.
According to the executive, the system processes a proprietary data lake consolidating more than 50 indices from over 150 FIDCs (Receivables Investment Funds). The practical objective of this algorithmic structure is not to outsource final buy-or-sell decisions, but to quickly identify any operational deviation in an asset compared to its historical pattern built over years.
In the Brazilian structured credit market, most routine decisions tend to evaluate only the recent behavior of the debtor and collateral. By contrasting an anomaly with long-term databases, the team aims to anticipate whether a spike in delinquencies is passing noise or structural stress that puts capital at risk.
What Changes in FIDC Risk Assessment?
The core shift lies in the ability to detect statistical ruptures before defaults contaminate fund units. In receivables-backed operations, assets enter and leave portfolios continuously, making manual tracking across dozens of vehicles a task prone to delays or blind spots.
Pegorini emphasized that Valora has operated in the FIDC segment since 2008 and 2009, a period when the instrument was still in its infancy in Brazil. This nearly two-decade track record generated the raw material that now feeds the firm's computational models, which currently manage R$ 36 billion.
Total volume managed by the firm over its 20-year history.
Receivables investment funds tracked in the database.
Operational indicators fed directly into the AI.
When a receivables fund exhibits a sudden increase in average settlement terms or portfolio replenishment rates, artificial intelligence compares these figures with metrics from similar periods in past crises. This gives the management team immediate comparative context to decide whether to demand credit enhancement or reduce exposure to the asset.
Why Proprietary Data Outperforms Generic Models?
The value of this technological architecture lies in the fact that artificial intelligence does not rely on public, standardized information, but rather on data the firm has collected directly from debt origination and structuring since the post-2008 crisis.
“We feed artificial intelligence with our proprietary database, which tracks more than 50 indices across more than 150 FIDCs. Instead of looking at an 18-to-24-month scenario, the AI allows us to review a decade-long spreadsheet.” Daniel Pegorini, CEO of Valora Investimentos, speaking to InfoMoney
Artificial intelligence models applied to finance depend on the quality of their training data. In over-the-counter credit markets and structured securities, public default data tends to be too aggregated to signal problems in specific supplier contracts, trade bills, or consumer receivables.
The technology acts as a constant screening filter: instead of human analysts spending hundreds of hours consolidating monthly balance sheets for dozens of vehicles, the machine points out exactly where the standard payment curve begins to deviate from statistical expectations.
- Early identification of deviations from the fund's own historical average.
- Instant cross-referencing of 50 simultaneous indicators per portfolio.
- AI does not eliminate credit risk or replace the committee's human decision-making.
What Does This Mean for Private Credit Investors?
For investors allocating capital to structured credit funds, continuous investment in technology and proprietary data serves as an extra layer of governance against portfolio surprises. In periods of high interest rates and corporate financial tightening, delays in diagnosing defaults are typically the primary catalyst for unrecoverable losses.
The presence of analytical technology at an asset manager does not nullify the credit risk of the debt issuer, but it shortens the management team's reaction time. In credit products held in portfolios for years, the ability to identify operational deviations before month-end preserves the integrity of distributions paid to unitholders.
Still, retail investors must remain vigilant. Cutting-edge technology does not guarantee returns without volatility, and monthly fund reports remain the primary source for verifying the risks assumed when lending money through fund units.
Exposure by assignor and debtor — verify whether the credit funds you invest in have excessive concentration in a few economic groups, regardless of the technology used in management.
Subordination level — in FIDC structures, monitor the cushion of subordinated and mezzanine units, which are designed to absorb initial losses in the event of receivables defaults.
Provisioning transparency — observe whether management recognizes deviations and settlement delays in a timely manner or postpones mark-to-market valuations for distressed securities.
Rico aos Poucos Verdict
Valora's move reflects the technical maturation of private credit in Brazil, where competitive advantage is no longer just about offered yields, but the depth of monitoring data. Artificial intelligence strengthens risk screening by extending analysis to a decade of history, but it does not replace basic unitholder prudence: diversifying issuers, respecting liquidity profiles, and monitoring underlying asset strength remain the best insurance against defaults.