RenaAi 레나에이아이
KORENG

Data assets

Three kinds of alternative
data RenaAi builds

RenaAi builds alternative data with its own methodology and connects it to global financial markets. This page covers the “building” part: the three data assets we have built ourselves.

Small-business alternative credit scoring (market area), equity alternative data and patent intelligence turn credit, corporate and technology value into assessable signals. Rating agencies, lenders, asset managers and investors receive them via API and data feeds. The market-area model is backed by a patent application filed, a paper submitted to Finance Research Letters, and an out-of-sample AUC of 0.83.

Market-area dataEquity dataPatent data

The three assets at a glance

Market area
Small-business probability of default (PD) · API supply to lenders and rating agencies
Equity
Normalized signals from filings, prices and trading flows · screening, monitoring, reporting
Patent
Patent quality score · base signal for valuation, collateral and securitization
Delivery
REST API · data feed
Who uses it
Rating agencies · lenders · asset managers · investors · REITs · appraisal

Market-area data

Small-business alternative credit scoring

We use market-area data to estimate the probability of default (PD) for businesses with no credit history. The same market-area and credit data opens the next market at each stage, from small-business credit to real-estate capital markets. The more data accumulates, the faster the next stage comes.

What we build and for whom

Output
Small-business probability of default (PD) · market-area risk leading index
Methodology
(Market area × business type × quarter) cell normalization · market-area network (R₀) engine
Coverage
3.16 million thin-file small businesses
Users
Lenders · rating agencies
Delivery
API
Evidence
Patent application filed (9 claims) · out-of-sample AUC 0.83

Five stages the same data opens

  1. 1

    Core market-area analytics technology

    IP · Patent

    A data ontology that normalizes scattered public market-area, sales and population data into a single (market area × business type × quarter) cell, plus a market-area network (R₀) engine. Every later business is built on top of it. Spatial contagion was confirmed as real in daily data, and we showed that quarterly aggregation erases a third of that signal.

    NowCore IP
  2. 2

    Small-business alternative credit scoring

    SBCR

    Estimates probability of default (PD) from market-area and sales data, even without credit history. Makes 3.16 million thin-file small businesses assessable and supplies the scores to lenders via API.

    Now~$110M+
  3. 3

    Regional leading index

    SDCRI

    Turns accumulated data into a real-time index of market-area risk and vitality. Sold as a data subscription to franchises, real estate and media, with the goal of becoming an industry-standard indicator.

    ExpansionHundreds of $M
  4. 4

    Real estate valuation

    Leading indicator

    Empirically shown that market-area risk leads retail cap rates, rents and vacancies (t≈7). Offers forward-looking risk and value diagnostics, rather than lagging appraisals, to financial institutions, REITs and appraisers.

    Expansion~$1B scale
  5. 5

    Real estate derivatives · RWA

    Capital markets

    Parametric insurance, cap-rate derivatives and real-world asset tokenization (RWA) using a validated index as the underlying. The final vision: the largest market, with the highest barriers to entry.

    VisionSeveral $B

The same engine widens the market at every stage: from small-business credit (~$110M+) to real-estate capital markets (several $B).

We Don't Just Automate. We Multiply.

Market sizes are rough estimates. KRW amounts converted at ₩1,340 = US$1 (October 2026).

Equity data

Signals from listed companies that move before financial statements

The value of a listed company moves before its results are announced. We normalize filings, prices and trading flows to a single standard and turn changes not yet written into the financial statements into signals.

  1. Normalization

    Scattered sources, one line

    We align filings, prices and trading flows, which differ in format and frequency, into time series on a single standard. It is the ontology we built for market-area data, carried over to individual stocks.

  2. Signal separation

    Separating noise from signal

    We read the persistence of growth and the build-up of risk separately. Telling what is noise and what is a signal tied to value is judgment learned from asset management.

  3. Delivery

    In a form that supports decisions

    Supplied as screening, monitoring and reporting. Instead of raw data, we deliver it in a form you can put straight into decision-making.

What we build and for whom

Output
Time-series signals for listed companies from normalized filings, prices and trading flows
Methodology
The same normalization engine as our market-area alternative credit scoring, applied to individual stocks
Users
Asset management · research
Delivery
Screening · monitoring · reporting

This is a data and indicator supply service. It is not investment advice or discretionary investment management, and it is not a recommendation to buy or sell any security.

Evidence

Why it can’t be copied,
even with the same public data

Patent · validated performance · academic credibility · demand pipeline.
RenaAi’s real value lies in the methodology that turns market-area data into credit risk.

Patent, 9 claimsOut-of-sample AUC 0.8325Submitted to SSCI journalDemand pipeline

AUC 0.8325 is the discriminatory power of predicting default among small businesses with no credit history using market-area data alone, scored on a period not used to train the model (out-of-sample). Closer to 1 is more accurate.

Four checks the methodology has passed

  • Patent (IP)

    Patent application filed with 9 claims through patent attorney Kyung-min Woo (우경민) of Meta Patent Law Office. It protects the thin-file scoring methodology itself (Claim 9), not just its output.

    Anyone using the same method has to design around the claims. The first to file sets the standard.

    claims9

  • Validated performance

    With zero credit history and market-area data alone, the model reaches an out-of-sample AUC of 0.83 (an accuracy measure; closer to 1 is better), above the 0.70–0.80 range of card issuers’ retail scoring built on decades of payment history.

    Accuracy rises further when personal credit data is added through pseudonymized data linkage and SCB.

    AUC0.8325

  • Academic validation

    Researchers affiliated with NUS AIDF, co-founded by MAS, NRF and NUS (Seyoung Park and Hyuntak Lee), and a paper submitted to Finance Research Letters (an SSCI-indexed journal) support the credibility of the methodology.

    Alternative credit depends on adoption by regulators and financial institutions, so peer-reviewed evidence is needed.

    SSCIFRL submitted

  • Demand & pipeline

    We have a pipeline of pilots and collaborations with financial institutions and the core lenders to small businesses, including savings banks, capital (consumer finance) companies and internet-only banks.

    Validated model + demand that already exists = lower commercialization risk.

    demandLenders
Result: out-of-sample discrimination0.8325

Validation

We raised performance and mapped the limits

We compared three generations on the same sample and the same split. We wrote more about what did not work, and why, than about what did.
v1 · original design
0.7470
v2 · Forward Intensity transplant
0.7847
v3 · time-flattened MLP
0.8325

Out-of-sample AUC · same sample, same split

See validation results by generation and the research team

The full report is provided individually to rating agencies, lenders and investors

We showed the limits of quarterly data in numbers

Closure contagion between market areas barely showed up in quarterly data (hazard ratio 1.036). To tell whether the effect was absent or the data was too coarse, we re-measured the same phenomenon using daily closure dates, based on 135,639 closures of general restaurants in Seoul.

1.85x z=49.9

Same building · same business type

1.24x z=16.2

Same building · different business type

The key is that the effect depends on business type. Demolition or landlord issues would not discriminate by business type. And when the same data is collapsed into quarters, 1.845 drops to 1.585. Quarterly aggregation erases a third of the signal.

So the accurate statement is not “there is no contagion” but “it is not visible at that level at this resolution.”It is a signal worth revisiting once daily or monthly sales data becomes available.

Free data PoC

Shall we validate alternative data together?

We propose a PoC that combines market-area and patent signals with your existing portfolio or scoring model. CEO Byung-Geun Choi, a former fund manager, runs it personally.

Data partnership inquiry

On the next screen you can send via Gmail, your mail app or copy-paste · we reply within 24 hours