- Public-market signals decay fast because securities are homogeneous, continuously priced, and cheap to trade — arbitrage closes the gap almost immediately.
- Asset-based finance signals persist because loans are heterogeneous, transactions are episodic, and the barrier to entry is infrastructure, not information.
- Signals here decay over reporting cycles, not trading sessions — success depends on learning accurately, not reacting fastest.
- The advantage compounds rather than decays: every realised outcome improves the model, so the platform’s edge grows the longer it operates.
Quantitative signals in public markets decay rapidly: once a predictive relationship is known, capital compresses it away. Asset-based finance operates under different economics. Information is revealed through realised cash flows rather than continuous prices, loans are heterogeneous, transactions are episodic, and the barrier to entry is infrastructure rather than information. The result: predictive signals whose half-lives are measured in reporting cycles, and a research platform whose accumulated learning compounds faster than its signals decay.
The decay problem in public markets
Quantitative investing in public markets is defined by rapid signal decay. Once a predictive relationship becomes widely known, capital flows toward it and expected excess returns compress. The academic record is consistent on this point: McLean and Pontiff (2016) find that abnormal returns on published equity anomalies decline materially once a strategy is documented in the literature. In highly liquid markets, alpha is a depreciating asset.
In our view, this outcome is structural rather than incidental. Public securities trade continuously on common information, transaction costs are low, and the instruments are standardised, so new information is incorporated into prices almost immediately. Competitive advantage has increasingly migrated toward execution speed rather than proprietary information—a race in which the marginal advantage narrows every year.
Asset-based finance operates under fundamentally different economics.
Information emerges through cash flows, not prices
Loans are contractual cash-flow instruments rather than continuously repriced securities. Their true risk is revealed gradually through borrower behaviour. A deterioration in affordability, a weakening regional labour market, or a loosening of underwriting standards does not appear instantly in any market price; it manifests over months and years through delinquency, prepayment, and recovery performance. The economic process itself unfolds slowly.
Predictive signals in asset-based finance therefore exhibit materially longer half-lives than comparable signals in liquid markets—they decay over reporting cycles rather than trading sessions. This changes the economics of quantitative investing at its foundation. Success depends less on reacting fastest and more on learning most accurately from realised outcomes.
Public Markets
Asset-Based Finance
Why arbitrage cannot eliminate these signals
We believe the persistence of these signals reflects market structure rather than temporary inefficiency, for three reasons.
First, loans are heterogeneous. Every mortgage or consumer loan differs in borrower profile, collateral, servicing arrangements, and contractual terms. There is no homogeneous instrument against which mispricing can be instantaneously arbitraged, as there is in listed equities.
Second, transactions are episodic rather than continuous. Portfolios trade infrequently and only after substantial due diligence, legal negotiation, and servicing transition. These frictions prevent rapid convergence between transaction prices and underlying credit fundamentals, while also constraining the pace at which new capital can enter.
Third—and most importantly—the informational advantage lies not in access to information but in the ability to transform it into investable knowledge. European loan-level disclosures contain billions of loan-level observations across residential mortgages, consumer loans, auto finance, and SME lending. Although much of this information is publicly available through regulatory reporting, it remains fragmented, inconsistent, and operationally difficult to use. Schema harmonisation, data quality validation, borrower linkage, collateral enrichment, and macroeconomic integration must all precede predictive modelling. In our experience that is the better part of a year of dedicated engineering before a first model can be estimated and validated across the full reporting history. The barrier to entry is infrastructure, not information.
Infrastructure compounds
MinVaris operates quantitative underwriting as a continuous learning system. We normalise heterogeneous disclosures into a single analytical framework across jurisdictions, originators, and asset classes. We then estimate borrower and collateral dynamics using models validated across time periods and lending institutions, enriching the regulatory record with macroeconomic, geographic, and collateral information. The resulting probability distributions inform pricing, structuring, and portfolio construction directly, and realised defaults, prepayments, and recoveries feed continuously back into the models.
Each completed investment contributes additional realised performance that improves future underwriting. Unlike a trading signal, which weakens as competitors discover it, the research platform becomes more valuable as performance history accumulates. A competitor may reproduce a modelling technique within months; reproducing years of cleaned loan-level history, originator relationships, servicing outcomes, and calibration experience is a materially harder problem. The durable advantage lies less in algorithms than in accumulated learning. The models improve because the evidence base improves, and the evidence base compounds with every repayment, delinquency, and recovery observed.
This distinction is fundamental. In public markets, predictive signals often decay faster than investors can learn from them. In asset-based finance, realised borrower behaviour accumulates more quickly than the underlying relationships change. The learning process therefore compounds faster than the signals decay.
Implications for institutional investors
The persistence of quantitative signals in asset-based finance is a consequence of market architecture. Because information is revealed through realised cash flows rather than continuous market prices, the value of superior underwriting compounds over time. Data engineering, historical depth, operational infrastructure, and realised performance become cumulative assets rather than transient advantages.
For institutional investors, this makes asset-based finance unusual among quantitative asset classes. Excess returns arise not from discovering signals before everyone else, but from building an investment platform whose accumulated learning compounds over time. The competitive advantage is not a model. It is the system that continuously improves the model.
In asset-based finance, learning is.