What SoFi's fair-value marks say about prime personal loan pricing

SoFi carries its personal loan book at fair value, one of the few public windows into how prime unsecured paper is priced. Q3 shows strong growth, lower losses, and a discount rate that rose less than rates did.

Originally posted on LinkedIn, October 2026. Expanded here.

This follows up my LightStream post. If that sale was a single transaction price, SoFi's quarterly disclosures are the closest thing to a running mark on prime personal loans. SoFi carries its loans at fair value and discloses the inputs. Few lenders do.

Here is what Q3 shows.

Growth

  • $12.6B originated in the quarter, up 69% year over year.
  • The balance-sheet book reached $30.9B, up 59%. More than half of it was originated this year.

The second point matters most. A book that is mostly this year's originations hasn't been through its loss curve yet. Personal loan losses typically peak well after origination, so headline loss rates on a fast-growing book flatter it.

Credit

  • All-in net charge-offs fell to 3.4%, from 4.4% in Q1.
  • The assumed default rate rose to 5.1%.

These two numbers point in opposite directions, and the gap makes sense. Realized NCOs are a ratio, and a fast-growing denominator of unseasoned loans pushes it down. The forward default assumption is SoFi's own view of what the book will actually do. The real credit test comes as these young vintages season.

Valuation

SoFi marks the book at ~103.3, down from ~104.7 in Q2. The interesting part is the discount rate:

Q2 → Q3
2-year SOFR +76 bps
SoFi discount rate +59 bps (4.97% → 5.56%)
Implied spread change ~17 bps tighter

Rates rose sharply, but the discount rate rose less, so the implied credit spread compressed by about 17 bps. That isn't obviously aggressive. It lines up with where recent consumer ABS has been pricing. Spreads on prime unsecured paper have been tightening in the securitization market too.

Why it matters

Fair value marks have three moving parts: a rate curve, a spread, and a credit assumption. When rates are this volatile, the mark can move a lot without anything changing in the loans themselves. For investors in whole-loan forward flow, ABS residuals, or fintech equity, it's worth separating the three:

  1. How much of the mark change is rates? Mostly outside management's control and hedgeable.
  2. How much is spread? Market-observable, and ABS pricing is the check.
  3. How much is credit? This is where judgment and model risk live, and it will surface as vintages season.

With interest rates this volatile, fair value marks will be worth watching. I'm interested in other perspectives on where prime unsecured discount rates should sit.