AFRM · Forward model · Levchin case

The Levchin case, 20 quarters out

Model as of 2026-08-31

Each segment is projected from its own operating driver, rolled up into consolidated cash flow, and discounted back to a fair value per share. The assumptions below are editable — change them and every number on this page moves with them.

Affirm reports five revenue lines on the face of every shareholder letter - merchant network, card network, interest income, gain on sales of loans, servicing income - and this model carries all five exactly as filed for seven quarters, from FQ2'25 (December 2024) to FQ4'26 (June 2026). Nothing is apportioned and no line is invented, so no quarter is flagged estimated. Quarter labels are FISCAL: '2026 Q4' is the quarter ended 30 June 2026, because Affirm's year ends 30 June. The FQ4'26 lines sum to $1,165,961K against $1,165,960K of reported total revenue - a $1K rounding difference and nothing else. Because this is a fiscal fourth quarter there is a full year to check against, and it checks: the four FY2026 quarters sum to merchant $1,149,932K + card $293,990K + interest $2,047,484K + gain on sale $596,553K + servicing $173,124K = $4,261,083K against $4,261,082K of reported FY2026 revenue, again $1K of rounding. Every line is disclosed; nothing here is derived. What IS assumed is the cost side. Affirm publishes no segment economics at all, so the split of the four transaction-cost lines across the five revenue lines is ours: funding costs and the provision for credit losses are charged entirely to interest income, the loss on loan purchase commitment entirely to merchant network (it arises on the 0% promotional loans merchants pay for), and processing and servicing is split across the four remaining lines. THOSE FIVE MARGINS ARE STILL CALIBRATED TO FQ3'26 AND THIS ROLL-FORWARD DELIBERATELY DID NOT RE-CUT THEM. At FQ4'26 revenue mix they produce $569.4M of RLTC against the $589.1M Affirm reported, a 3.3% shortfall, and 21% corporate overhead then produces a 27.8% adjusted operating margin against the 30.3% reported. The model is therefore conservative on margin by about 2.5 points at the basis. It was left that way on purpose: recalibrating the five to sum to $589M, on the same allocation rule, pushes card network to 49-54% and gain on sale to 92-93%, at or above their own terminal margins of 48% and 92% - so the glide would run backwards. That cannot be fixed without moving terminal margins, which is a re-thesis rather than a roll-forward. It is the first thing a future revision should settle. Two known distortions, stated rather than tuned away. First, this spec carries no seasonality array even though the engine supports one (four factors per vertical, supported by the model engine), so merchant network and gain on sale grow at a smooth sequential rate. That is why the model cannot be solved onto the FQ1'27 revenue guide of $1,190-1,220M: June is Affirm's seasonal peak and September steps down from it, so any growth rate that lands on the September guide is negative and destroys the path. Growth here is therefore calibrated on the FY2027 guide instead, and the near-quarter residual is reported in each vertical. Second, corporate overhead is a fixed share of revenue here, so the operating leverage Affirm promises - the same RLTC yield carrying a lower opex ratio - has to be carried in each vertical's terminal margin instead of in a falling overhead line. One sourcing limitation, stated rather than worked around: the FQ4'26 shareholder letter is furnished to the SEC as an image-only HTML exhibit, and the IR PDF's extractable text drops several letters, so no new verbatim sentence could be taken from it beyond the management quotes captured for the quarter. Two source items therefore still quote the FQ3'26 and FQ2'26 letters where the FQ4'26 print has superseded the figures - funding capacity is now $30.0B against the $28.0B in the quoted sentence, and Affirm Card grew 124% with cardholders up 125% against the 159%/121% in the quoted one. The quotes are accurate as of their own dates; the current figures are in the vertical notes.

Where each case comes from

From cash flow to fair value

The projected path

The published model's 20 quarters. The horizon and sliders above do not change this table.

Every row is projected. A year-over-year change is shown only where the quarter it compares against exists — an em dash means there is no comparable quarter, not a flat year.

Model revisions

Assumptions are marked to reality as each quarter prints. Every change is appended here, with the fair value the model produced at the time, so the model's own history stays visible.