PRDO — Perdoceo Education Corporation

Deep-value analysis — PRDO (Perdoceo Education Corporation) · REVISIT

Nasdaq: PRDO · reference price … (close 09/04/2026, yfinance) · 62.577mn shares (10-Q Q2 2026, balance sheet) · market cap …mn · reporting and quoted in USD (no conversion risk)

Regime: REFRESH over the 08/14/2026 thesis. Delta pack: delta-PRDO-20260904.md. Reason for revisiting: one falsifier (organic revenue growth) had remained unverifiable — the mechanical triage needed two consecutive quarters and was missing one.

Primary sources: 10-K FY2025 filed 02/19/2026; 10-Q Q1 2026 filed 05/07/2026; 10-Q Q2 2026 filed 08/06/2026; 8-K + Ex-99.1 08/06/2026; DEF 14A 04/09/2026; Form 4 filed 08/12/2026 and two Form 144s filed 08/11/2026; SEC XBRL companyfacts and companyconcept, CIK 0001046568. Market data: yfinance 09/04/2026. Consensus context: research-PRDO-20260905.md, with corrections noted in the text.

What I inherit and what I re-derived

The refresh regime shortens the reading work, not the report’s structure. I state the boundary explicitly.

Inherited unchanged (no new fact touches these — between 08/14/2026 and 09/05/2026 the company filed no 10-K, 10-Q, or 8-K, verified directly in the issuer’s EDGAR filing stream):

• the business description, moat structure, and competitive position (ch. 2);
• the capital-allocation history, incentive structure, and DEF 14A analysis (ch. 3);
• the line-by-line balance sheet at 06/30/2026 and the Q2 operating-profit breakdown (ch. 4);
• the entire O’Glove analysis — receivables, provision, debt, discretionary expenses, accruals (ch. 5);
• the Thorndike grid and the 1.5/5 Outsider score (ch. 6);
• the nine accounting red-flag items (ch. 7);
• the FCF bridge and the … thousand FCFE definition (ch. 8) — the relevant TTM is still Q3 2025 – Q2 2026, since there’s no new quarter;
• the valuation assumptions: OE …M, g1 3.5%, r 11%, gt 2.0%, net cash −…M.

Re-derived in this run:

1. The growth falsifier, verified directly in XBRL RevenueFromContractWithCustomerExcludingAssessedTax across eight quarters (not from aggregators) — the result is more nuanced: the status remains “undecided,” but for a different reason than the triage assumed.
2. Three insider filings the August report didn’t contain — a Form 4 (08/12/2026) and two Form 144s (08/11/2026). These are the only new information filed with the SEC since the thesis, and they touch directly on the skin-in-the-game chapter.
3. The borrower-defense loan-cap litigation — moved materially in the company’s favor; the government’s appeal deadline expired 08/24/2026 with no appeal. I lower the probability of pre-mortem scenario 3.
4. The Monte Carlo simulation, re-run. The automated 09/04 run used oe and nd in thousands instead of millions and produced a median MOS of … — a unit error, not a result. The correct run is in the file mc-PRDO-20260905.json.
5. All five valuation models, recalculated against the … price instead of …

The price move (… since the thesis) is context, not thesis. I build nothing around it. But it mechanically consumes the margin of safety, and this report says by how much.

Executive summary

The thesis, unchanged in substance. Perdoceo is a for-profit online university operator — Colorado Technical University (54% of H1 2026 revenue), American InterContinental University System (26%), University of St. Augustine for Health Sciences (19%) — trading at …× guided 2026 GAAP profit and at 6.7× enterprise value relative to TTM EBIT, because a third of the market cap is net cash. At 06/30/2026 the company had …M cash, restricted cash, and short-term investments against interest-bearing obligations of …M (finance lease …M + failed sale-leaseback financing …M). Net cash of …M = …/share, i.e. 32.2% of the market price (was 34.1% at …). The remaining business — …M TTM EBIT, …M ex-cash free flow — is bought at an enterprise value of …M, i.e. 7.3× operating free flow and a 13.7% yield on EV (was 6.7× and 14.9%).

The falsifier that triggered the revisit: verified, but not resolved. The triage required two consecutive quarters of revenue growth and found only one. I checked the full series directly in XBRL: Q1 2026 …M versus …M (…) and Q2 2026 …M versus …M (…). So the two consecutive quarters DO exist in the filings; what was missing was the comparison quarter in the automated feed, not in reality. Both are positive, both below the historical pace, and the direction is deceleration (… → …). The falsifier’s threshold isn’t met — the thesis isn’t refuted — but it isn’t confirmed either: total enrollments at 06/30/2026 are 46,830 versus 46,500 (…), and all of the growth comes from USAHS (…), against CTU … and AIUS …. The honest verdict: the falsifier stays open until the Q3 reporting, scheduled for 11/09/2026, which is the first quarter under the identity-verification checks imposed by the Department of Education for academic year 2026-27.

What genuinely changed since 08/14. Three things, all individually small, two in opposite directions:

1. Insiders sold in the post-earnings window. On 08/11/2026, five days after Q2 results, board chairman Gregory L. Jackson exercised 11,271 options at … and immediately sold them at … (Form 4 filed 08/12/2026), and officer John Kline notified the sale of 13,511 shares for … from restricted shares granted March 2024 and performance shares vested 03/14/2026 (Form 144, 08/11/2026). Small amounts in absolute terms; a relevant signal in context, since they reinforce exactly the criterion August’s analysis had scored at zero.
2. Loan-cap risk fell materially. In American Association of Nurse Practitioners v. McMahon (consolidated with PA Education Association v. Dept. of Education, D.D.C.), the government’s deadline to appeal the 06/24/2026 stay expired on 08/24/2026 with no notice of appeal on the docket. The cases proceed on an expedited summary-judgment track, with filings closing on 12/04/2026. The stay of the restrictive “professional program” definition remains in effect, and the Department has issued interim guidance treating eligible nursing programs as professional programs.
3. The price rose 5.8% (… → …) with no new filing. Arithmetic consequence: the stock moves from 2.2% below the Graham Number to 3.7% above it (… at EPS … and BVPS … per the tracker), and the simulation’s pessimistic decile moves from positive to negative.

Estimated value. The five triangulated models give an MOS range of ** … to …**, with the median at … (was …) and the base DCF at … (was …). The Monte Carlo simulation with the same assumptions (20,000 scenarios, OE …M, g1 3.5%, r 11%, gt 2.0%, net cash −…M) gives a median intrinsic value of … P(undervalued) …% and P10 = …. Public consensus — three converging sources: price target … (the Simply Wall St narrative updated 05/16/2026, with r = 7.11% and a forward P/E of 12.87×), … Barrington Research (11/05/2025), and the tracker’s discount-to-consensus figure (… at … implying a target of …) — sits above my base model, but starts from a discount rate nearly four points lower than mine.

Verdict: INTERESTING — small position, but the price has eaten almost all of the modeled margin. I maintain August’s verdict because the facts haven’t deteriorated, and one of the three risk tails has actually thinned. I don’t upgrade it, because at … the most conservative model that still makes economic sense (EPV with zero growth) gives only …, and the simulation’s pessimistic decile is below zero. Concretely: I don’t add above ~…; below … the thesis becomes comfortable again.