KMX: The Volume Recovery Still Has a Credit Test

The 60-second read

Five things to know

1. More cars, less profit per car. Retail used units rose 13.8%, but retail gross profit per unit fell to $2,105.
2. Overhead leverage returned. SG&A per total unit fell $157 even as absolute SG&A increased 4.6%.
3. Finance helped the recovery. Lower provisions and a $16.6 million loan-sale gain supported CAF income.
4. Capital returns are restarting carefully. Management plans modest Q3 repurchases; authorization is not a cash balance.
5. Price asks for persistence. Our three valuation checks support roughly $55 against the saved $59.10 quote.

Three scenarios, one probability-weighted number

Analyst scenarios, not company guidance

Signposts: what would change our mind

Thresholds are Charged Alpha checkpoints, not management guidance. Reassess the valuation if credit or volume conditions change.

Price and expectations

The stock is recovering faster than the evidence needed to call the turnaround durable. The saved September 29 quote is $59.10, up 4.51% from its previous close. This is a timestamped intraday observation, not a closing price or a promise of the eventual earnings-day return. The quote sits above our $55 fair-value anchor and above the provider's $53.33 average analyst target. Neither comparison means the shares must decline immediately. It means the current price already asks the recovery to continue.

A large price recovery can reflect less risk as well as higher future profits. Keith Barr's strategy, improved value for customers and resumed buybacks may justify investors becoming less pessimistic. But a lower perceived risk premium is not the same thing as more cash distributable to shareholders. We separate those two explanations throughout this packet. A stronger multiple without stronger normalized earnings leaves less room for disappointment in used-car demand or credit costs.

The longer chart provides context rather than a trading signal. Historical reactions use the last available close before an earnings date and the first close on or after that date. That calculation combines earnings with market and industry developments; it does not isolate the causal effect of the release. We exclude today's unfinished session from that reaction series. The current quote and historical daily closes are separate observations, with their own timestamps and sources. These distinctions matter when a stock moves sharply before an eventual close.

The reported quarter

The second quarter ended August 31 delivered $7.878 billion of revenue and $165.287 million of net income. Revenue grew 19.5%, while reported diluted earnings rose to $1.16 from $0.64. A falling share count helped earnings per share grow faster than total earnings, but most of the improvement came from higher profit. The current release also shows first-half revenue of $15.891 billion and earnings of $350.914 million, which helps reconcile the quarter with the preceding release.

The sales increase is not simply a higher sticker-price story. Retail used units rose 13.8% and comparable-store used units increased 13.0%. Wholesale units increased 15.9%. Higher retail selling prices contributed as well, while the company deliberately accepted less gross profit per retail vehicle. The combination is strategically coherent: win more transactions, then spread operating costs over the larger base. It remains economically attractive only if the extra units bring enough incremental gross profit to cover the margin concession and associated credit exposure.

FMP's earnings calendar lists $0.73 estimated EPS and approximately $7.092 billion estimated revenue for this release. We preserve these as provider estimates, not management guidance. Older provider records mix adjusted and GAAP earnings, and some historical standardized operating lines disagree with actual company tables. Therefore our historical GAAP series comes from saved SEC exhibits. In particular, first-quarter fiscal 2027 gross profit is $854.414 million, and fourth-quarter fiscal 2026 GAAP EPS is a loss of $0.85. Using the provider's alternative values would distort the turnaround comparison. A headline surprise should never substitute for understanding which earnings measure was expected.

Two businesses, different economics

CarMax reports Sales Operations and CarMax Auto Finance as its two reportable segments. The retail, wholesale and other gross-profit categories below are useful operating components within the sales business, not additional reportable segments. This distinction prevents an apparent segment margin from being mistaken for a fully allocated operating margin. The release does not allocate all indirect overhead to CAF, so a direct comparison of its income with retail gross profit would also be misleading.

Retail generated $478.6 million of gross profit, wholesale $137.6 million and other activities $183.3 million. Retail profit rose because volume more than offset the decline in profit per vehicle. Wholesale volume grew strongly, but gross profit was almost unchanged because profit per unit declined to $858. Other gross profit benefited from extended protection plans and improved service economics. These businesses provide real support, but warranty-related revenue and cancellation estimates deserve monitoring rather than an assumption that every recent improvement repeats indefinitely.

CAF income of $135.6 million is already net of its interest expense, credit provisions and direct expenses. It is not revenue and should not be stacked with retail vehicle sales to suggest an enlarged top line. The company financed a larger share of Tier 2 business, increasing potential lifetime finance income while also requiring upfront provisioning. The strategic question is whether better underwriting and scale produce adequate returns through a full credit cycle. One strong quarter of provisions cannot answer it. Sales growth, loan growth and credit risk must be evaluated together, especially when management is expanding the range of borrowers it serves.

Where the earnings improvement came from

The clearest operating evidence is the dollar bridge from the year-ago quarter. Gross profit added $81.8 million and CAF income added $32.9 million. Higher selling, general and administrative costs absorbed $27.5 million, with smaller offsets from depreciation and corporate interest. Increased other income added another $14.9 million. Together those movements reconcile the $96.0 million increase in pretax earnings. The reconciliation treats expense signs consistently and does not silently add back depreciation that the company separately reports.

SG&A increased in absolute dollars even though the efficiency story improved. Cost per combined retail and wholesale unit fell from $1,778 to $1,621, a reduction of $157. That is operating leverage: a larger sales base carries each dollar of overhead further. Compensation comparisons and variable costs still pushed the total expense line upward. Investors should distinguish a lower cost per transaction from an already achieved reduction in every expense category. The promised $200 million exit-rate savings target is also a run-rate objective, not a statement that the full amount has already flowed through annual earnings.

The profit bridge supports a recovery thesis without proving a new steady-state margin. Retail pricing actions reduce gross profit per car now in exchange for better demand and throughput. If transactions slow before the cost structure adjusts, leverage reverses. If the stronger throughput persists while reconditioning and logistics improve, profit can compound with less promotional support. Our valuation requires evidence of the latter over subsequent reporting periods. It does not capitalize the entire latest EPS increase as a permanent annual growth rate.

Nine repeatable earnings-quality checks

We rate earnings quality as mixed rather than dismissing the improvement. The release says the credit-loss provision declined by $28.8 million to $113.4 million as older-vintage performance compared more favorably. CAF also recognized a $16.6 million gain on sale of auto loans. Separately, other income rose to $18.6 million, primarily from unrealized investment gains. These are distinct channels. Lower provisioning reflects portfolio expectations; loan-sale gains reflect transaction economics; investment marks reflect asset values outside the core vehicle transaction.

A conservative sensitivity removes the full loan-sale gain and all other income using this quarter's effective tax rate. That reduces EPS by about $0.18 to about $0.98. This is our analytical sensitivity, not a company-defined adjusted result, and excluding all other income overstates the adjustment if some portion is recurring. We do not simultaneously remove the provision decline in that same sensitivity because doing so would assume that every favorable credit development is temporary. Instead we stress credit costs separately in the bear case.

The nine checks below are deliberately repeatable. SBC has an economic cost even when it does not require an immediate cash payment. Reported operating cash flow is difficult to compare with net income because originating and selling auto loans affects the cash statement. Trade receivables should not be combined with auto-finance receivables to calculate retail collection days. Inventory days use vehicle cost of sales as a directional proxy and do not measure loan duration. The lack of a standardized adjusted-income series is itself a disclosure constraint: we preserve GAAP history, identify specific charges and avoid inventing a smoother management earnings measure.

Corporate cash versus finance funding

The company has $170.5 million of unrestricted cash at quarter-end, while separate restricted cash largely relates to customer loan collections and securitization arrangements. Treating all restricted balances as freely available buyback cash would materially overstate flexibility. Corporate debt is approximately $1.681 billion, whereas nonrecourse notes total approximately $15.694 billion. Those notes finance the auto-loan portfolio; they are not an interchangeable claim on the retail business's cash flows. Equally, they are not harmless simply because they are nonrecourse: credit losses and funding-market access still affect the economics of CAF.

First-half operating cash flow of $900.5 million less capital expenditures of $180.4 million equals $720.1 million. This arithmetic is a conventional cash-flow subtotal, not our normalized owner-earnings estimate. Loan-sale proceeds of $579.9 million are included in operating cash flow, and changes in originated receivables, collections and inventories can produce large swings. The full loan-sale proceeds cannot be treated as a gain, nor can they simply be subtracted without considering the financing and origination activity around them. For that reason our equity cash model uses a transparent normalized assumption instead of annualizing the reported subtotal.

Management plans to resume buybacks at a modest level in the third quarter, with $1.31 billion of authorization remaining. Authorization is permission, not cash and not a commitment to retire that amount of stock. The latest quarter contained no program repurchases. Better leverage and profitability create capacity to return cash, but paying too much for shares would weaken returns on that capacity. A sensible scorecard asks whether buybacks occur below a conservative estimate of value while the credit portfolio and retail inventory remain appropriately funded. Our $44 entry threshold is a valuation margin-of-safety rule, not management's repurchase target.

Three routes to value

Our central estimate is $55 per share, with high uncertainty and a HOLD rating at the saved $59.10 quote. This is an estimate of business value under explicit assumptions, not a forecast of where the market will trade next week. The three approaches converge around the mid-fifties because they use related but distinguishable descriptions of the same business: sustainable earnings, distributable equity cash and the balance-sheet capital supporting those earnings. Convergence does not create independent certainty when all three depend on the durability of the operating recovery.

The earnings route uses $3.60 of normalized annual EPS at 15 times, producing $54. The EPS input is an analyst assumption below a simple annualization of the latest quarter after our gain sensitivity. It allows for seasonal variation and ongoing credit costs. The equity cash route begins with $475 million of normalized distributable annual cash, grows it 5% for five years and 3% thereafter, and discounts it at a 10% cost of equity. The resulting $7.609 billion equity value divided by 141.934 million shares produces $53.61. Because the cash input is already an equity measure after financing and reinvestment needs, we do not subtract securitization debt again.

The book-value check applies 1.25 times to $44.42 of equity per outstanding share, producing $55.52. That premium assumes the company can earn adequate returns on its retained capital; book value is not liquidation value and does not guarantee recovery of loan assets at stated amounts. At $59.10, the stock requires about $3.94 annual EPS at our 15-times multiple. Our base assumption is $3.60. The bear, base and bull values are $33, $55 and $78, weighted 25%, 50% and 25%, for $55.25. The rounded scenario endpoints represent stressed analytical cases, not precise outputs of management guidance. A 20% discount to central value gives a $44 potential entry threshold, conditional on fundamentals remaining intact.

Our call versus Wall Street

The fresh consensus snapshot shows a Hold rating, with one strong buy, ten buys, twenty-three holds and two sells. Its average target is $53.33, the median is $50 and the range is $43 to $74. These are aggregated opinions with different publication dates and methods. A newly fetched consensus is not necessarily a set of newly rewritten analyst models. We therefore show the distribution and dates rather than describing every target as an immediate response to this quarter.

The latest available target-news records include Stephens at $74 on September 25 and Morgan Stanley at $47 on September 23. Both precede the September 29 earnings release. Their wide gap illustrates disagreement over sustained earnings power and the multiple investors should pay for it. Our $55 estimate sits much closer to the aggregate than to the bullish endpoint. We align with the broad Hold stance, but our work should be judged by its assumptions and checkpoints rather than agreement with a vote count.

Provider annual consensus EPS for fiscal 2027 is $2.785 with fourteen EPS analysts in the saved response. That estimate is lower than our $3.60 normalized earnings-power input, but the two are not interchangeable: consensus is a dated specific-year forecast, while our input represents a sustainable scenario used for valuation. The forecast may also be slow to incorporate the latest print. A target-news record whose headline named one broker while the structured broker field named another was excluded from our firm-specific discussion. This is a practical reminder that even licensed provider data requires consistency checks. We do not average contradictory broker labels into apparent precision.

Management promises and evidence

Keith Barr describes the four-pillar strategy as improving the customer offering, making the buying experience easier, adding value and running lean. Those categories become useful for investors only when linked to measurable operating changes. This quarter provides evidence of better demand and SG&A leverage, but retail and wholesale profit per unit weakened. Management is choosing a different balance between margin and throughput rather than delivering improvement in every line at once. We judge that choice by total profit and cash returns over time.

The company says it remains on track for $200 million of SG&A exit-rate savings by the end of fiscal 2027. A virtual strategic update on November 3 should provide more detail on initiatives and milestones. This is a separate event from the September 29 earnings call and should not be described as the earnings release date. The next earnings release is planned for December 17, covering the quarter ending November 30. Dates are management plans and can change; the point is to establish the next concrete opportunity to test the thesis.

The historical Starboard director contest is resolved for eligibility purposes. The April agreement withdrew the nominations and the June annual meeting elected the agreed directors. That resolution is not an endorsement of execution or a reason to ignore governance. The operational scorecard still asks whether management can translate the refreshed board and strategy into better returns. We would ask how much unit growth comes from pricing, what repeatable contribution loan sales should provide, how Tier 2 expansion changes required capital, what portion of savings is structural, and what valuation discipline will govern restarted buybacks. Answers that distinguish timing benefits from recurring economics would improve confidence more than another broad expression of optimism.

The cases against our view

Credit normalization is the most direct threat to the apparent earnings improvement. A $50 million increase in quarterly pretax credit costs would reduce earnings by approximately $0.26 per diluted share at the current effective tax rate and share count. That is a sensitivity rather than a predicted loss. It shows why a modest change in loan performance can matter more than a small movement in vehicle gross profit. Expansion in Tier 2 originations increases the importance of tracking vintages rather than relying on a single aggregate reserve ratio.

Vehicle-price and affordability risk follow closely. Tariffs, interest rates and household finances influence what customers can afford, while used-vehicle depreciation affects wholesale recovery values. Retail volume can remain strong for a time because the company offers better prices, yet the associated margin sacrifice may prove harder to reverse. Fixed store and technology costs amplify that tension. A competitor's aggressive pricing can also force CarMax to choose between defending unit share and protecting near-term profit.

The strongest bull case is that the company has finally repaired its value proposition and now gains enough traffic to spread its existing infrastructure over a much larger base. Better EPP economics, more efficient servicing and disciplined CAF underwriting could then lift earnings beyond our central assumptions. The strongest bear case is that incentives and favorable credit comparisons are front-loading the recovery while the stock already discounts durable success. Neither case can be settled by the quarterly EPS growth rate alone. Our Hold stance reflects an improving business with limited valuation cushion, not a claim that the strategy cannot work. Position sizing and diversification remain investor-specific decisions outside this packet.

What could change the estimate

The November 3 strategic update is the next major opportunity for management to describe the bridge from current recovery to sustained returns. Useful disclosures would quantify milestone dates, incremental capital requirements and the distinction between gross savings and net earnings benefits. A presentation filled with long-term aspirations but few measurable commitments would not strengthen the valuation. We will compare whatever targets are announced with the baseline in this packet rather than retrospectively changing that baseline to make the strategy look successful.

The planned December 17 earnings release supplies the next operating test. We want comparable-store used-unit growth to remain positive while SG&A per total unit stays at or below the current level and retail gross profit per unit does not collapse. We also want enough credit detail to distinguish improving portfolio performance from gain-on-sale timing. If management provides a new reporting metric, it must be reconciled to the existing series before becoming a substitute for the old measure.

Buyback execution and the financing environment are continuing catalysts rather than one-day events. Repurchases below fair value can improve per-share economics, but they do not repair weak underwriting or justify borrowing indiscriminately. Lower market rates might support affordability and funding costs, while faster depreciation could work in the opposite direction through collateral values. The dated signposts at the top of the packet convert these broad forces into observable thresholds. They are our analytical checkpoints, not company guidance and not automatic buy or sell orders. A later report should grade them honestly, including checks that deteriorate even if the share price rises.

Data, sources and method

The nine-quarter history is extracted from actual quarterly SEC earnings exhibits, including year-ago comparison columns where appropriate. Dollar statements are converted from thousands to millions for presentation, while per-share figures and unit counts retain their original dimensions. Column headers were inspected before extracting values. Prior-year comparison data must not be mixed with current-quarter values, and first-half cash-flow amounts must not be mistaken for a standalone quarter. Raw HTML, full submissions, extracted tables and source hashes remain attached to this episode.

The fiscal 2026 fourth quarter contains an impairment and restructuring charges, so reported EPS and company-adjusted EPS differ materially. The historical chart deliberately keeps the GAAP loss; replacing it with adjusted earnings would hide the event that affected shareholders' reported results. Our current-quarter gain sensitivity is separately labeled and is not a replacement GAAP series. The same discipline applies to the finance operation: CAF income is after certain funding and credit costs, while retail gross profit precedes corporate operating expenses.

Valuation assumptions are ours. Probabilities express analytical judgment and do not come from option prices or a statistically calibrated forecast. The equity cash estimate is normalized rather than copied from reported operating cash flow less capital expenditure. Market prices, analyst counts and targets reflect the saved provider responses at their indicated times. No same-quarter predecessor research packet was found; earlier channel videos cover fiscal 2027 first quarter and fiscal 2026 fourth quarter. Their headline beat-and-stock-reaction angles are deliberately not repeated here. This packet establishes measurable checkpoints for the next fiscal quarter instead of claiming a track record that has not yet been observed.

• Q2 FY2025: SEC earnings exhibit
• Q3 FY2025: SEC earnings exhibit
• Q4 FY2025: SEC earnings exhibit
• Q1 FY2026: SEC earnings exhibit
• Q2 FY2026: SEC earnings exhibit
• Q3 FY2026: SEC earnings exhibit
• Q4 FY2026: SEC earnings exhibit
• Q1 FY2027: SEC earnings exhibit
• Q2 FY2027: SEC earnings exhibit
• FY2026 Form 10-K: structure, financing and audit
• FMP saved quote, daily prices, earnings calendar, annual estimates, targets and ratings; exact sanitized request URLs and hashes in research source manifest.