SAH · Sonic Automotive, Inc. · Stock research

Current Thesis

Sonic beat on both headline lines on 2026-07-30 — revenue $3.934B vs $3.772B consensus, adjusted EPS $1.82 vs $1.74 — and the stock went the other way. The reason sits one line below the headline: adjusted diluted EPS fell 17% YoY, franchised segment income fell 23% YoY to $70.7M, EchoPark segment income fell 38% YoY to $7.2M, and management cut EchoPark's full-year total GPU guidance to $3,100–$3,300 from $3,400–$3,600. The gross-profit-per-unit normalization the market had been pricing as finished is still running.

The price record is unambiguous about how that landed. The last completed daily close was $82.44 on 2026-08-07 against a 52-week high of $112.66 — a 26.8% drawdown — while the three-month return is +2.0%. The entire June–July advance has round-tripped to roughly where the stock traded in early May. RSI(14) sits at 36.7.

Bullish and bearish views on Sonic Automotive, Inc.

The model's bull view on Sonic Automotive, Inc. (SAH), in brief: Fixed operations and F&I together produced more than 75% of total gross profit in Q2 2026 (reported 2026-07-30). The bear view: Every per-unit line compressed in Q2 2026. Both cases follow in full.

Bull Case

• Fixed operations and F&I together produced more than 75% of total gross profit in Q2 2026 (reported 2026-07-30). Same-store fixed operations gross profit +2% YoY, warranty +3%, customer pay +1% — the recurring-revenue layer held while transaction margin compressed.
• Powersports is compounding off a small base: Q2 revenue $73.5M (+53% YoY), gross profit $19.7M (+58%), segment income $2.3M against roughly breakeven a year earlier, adjusted EBITDA $4.9M (+145%). Five Harley-Davidson stores acquired in April 2026 are guided to add ~$100M annualized revenue.
• EchoPark volume is scaling even as unit economics compress: 19,601 retail used units (+17% YoY), revenue $582.9M (+15%), record Q2 gross profit $64.3M. Management raised the FY26 volume-growth target to 12–15% and held FY26 adjusted EBITDA at $35–40M, with footprint expansion starting in Q4 2026.
• Share count is shrinking fast.
• The published sell-side has not capitulated. Citi Buy PT $115 (2026-08-03), Needham Buy PT $139 (2026-07-31), UBS Buy PT $109 (2026-07-10), BofA Buy PT $108 (2026-07-09), Barclays Equal-Weight PT $92 (2026-07-15) — all above the 2026-08-07 close of $82.44.
• Balance sheet at 2026-06-30: cash and floor plan deposits ~$294M, total available liquidity ~$676M. Quarterly dividend of $0.41 declared 2026-07-30 ($1.64 annualized), record 2026-09-15, payable 2026-10-15.

Bear Case

• Every per-unit line compressed in Q2 2026. Same-store new-vehicle GPU −16% YoY to $2,872; same-store used GPU −13% to $1,401; F&I gross profit per retail unit −4% to $2,619; EchoPark combined used + F&I GPU −12% to $3,292. Volume is buying revenue while price gives margin back.
• Franchised same-store gross profit fell 3% YoY on same-store revenue up 2%. Consolidated gross profit grew 2% on revenue up 8%.
• Adjusted SG&A was 72.0% of gross profit versus management's stated low-70s ambition, and the denominator is the line under pressure.
• The guidance revision at the 2026-07-30 print moved the wrong way where the differentiated growth story lives: EchoPark total GPU cut roughly $300 at the midpoint.
• Even the annuity softened — fixed operations margin fell 30 basis points to 51.0%.
• CEO David Smith's own call framing was that affordability pressure may push consumers to repair current vehicles instead of replacing them. That helps parts and service and works against unit volume and new-vehicle GPU simultaneously.
• JPMorgan has cut twice into the move: Underweight with PT to $76 on 2026-07-13, then to $72 on 2026-08-04 — the only target published since 2026-07-09 that sits below the last close.
• Reported and adjusted results diverge sharply: reported net income +226% YoY, adjusted net income −23% YoY. Screens reading the GAAP line will misprice the trend.

Setup & Price Structure

Life-cycle: DEAD. The narrative failed on a specific date and the structure broke with it. Shares set a 52-week high of $112.66, met a beat-and-guide-down on 2026-07-30 — press coverage of the results presentation reported a roughly 9% single-day decline, with further weakness reported 2026-08-02 — and closed at $82.44 on 2026-08-07, 26.8% below the high. The three-month return of +2.0% means the whole advance has been given back to its launch area. This label describes the narrative leg, not the enterprise: revenue grew 8% YoY and two segments printed Q2 records.

Structure observables:

• RSI(14) 36.7 as of 2026-08-07 — weak, but above the sub-30 zone where exhaustion usually registers. No higher low has formed since 2026-07-30; there is no base to reference yet.
• The pre-advance shelf near $80 — the area the June–July move launched from — sits directly beneath the current close. The +2.0% three-month return is what puts it there.
• Positioning and crowding evidence runs the opposite way from a typical late-stage momentum name. Price is below, not extended above, its rising averages, so distance-from-MA crowding does not apply here.
• What does apply: sell-side target dispersion of $72 (JPMorgan, 2026-08-04) to $139 (Needham, 2026-07-31) on the same quarter — a 1.9x spread, which is where estimates go when normalized earnings power is genuinely unsettled.
• The largest identified incremental buyer of the stock has been the company: ~2.1M Class A shares in Q1 2026, 6% of shares out, with ~$528M of authorization remaining as of April 2026. The buyback was running hard before the drawdown; whether it accelerated into it is not yet observable and will show in the Q3 10-Q.
• No forced repricing event exists in the near term. The next scheduled company disclosure is a dividend record date, and the next print is roughly eleven weeks away.

Catalyst Calendar (next 30 days)

• ~2026-08-18 (est.) — Cox Automotive Manheim Used Vehicle Value Index mid-month read. Wholesale used values set the cost side for both the franchised used desk and EchoPark.
• ~2026-09-03 (est.) — August US light-vehicle SAAR from industry/automaker reporting. Direct read on whether affordability is throttling unit volume beyond Q2's flat same-store new units.
• ~2026-09-04 (est.) — Manheim full-month August index.
• 2026-09-15 — Record date for the $0.41 quarterly dividend declared 2026-07-30 (payable 2026-10-15). Eight days past the 30-day window and the only hard-dated company event on the near calendar.
• No company reporting event falls inside the window. Q3 2026 results are expected ~2026-10-22 (est.); EchoPark footprint expansion is guided to begin in Q4 2026.

What Would Change Our Mind

The recovery case rests on one shelf: the roughly $80 area from which the June–July advance launched, which the 2026-08-07 close of $82.44 is sitting directly on. A weekly close below $78 would say that shelf failed and the round-trip has extended into a fresh downtrend leg — at which point JPMorgan's $72 becomes the reference rather than an outlier.

On the other side, the evidence that would flip the read constructive is specific and dated: consolidated new-vehicle GPU printing inside the $2,850–$3,000 FY26 band and used GPU inside $1,350–$1,450 at the Q3 report (~2026-10-22, est.); adjusted SG&A back below 71% of gross profit; EchoPark segment income above Q2's $7.2M rather than down another 38%; and a multi-week higher low built above $80 instead of a single reflex bounce. A Q3 print that comes and goes with the GPU bands intact and the repurchase pace still near Q1's ~2.1M shares would materially weaken the "normalization is still running" case.

A reiteration or cut of the $35–40M EchoPark FY26 adjusted EBITDA range at that print is the cleanest single datapoint, because it is the only figure management has committed to in writing for the segment carrying the growth narrative.

Correlation Notes

• Sonic trades inside the franchised-dealer complex — ABG, AN, GPI, LAD, PAG. Q2 2026 GPU compression is an industry condition; a peer print showing per-unit stabilization would re-rate the group and a second peer confirming compression would deepen it.
• EchoPark's standalone used channel overlaps CVNA's demand pool, and both are priced off Manheim wholesale values, which reach the P&L with roughly a quarter's lag.
• Rate sensitivity is two-sided: consumer auto-loan APRs gate affordability on the revenue line, while floor plan interest sits directly against inventory in the expense line (2026-06-30: 56 days' supply new, 35 days used franchised, 47 days EchoPark).
• Tariff pass-through on imported vehicles and parts is a shared sector input; management cited tariff-related pricing pressure on the 2026-07-30 call, alongside a BEV mix drag of approximately $125 on average GPU year-to-date.
• Powersports (Harley-Davidson) is a separate discretionary cycle from auto and is small enough that its 53% revenue growth does not offset franchised margin at the consolidated level.