AXP — Lean bullish · edge 6/10

Thesis

The market has de-rated AXP as a consumer lender (tracking COF/XLF, not Visa) in response to a September Fed hike, 12.5-year-low consumer confidence, and tariff-squeezed aggregate spending. AXP's actual reported data contradicts the credit-cracking narrative: 2.0% net write-offs in both Q1 and Q2 2026, delinquencies in a 1.2-1.3% range management says has persisted for three years, and $215M of reserve releases in H1 2026. The divergence between rising EPS estimates and a falling stock is classic multiple compression. If Q3 confirms the affluent consumer remains decoupled from the aggregate, the multiple has room to recover — the 'low-20s forward P/E earlier in 2026' figure circulating in commentary is unverified from a primary source, but even a modest move from ~17x FY26 to 18-19x FY27 consensus implies meaningful upside without relying on that anchor.

Bull case

The de-rating is macro mechanical, not fundamental. AXP's affluent cardmember base is demonstrably decoupled from the aggregate consumer (+11% US consumer spend vs ~1.3% aggregate spending growth), credit is stable-to-improving with reserves being released, and FY26 EPS estimates have risen through the selloff. At ~15x FY27 EPS of ~$20, even a partial multiple re-rating to 18-19x FY27 implies $360-380, or +18-25% from $304, within the 3-6 month horizon — and that arithmetic stands on consensus estimates alone, without needing the unverified 'low-20s forward P/E earlier this year' commentary to be accurate. Q3 on Oct 23 confirming 2.0% write-offs and ≥9% billed business is the trigger.

Bear case

The market may be correctly front-running the affluent consumer's eventual crack: credit metrics are lagging indicators, and the macro backdrop (12.5-year-low confidence, tariff drag, a Fed that is HIKING) has historically preceded credit normalization with a delay. Management's refusal to raise EPS guidance while raising revenue guidance is a tell that expense growth (+12%) is structurally outpacing revenue (+10%) — Platinum engagement costs are sticky, and if billing growth decelerates to 6-7%, operating leverage goes negative fast. A second Fed hike extends the lender de-rating regardless of AXP's print. And while the stock is only ~5% above its 52-week low with RSI at 26, oversold conditions can persist for months in a hiking cycle — there is no mechanical floor here.

What would change our mind

Q3 2026 net write-off rate rises above ~2.2% or 30-day delinquencies break above ~1.4% — would validate the market's front-running of credit deterioration
Q3 billed business growth decelerates below ~7% FX-adjusted, breaking the affluent-decoupling thesis
Management cuts or narrows FY26 EPS guidance below $17.30 on Oct 23
A reserve BUILD in Q3 after two quarters of releases — the clearest forward-looking credit signal management controls

**AXP has traded like a lender, not a payments stock.** Rebased over the past year, AXP tracks COF and XLF far more than Visa, supporting the thesis that multiple compression reflects lender-style de-rating rather than worsening company fundamentals.

Since publication vs SPY · closed horizons are frozen

1 weekfinal

+2.0%

+2.5% raw · hit

1 monthrunning

+3.6%

+2.9% raw

3 monthsrunning

+3.6%

+2.9% raw

6 monthsrunning

+3.6%

+2.9% raw

The full research everything the view rests on

Direct answer to the question asked

AXP's 20% drawdown is a macro/sector de-rating of consumer-credit exposure, not a deterioration in AXP's own fundamentals: Q2 beat, revenue guidance was raised, consensus EPS has drifted UP while the stock fell 13%, and credit metrics (2.0% write-offs in both reported 2026 quarters, delinquencies below 2019, reserve releases) show zero cracking in the affluent cohort. At ~17x FY26 EPS and ~15x FY27 with an oversold technical setup and a Q3 print on Oct 23, risk/reward skews favorable over 3-6 months. The honest caveat: aggregate consumer data is genuinely deteriorating and management capped EPS upside by reinvesting, so this is a measured overweight, not a high-conviction table-pounder.

Where the edge would come from

What the price assumes

At ~17x FY26 consensus EPS of $17.64, the market is pricing meaningful risk that AXP's premium cohort eventually follows the aggregate consumer down — i.e., that current credit metrics are a lagging indicator. The post-Q2 reaction (stock -6% despite a beat and raised revenue guidance) shows investors also doubt management's reinvestment will convert to future EPS, treating the unchanged $17.30-$17.90 EPS guide as a ceiling rather than a choice.

Why the market may be wrong

Three pieces of hard data argue the selloff front-runs a deterioration that hasn't arrived: (1) AXP released reserves in both Q1 ($24M) and Q2 ($191M) — companies building toward a credit problem don't release; (2) FY26 consensus EPS rose from $17.41 (Jan) to $17.64 (Oct) while the stock fell 13%, meaning the drawdown is pure multiple compression; (3) AXP's affluent-skewed US consumer spend grew +11% in Q2 against aggregate consumer spending of ~1.3% annualized — the decoupling is the business model, not a coincidence.

The one thing to watch

The single most useful reframe: AXP is currently being priced as a subprime-adjacent lender (1y return ≈ COF's -8.7%, correlation with XLF 0.74) while its own credit book behaves like a payments franchise (delinquencies below pre-COVID, reserve releases). The variable to watch is not billed business — it's whether Q3 net write-offs stay pinned at ~2.0%. If they do, the stock is misclassified and the lender-multiple discount is misapplied; that misclassification, not spend growth, is the trade.

Key evidence (9)

Q2 2026 (Jul 24): adjusted EPS $4.53 vs ~$4.41 consensus; revenue $19.64B +10% y/y; FY26 revenue guidance raised to 10% growth, EPS guide reaffirmed $17.30-$17.90 [T1 8-K exhibit; T2 Reuters]
Credit: net write-offs 2.0% in both Q1 and Q2 2026 (Q1 vs 2.1% a year earlier; Q2 flat versus the prior-year period per the 8-K exhibit); 30-day delinquencies 1.2-1.3%, below 2019 levels; reserve releases of $24M (Q1) and $191M (Q2) [T1 SEC exhibits]
FY26 consensus EPS rose from $17.41 (Jan 30, LSEG via Reuters) to $17.64 (Oct 6) while the stock fell ~13% over 3 months [T2 Reuters, Oct 6, 2026]
Billed business +9% FX-adjusted in both Q1 and Q2 2026; US consumer services +10% Q1, +11% Q2; Platinum refresh driving retention and acquisition [T1]
Sector context: 1m returns AXP -7.8%, XLF -8.0%, COF -11.5%, V -2.4%; AXP 1y return -7.0% ≈ COF -8.7% vs Visa +7.5% — lender de-rating, not payments de-rating [computed from price history]
Macro drivers: Fed hiked mid-September 2026 [T2 Reuters]; consumer confidence at ~12.5-year low in September [T2]; Fed July MPR: consumer spending growth slowed to ~1.3% annualized, tariffs cited [T1 Federal Reserve]
Valuation: ~17x FY26 consensus EPS and ~15x FY27 (~$20); commentary earlier in 2026 had forward P/E in the low-20s [T4 commentary — no primary-source P/E series verified]
Q2 expenses +12% vs revenue +10%, driven by engagement costs and Platinum benefit usage; management explicitly chose to reinvest upside rather than raise EPS guidance [T1/T2]
Q3 2026 earnings call scheduled Friday, Oct 23, 2026, 8:30am ET [T1 IR]

Catalysts and risks

Catalysts

Q3 2026 earnings, Friday Oct 23, 2026 8:30am ET — billed business trend, credit metrics, and any EPS guidance raise [T1 IR]
Q4 2026 earnings Jan 28, 2027 — first look at FY27 guidance vs ~$20 consensus
Fed policy path — whether the September 2026 hike is followed by another in Q4
Monthly consumer data (confidence, retail spending, delinquency aggregates) that either confirms or refutes the affluent-cohort decoupling
Potential analyst re-rating cycle if Q3 confirms stable credit

Risks

Aggregate consumer deterioration finally reaching the affluent cohort — delinquencies/write-offs inflecting above 2.2-2.3%
Negative operating leverage: expenses +12% vs revenue +10% persists and compresses margins into 2027
Further Fed tightening extending the consumer-finance multiple de-rating independent of AXP's fundamentals
Platinum refresh economics disappointing — elevated acquisition/engagement costs without commensurate spend lift
Regulatory/competitive overhang (Credit Card Competition Act, DOJ case, Chase/Capital One premium competition) was not deeply verified in this research round

What to watch

Q3 net write-off rate vs the 2.0% level reported in both Q1 and Q2 2026 — the single most important number
Q3 billed business growth (≥9% FX-adj confirms thesis; <7% breaks it)
Whether FY26 EPS guidance ($17.30-$17.90) finally gets raised — a raise after Q2's reinvestment choice would be a strong signal
Expense growth vs revenue growth spread (12% vs 10% in Q2)
AXP's performance relative to COF vs Visa — if it starts trading with V instead of COF, the misclassification is correcting
Fed communications into year-end and monthly consumer confidence/spending prints
Any reserve build/release direction in the Q3 8-K exhibit

Market snapshot when the call was made 2026-10-05

Price

Return 1w	-0.8%
Return 1m	-7.8%
Return 3m	-13.0%
Return 6m	-0.3%
Return 1y	-7.0%
From 52-week high	-20.4%
Realized vol, 20d	20.1%
SPY 3m / 1y	+3.6% / +15.8%
QQQ 3m / 1y	+6.6% / +24.8%

Technicals

AXP technicals as of 2026-10-05: last close $304.02; 20d SMA 312.03 (bearish); 50d SMA 327.01; 12d EMA 307.21; 26d EMA 314.51; RSI 14 26.3 (oversold); 20d vol 19.9%; volume x0.8 vs 20d avg

Next earnings: 2027-01-28 AMC

Verifier notes REVISE · opinion revised once

Unsupported claims

what_to_watch claims 'the 2.0% [net write-off] level held for four straight quarters' — the evidence supports only two quarters at 2.0% (Q1 and Q2 2026). No Q3/Q4 2025 write-off data appears anywhere in the bundle, so 'four straight quarters' is a fabricated specific that overstates the stability track record.
bear_case claims 'the stock already trades above pre-2026 levels on a multi-year view' — the bundle contains no multi-year price history beyond the trailing 1 year, and the 1y return is -7.0% (i.e., the stock is BELOW its October 2025 level). This claim is unverifiable from the evidence and plausibly wrong.
key_evidence states net write-offs were '2.0% in both Q1 and Q2 2026 (vs 2.1% prior year)' — the '2.1% prior year' comparison is sourced only for Q1; the Q2 evidence says 2.0% was 'flat versus the prior quarter / prior-year period,' implying Q2 2025 was also ~2.0%. Minor but the parenthetical overgeneralizes the prior-year comparison.

Weakly sourced

The thesis asserts the multiple 'should recover toward the low-20s forward P/E it carried earlier in 2026' — the low-20s forward P/E is T4 commentary only, with the evidence explicitly noting no primary-source P/E series was verified. The opinion does flag this in key_evidence, which mitigates it, but the thesis and bull case state it as established fact and the re-rating argument leans on it materially.

Corrections applied

Change 'the 2.0% level held for four straight quarters' to 'the 2.0% level held in both reported quarters of 2026 (Q1 and Q2)' unless Q3/Q4 2025 write-off data is sourced.
Remove or rewrite 'the stock already trades above pre-2026 levels on a multi-year view' in the bear_case — e.g., replace with the verifiable point that the stock is only ~5% above its 52-week low and oversold conditions can persist, or source actual multi-year price data.
In key_evidence, scope the prior-year comparison: 'net write-offs 2.0% in both Q1 and Q2 2026 (Q1 vs 2.1% a year earlier; Q2 flat y/y per the 8-K exhibit)'.
In the thesis, attribute the low-20s historical forward P/E explicitly as unverified commentary (as already done in key_evidence) rather than stating it as fact, or drop the specific 'low-20s' anchor and frame the re-rating case on the 17x→18-19x arithmetic alone.

Sources (25 · T1 = primary filings)

Earlier and later calls on AXP

Oct 26, 2026Lean bullish — sentiment reset, not fundamentals.Lean bullish

Pullback looks overdone — lean bullish.

Thesis

Bull case

Bear case

What would change our mind

Since publication vs SPY · closed horizons are frozen

The full research everything the view rests on

What the price assumes

Why the market may be wrong

The one thing to watch

Catalysts

Risks

Price

Technicals

Earlier and later calls on AXP

Want a fresh look at AXP, or another name?