Citi Clears Its Own Bar and Falls 5.6% Anyway: At 1.22x Book, We Upgrade

Rating Action: Upgrading to Outperform

We have spent two quarters on the sidelines of a franchise we like, and the reason was price. In April we stepped aside at 1.31x tangible book because a ~13% seasonal peak return was being capitalized as if it were durable. In July we stayed there at 1.32x because the Investor Day had framed a staircase rather than a reset, and because management refused to commit to anything above 11% for the year. The April note asked for a meaningful pullback or an Investor Day that framed returns well above 11%. The July note made the test concrete: come back on a genuine de-rating toward 1.15-1.20x book, or on a second half that shows the 2026 “waypoint” was conservatism rather than a warning.

Q3 delivered the second condition and most of the first. Markets mean-reverted exactly as we expected, from 17.0% to 12.3%, which is inside its own 11.5-13% target. The firm still earned 11.5%. That is the number we had estimated in July as the “normalized” return underneath the Q2 headline, and it arrived with about $250M of pulled-forward severance and investment in the cost base. Management then did what it declined to do in July and raised the full-year guide to above 11%. Year to date the firm has earned 12.6%. Taken together, the durable return sits in the middle of the 11-12% range we use for valuation, and the evidence for it is stronger than it was in July.

The price moved the other way. The stock fell 5.6% on the day to $124.53 and closed the week at $124.26, which is 1.22x the $101.59 tangible book. That is about 8% below the bottom of our fair value range, and it is a lower multiple than at our April downgrade. The Hold has done its job: since we downgraded on April 15, C has fallen 5.6% against a 9.1% gain in the S&P 500 (price only). We would rather be early on the way back than chase the stock once the second-half picture is obvious to everyone.

We are not upgrading on a clean bill of health. Cards missed management's own revenue guide by about $100M and its earnings fell 8%. The buyback has stepped down in each quarter since Q1, and the CEO answered the buyback question by talking about investment. Under current rules the regulatory minimum rises 50 bps by January 2028, which means capital has to be built. Those are the reasons the stock trades at 1.22x rather than 1.35x, and we think they are now priced. At 1.33-1.45x tangible book on a durable 11-12% return, fair value is $135-147. The midpoint is about 13% above the reaction close, before a ~2% dividend. That is better than a market return, which is what Outperform means in our framework.

What would make us wrong: a Q4 that misses the high-single-digit seasonal floor management described, a further step down in buybacks below $3B a quarter without a regulatory reason, or cards contra-revenue that keeps outrunning its guide. We set these out under Commitments to Watch below.

Results vs. Consensus

A smaller beat than Q1 or Q2, and a cleaner one. EPS beat by 7% while revenue beat by less than 1%, so most of the beat came below the revenue line. Inside revenue, net interest income grew 14% and non-interest revenue fell 4%. Much of that split is one business: Markets earned more of its revenue as financing NII (Markets NII +50%, Markets NIR -20%). Cards contra-revenue also grew.

Q3 2026 Scorecard

Year-Over-Year Comparisons

The prior-year quarter carried a $726M Banamex goodwill impairment, so the reported growth rates flatter. On a like-for-like basis revenue grew 8% against expense growth of 3%, about five points of positive operating leverage. Strip out our estimate of the ~$250M pulled forward from 2027 and expense growth was closer to 1%. The tax rate fell to about 26% from 29%, and the diluted share count fell 8%, so EPS growth outran net income growth by a wide margin.

Quarter-Over-Quarter Comparisons

Two lines more than account for the sequential revenue decline: Markets fell $894M and Investment Banking $198M, together $1,092M against a total decline of $921M. Services, Wealth and Cards all grew sequentially, while All Other fell $305M. That pattern is the one we predicted in July, when we argued that the reported second half would be worse while the underlying business would not.

Revenue assessment. In Q3, unlike Q2, the composition of revenue growth looks like the franchise Citi says it is building rather than the capital markets cycle. Services and Wealth together grew $1.31B year over year; Markets and Banking together added $462M. The best third quarter in a decade on a normal trading quarter is a better data point than a decade-best first half on a peak one. The weakness is concentrated and identifiable: cards NIR, and a Corporate/Other line that keeps absorbing asset-liability management costs.

Margin assessment. Expenses fell sequentially while management pulled 2027 costs into 2026. Firmwide expenses ex-notable rose 3%, and on our estimate about two points of that is the pull-forward. The cost roll-off story we gave management credit for in July is now visible in the All Other line, where expenses fell 14%. The 58.5% efficiency ratio is higher than Q2's 57.4% only because the revenue denominator lost a peak trading quarter.

EPS assessment. $2.85 against $2.66 is a 7% beat with no help from the trading desk and a modest help from tax. The sequential decline from $3.15 is seasonal and was guided. Year-to-date EPS is $9.07, up 47% on an ex-notable basis. That is the self-help case delivered for a third straight quarter, at a lower multiple than in April.

The Normalization Test, One Quarter Later

In July the whole debate came down to one question: what does Citi earn when trading is normal? We answered it with management's own Investor Day targets. Marking Markets from 17.0% to the midpoint of its 11.5-13% target took Q2's 13.0% to about 11.4%; also marking Services to its mid-20s through-cycle target took it to about 10.2%, which was the full-year guide. We concluded that the 13% was a cycle on top of a genuine 11%. Q3 is the test of that analysis, because Markets did the normalizing for us.

The second-half arithmetic we laid out in July holds up as well. The literal July guide implied a second-half RoTCE of roughly 8-9%; we said we would model it nearer 10-11%. Q3 came in at 11.5%, and management described Q4 as landing at a “high single digits rate” once Markets seasonality is applied. On our Q4 estimate of about 9.5%, the second half averages roughly 10.5% and the full year about 11.7%, above the new “above 11%” guide. Markets is tracking the “20% or worse” first-half-to-second-half decline: $14,253M in the first half against about $11.0B in the second if Q4 falls 20% from Q3, a decline of roughly 23%.

Assessment. The quarter validates the July analysis rather than the July selloff. The market sold Q2 on the belief that management was guiding to a much worse business, and we argued the guide was mostly Markets arithmetic. Q3 shows the business earning what we said it would once the arithmetic played out. That does not make the stock cheap at any price. It does mean the 1.22x multiple is being applied to a return we now have direct evidence for, rather than to a seasonally inflated one.

Segment Performance

Four of five businesses improved their return year over year, and the fifth (Markets) is exactly where its own target says it should be. The table compares each business against the targets set in May.

The firm is printing inside its 2027-2028 band in 2026 with its trading business at a normal return. In July the firm sat at the top of its band only because Markets was over-earning. In Q3 the over-earning is concentrated in Services, the business with the most structural reasons to stay above its through-cycle target in a higher-rate world. The remaining unbanked upside is in Wealth and in the All Other drag.

Services: 31.2% and Still Compounding

Revenue rose 15% to a record $6.61B, with Treasury and Trade Solutions up 16% to $4.90B and Securities Services up 13% to $1.71B. NII grew 17% on higher deposit balances and spreads, and fee revenue grew 12%. Average deposits rose 16% to $1,037B, cross-border transaction value rose 10%, commercial card spend 10%, and assets under custody and administration 16% to $35T. Net income rose 27% to $2.64B for a 31.2% return. Expenses rose 10%, partly severance.

“This was another standout quarter in services. With revenues up 15%. Growth was broad based across every driver, from deposits to commercial card spend.” — Jane Fraser, CEO

Management fielded a direct challenge on why a business earning 30% should be modelled at mid-20s. The answer was that the target is through-the-cycle and that the business has grown revenue at about an 8% compound rate and deposits at about 6% across the 2016-2025 period, which spanned zero, medium and high rates. Deposit growth is normalizing slowly, from 19% in Q2 to 16% now, and management repeated that it does not expect to grow at a multiple of money supply indefinitely.

Assessment: Services is now the main source of the firm's excess return, and we think more of it is durable than the mid-20s target concedes. Higher policy rates outside the US, which management flagged, favour a non-USD-asset-sensitive deposit franchise, and the operating-deposit mix grew faster than total deposits. The risk is the one management named: deposit growth normalizing toward low single digits while technology investment continues at about $2B a year. We keep Services marked at mid-20s for valuation purposes. That conservatism is part of why we think 11-12% durable is a floor-ish estimate rather than a stretch.

Markets: Back at Target, Exactly as Planned

Revenue rose 6% year over year to $6.11B but fell 13% from Q2. Fixed income rose 4% to $4.39B: spread products and other fixed income rose 13% on financing revenue, while rates and currencies were flat as FX volume growth offset weaker rates. Equities rose 14% to $1.73B on prime services, with prime balances up more than 40%, partly offset by lower derivatives. Average loans rose 30% to $191B on spread-product financing. Net income rose 2% to $1.74B for a 12.3% return.

“In markets, we expect a typical seasonal decline in the fourth quarter around the same magnitude as last year, which was 20% or possibly a bit more given the strong performance so far this year.” — Gonzalo Luchetti, CFO

Assessment: This is the most important segment result in the note, and the reason is that it is boring. Markets earned 12.3%, inside its own target band, on a mix tilted toward financing revenue and away from trading volatility. Average loans in Markets are up 30% and VaR only 11%, which is the “more stable financing revenue” strategy working. Equities revenue growth of 14% against prime balance growth above 40% suggests either pricing for share or a lag in monetization, and management attributed it to being late to invest. Q4 will likely drop toward $4.9B on the seasonal guide. That is priced into the full-year guide and should not surprise anyone.

Banking: ECM Carries a Softer Fee Quarter

Banking revenue including loan hedges rose 6% to $1.74B. Investment Banking revenue rose 9% to $1.35B, driven by ECM up 77% to $343M, while advisory fell 4% to $406M and DCM fell 3% to $601M. Sequentially IB fell 13%. Corporate lending fell 16% to $382M on lower spreads. Credit costs dropped to $23M from $157M a year ago, which lifted net income 40% to $378M and the return to 19.2%.

“I would say we expect to see an above average wallet even if it pulls a bit back from the 2026 near-record levels for the rest of the year.” — Jane Fraser, CEO

Assessment: A good quarter in a softer wallet, and the composition is worth noting. ECM strength came from follow-ons and converts, with lead roles on the SK Hynix and Intel offerings, while DCM and advisory, the more relationship-driven lines, both slipped. Much of the 40% earnings growth came from a lower provision rather than from fees. Banking contributes $378M of quarterly net income on $8B of capital. It is a well-run business improving its return, and it remains too small to swing the firm.

Wealth: Strong Flows, Return Stuck Just Below 15%

Revenue rose 15% to $3.27B across all three businesses: Citigold and Retail Banking up 16%, Private Bank up 17%, and Wealth at Work up 7%. NII grew 20% on deposit spreads; NIR grew only 5%. Net new investment assets were $38B in the quarter, more than double a year ago, and more than $75B over twelve months, or 11% organic growth. Expenses rose 3%. Net income nearly doubled to $585M, but the return was 14.3%, flat with Q2.

“So that gives us confidence that soon we will be able to get into the 15-20% near term target window that we spoke about at Investor Day.” — Gonzalo Luchetti, CFO

Assessment: The flows are excellent and the return has stalled for a quarter. Wealth went from 7.8% a year ago to 10.8% in Q1 to about 14% in Q2 and Q3. Management attributed the gap between 15% asset growth and 5% fee growth to strategic decisions taken a year ago that are still being lapped. We accept that explanation for now, because the flow numbers are not a business losing momentum. But this remains the business that decides whether 11-13% arrives in 2027 rather than 2028, and two quarters at 14% is a pause we will grade again in January.

U.S. Consumer Cards: The Guide Missed, and the Return Flatters

Revenue was flat at $4.66B. NII rose 6% on higher interest-earning balances, but non-interest revenue fell to ($758M) from ($468M) a year ago on higher partner payment accruals and rewards costs. Expenses rose 2%. Net credit losses fell 4% to $1.67B, but a $207M reserve build, largely macro-driven, lifted the provision to $1.88B. Net income fell 8% to $851M. General purpose spend rose 13% and loans 9%; private label continued to shrink.

“As well as remember, we have, in place some profit sharing agreements with some of our partners. So the fact that losses are performing quite well that some of that gets shared. Still accretive for us because we do not share 100%, but that is what has a little bit of an impact on the revenue.” — Gonzalo Luchetti, CFO

Assessment: This is the weakest part of the quarter and it confirms the cards bear point we added in July. Some of the NIR overrun is benign: partner profit-sharing rises when losses run better than expected, and losses did run better. But a specific guide missed by 15% within one quarter, with no re-guide offered, is a credibility cost. New account acquisitions fell 4% year over year despite the higher marketing spend. Management's line is that the business will hold “around” its low-20s through-cycle return while expenses outpace revenue in some quarters. That is acceptable at 1.22x book. It would not have been at 1.32x.

All Other: The Drag Is Partly an Accounting Artifact of the Banamex Sale

All Other lost ($955M) on a managed basis against ($923M) in Q2 and ($756M) a year ago. Revenue fell 3% to $1.43B: Mexico consumer and SBMM revenue rose 18%, but Corporate/Other was ($415M) and Legacy Holdings Assets swung to ($203M). Expenses fell 14% on lower transformation and exit costs. The provision rose 61% to $532M on Mexico consumer losses, where the loss rate is 6.68%, up from 5.46% a year ago.

The line we flagged in July needs a correction to our own framing. Net income attributable to noncontrolling interests in All Other was $146M this quarter against $3M a year ago. That is the minority shareholders' share of Banamex earnings after Citi sold 49% of it. It accounts for $143M of the $199M year-over-year widening in the net loss. Before minority interest, the All Other loss was ($809M) against ($752M) a year ago and ($761M) in Q2. The year-over-year widening that remains is a tax effect. Before tax, the All Other loss narrowed from ($1,029M) to ($974M): expenses fell $295M, which more than covered a $201M rise in provisions, mostly Mexico consumer credit, and a $39M dip in revenue. The tax benefit shrank from $277M to $165M.

Assessment: Better than the headline, and better than we feared in July. About 70% of the year-over-year widening is the arithmetic of having sold a stake, and that capital came in the door when the stake was sold. The rest is a smaller tax benefit. Before tax the line improved by $55M, as cost roll-off outran Mexico credit. It did slip sequentially: the pre-tax loss was $53M wider, as a $94M rise in provisions outweighed expenses falling $41M more than revenue. Deconsolidation in early 2027 removes the Mexico credit line and the minority interest line together. The cost side is now doing its job. The revenue side, Corporate/Other at ($415M) and Legacy Holdings Assets at ($203M), is the piece of the bridge from 11% to 13% that is not yet moving.

Key Topics & Management Commentary

Overall Management Tone: Calmer and more forthcoming than July. Having declined to size anything on the July call, management this time placed the pull-forward inside the $5B envelope, split it by quarter and named the severance component, raised four separate guides, and recommitted to the 11-13% framework without hedging. The firmness that unsettled the market in July has moved from the expense line to the capital line, where the CEO answered a buyback question by defending investment. The one area handled thinly was cards, where a missed guide went unaddressed.

1. The Waypoint Was Conservatism: Full-Year Guide Raised Above 11%

The single most important change from July. The 10-11% target that management called a “waypoint” and refused to exceed is now a floor. The CFO noted this is the second raise in about a month, after guidance of “a bit better than 11%” given about 30 days earlier.

“We expect to exceed this year's return target, even as we pull forward investments toward our 11-13% near-term goal” — Jane Fraser, CEO

Year-to-date RoTCE is 12.6%, and the implied Q4 is a high-single-digit return after the seasonal Markets decline. Our estimate puts the full year at about 11.7%.

Assessment: This is the upgrade trigger we set in July, and it was met in the way that matters. We wanted evidence that the waypoint was a conservative number rather than a signal of a worse business. Management has now raised it twice and delivered 11.5% in a quarter that carried pulled-forward costs and a normal trading result. The credibility gap that cost the stock 5.3% in July has narrowed. We would not call it closed until Q4 lands inside the high-single-digit range described.

2. Markets Normalizes, and the Firm Still Earns 11.5%

Markets fell from 17.0% to 12.3% and from $7.0B to $6.1B of revenue, a 13% sequential decline, and the firm return fell 150 bps. That is the mean reversion the bears said was coming and the guide anticipated.

“I think you are reading too much into it. This is just the normal seasonality that we have seen over many, many years between a strong third quarter and the typical fourth quarter, it is no more than that.” — Gonzalo Luchetti, CFO

Assessment: The bear case on Citi in July was that the return would fall when trading normalized. It did fall, to 11.5%, which is the level we argued was the real one. The Markets mix is also improving: spread-product financing and prime balances are more stable revenue than flow trading, and Markets loans are up 30%. The stock is now being priced as though the 11.5% were also cyclical. We do not think it is.

3. The Pull-Forward Gets a Shape

In July management declined four times to size the second-half investment ramp. The ~$500M figure surfaced in the weeks before this call. This time the CFO placed it fully inside the $5B three-year Investor Day envelope and said roughly half landed in Q3, including about $200M of severance.

“You can think of it as about half of it landed in this quarter. You can see about $200 million of severance in the quarter. So that is a good portion of it.” — Gonzalo Luchetti, CFO

Year-to-date severance is now above $1B, against about $800M for all of 2025. Exits and wind-downs are at a ~$100M quarterly run-rate, and the CEO told a questioner to “feel good about how we exit this year.”

Assessment: This is the disclosure the market wanted in July, and it is smaller than the July selloff assumed. A $250M quarterly pull-forward is about 2% of the cost base and about 45 bps of annualized RoTCE. It is also funded from inside the existing $5B envelope rather than added to it. The severance figure tells us management is continuing to take structural cost out while it invests. That is the version of “spend, not bank” we said we could live with.

4. Capital: The Buyback Cadence Steps Down Again

Citi repurchased $3.0B in Q3, against $4.0B in Q2 and $6.3B in Q1. Year to date that is $13.3B, already above the $13B for all of 2025, and the payout ratio was 85% against 92% in Q2. CET1 held at 12.8%, about 120 bps above the 11.6% minimum, after an AOCI hit from higher rates. Under current rules the minimum rises to 12.1% in January 2028, and the directional target is 13.1%.

“We know buybacks are important. We know return of capital is important to our shareholders. And we make that judgment as to what the right balance is. But you think all of our investors will want us to make sure that we are investing behind these long term, high return growth opportunities that will drive the performance and competitiveness of the bank.” — Jane Fraser, CEO

The CFO added that the glide path assumes current rules only, and that pending changes to Basel III, the G-SIB surcharge and the stress capital buffer are not in it. He expects full visibility on the new framework around the middle of next year.

Assessment: This is the most important new negative in the quarter. Two consecutive step-downs in the buyback, combined with a 50 bp rise in the minimum under current rules, imply a slower capital return in 2027 than the $30B program headline suggests. A $3B quarterly pace is still about 6% of the market capitalization a year, and at 1.22x book each dollar buys more tangible book per share than it did at 1.32x. The upside case is regulatory. In our view the pending rule changes are more likely to reduce Citi's requirement than to raise it, and none is in the 13.1%. We treat that as a free option, not a base case.

5. Cards: A Missed Guide and a Profit-Sharing Explanation

Cards NIR was ($758M) against a July guide of holding ($659M). The CFO attributed it to rewards costs and to partner profit-sharing that rises when losses run better than expected. General purpose cards keep growing: spend up 13%, loans up 9%.

“As I mentioned last quarter, while we expect ROTC to remain around our through the cycle target for the business, in certain quarters we do expect expense growth to outpace revenue growth as we invest in the business to drive engagement and acquisitions with some of these investments reflected as contra revenues and others as expenses.” — Gonzalo Luchetti, CFO

Assessment: The explanation is plausible and the miss is still a miss. Management did not re-guide the fourth quarter, so the July commitment of ($659M) for Q4 is technically still standing against a Q3 run-rate $99M worse. Net income fell 8% and, on constant capital, the return fell year over year. The strategic direction (general purpose growth, private label converting toward co-brand) is right, and the portfolio is prime: 86% of balances are at FICO 660 or above. But this is the one business where Citi is paying to catch up, and the bill is arriving faster than guided.

6. Services: “A True Moat” Under Higher Rates

The CEO used the Services question to restate the franchise case: four growth pillars, a roughly 40% improvement in win rate, minimal attrition, and about $2B of annual platform investment. Higher rates outside the US add to it, because Citi's asset sensitivity sits mostly in non-USD balances.

“And that is giving us a model that is pretty almost impossible to replicate. It is a true moat.” — Jane Fraser, CEO

Assessment: We agree with the substance. Services is the asset that justifies a premium to book for this bank, and a 31.2% return with 16% deposit growth and record revenue is the franchise doing what the Investor Day said it would. Our caution is valuation discipline, not doubt about the business: we keep the mid-20s target in our normalization, which leaves Services upside outside our fair value.

7. Rates: Net Positive, Mostly Outside the US

With short-end rates rising in the US and abroad, the CFO said a higher-for-longer environment is a net positive. NII ex-markets guidance moved to mid-to-high 6% from 5-6% at the start of the year. The firm is roughly neutral to US dollar rates within twelve months and turns asset-sensitive in year two as the short-duration portfolio reinvests.

“And then you will see also we are relatively neutral on a USD basis. At least within 12 months because given the short duration of our portfolio, we actually do turn asset sensitive in year 2.” — Gonzalo Luchetti, CFO

Assessment: A quiet positive that the reaction ignored. The same rates that hurt AOCI and capital this quarter support Services and Wealth NII into 2027. The near-term cost shows up in CET1; the benefit shows up over several quarters in revenue. For a bank whose problem has been a low return rather than a weak balance sheet, that is the better side of the trade.

8. The DTA: The Second Dot

In July the CFO said “you need 2 dots to draw a line” on the disallowed deferred tax asset, after $500M was consumed in the first half. Q3 is the second dot. The deduction fell by about $360M in the quarter, to roughly $900M year to date, ahead of the $800M full-year target.

“And we have reduced the total firm wide DTAs deducted from CET1 capital by approximately $900 million year to date.” — Gonzalo Luchetti, CFO

Assessment: Two dots make a line, and the line is ahead of plan. The DTA burn is the cleanest way for Citi to lift its return without new revenue, because it converts disallowed capital into usable capital as US earnings come through. The remaining deduction is about $13.0B against a path to about $7B over two years. We still keep it out of the base case, but it now has the track record to sit in the upside case with some weight.

9. Banamex: A $9 Billion Headline Loss That Is Not a Loss of Capital

Citi closed the sale of a further 1.4% stake, taking the total divested to 49%. Deconsolidation is set for 2027, with an IPO after that. The CFO repeated the accounting consequence and added that the net effect on capital is positive.

“Second, as we have shared previously, as part of the deconsolidation process, we expect to recognize an approximate $9 billion CTA related loss through the P&L in early 2027. And third, we expect the impacts related to the deconsolidation process to be slightly accretive to our CET1 capital ratio.” — Gonzalo Luchetti, CFO

Assessment: Investors should prepare now for a Q1 2027 print that shows a large GAAP loss, roughly $5 a share on the current count. The CTA is already in accumulated other comprehensive income, so moving it through the P&L does not change equity, tangible book or capital, and management expects the overall deconsolidation to add slightly to CET1. It will look alarming and it will not matter. It also removes the Mexico credit and minority-interest lines from All Other.

10. AI: Numbers Instead of Adjectives

Management gave a far more specific AI disclosure than in July: 14,000 engineers on its main agentic coding tool, technology vendor spend down 30% year over year from insourcing, product launch cycles 30-40% faster, about 100 of the most manual processes being rebuilt end to end, and client service response times 21% faster in two pilot countries. The CEO said the financial benefits this year significantly outweigh the investment.

“We have reduced our tech vendor spend by 30% year over year as a consequence, and that is still got a long way to run.” — Jane Fraser, CEO

Assessment: In July we called the AI story directionally credible and analytically weightless. The vendor-spend figure is the first item that maps to a specific line in the cost base, and the insourcing mechanism is plausible. We still do not have a dollar figure for the benefit, and we would want one before crediting it in the efficiency outlook. The cyber discussion, which framed defense as part of the AI agenda, was a sensible response to a risk the market is starting to price across banks.

11. Credit and the Macro: Stable Losses, a Maturity Wall Ahead

The CEO said conditions softened a bit during the quarter as energy prices rose and monetary policy tightened, but that the US consumer and labor market are holding up. Firmwide losses were flat, card loss guidance moved to the low end at about 4%, and the reserve build was driven mainly by macro assumptions. On the corporate side she flagged an upcoming three-year maturity wall and a shift from syndicated debt toward loans and revolvers.

“Our reserves incorporate an 8 quarter weighted average unemployment rate of 5.3%, which includes a downside scenario average unemployment rate of nearly 7%,” — Gonzalo Luchetti, CFO

Assessment: Credit is not the risk in this name right now, and the reserves are positioned conservatively. The two pockets to watch are Mexico consumer, where the loss rate is up 122 bps in a year, and the corporate refinancing wall in a higher-rate world. Citi's corporate book is 79% investment grade and its card book is prime-heavy, which is why we keep the macro bear point contained.

Guidance & Outlook

Four guides went up and none went down, which is the reverse of the July call's message. The only unresolved item is cards NIR, where the Q3 print broke the July guide and management offered no new number for Q4.

Implied fourth quarter: Management's own description is a high-single-digit RoTCE after a ~20% Markets decline. We model Q4 revenue near $22.6B, expenses near $14.0B including the rest of the pull-forward, and EPS of roughly $2.30-2.45. That puts FY2026 EPS at about $11.35-11.55 and FY RoTCE near 11.7%.

Street at: The pre-print EPS consensus had been revised down in the week before the report, so the Street came in leaning cautious. The consensus price target sits around $150, well above the current price.

Guidance posture: Conservative, and now visibly so. Management guided 10-11% in January, held it through two 13% quarters, called it a waypoint in July, then raised it twice inside a month. That pattern looks like a team that would rather raise than explain, and it is easier to own a stock under that kind of management at 1.22x book than at 1.32x.

Analyst Q&A Highlights

The Fourth-Quarter Return and the Path Into 2027

The opening question went straight at the seasonal gap: a 12.6% year to date and 11.5% quarter imply a weak Q4. Management answered with Markets seasonality, the guide raise and a flat refusal to move the Investor Day framework, and added that the targets had been stress-tested for a range of environments.

Q: “Understanding there is lots of factors around seasonality and how much spending you pull forward, could you speak a little bit to the near term expectations and trajectory for ROTCE for the fourth quarter? And then as you look out for the next 2 years?”
— John McDonald, Truist

A: “We have not moved an inch from our Investor Day commitments. The near term targets for returns Just as a reminder for everyone, are 11% to 13% into the 2027 and 2028 time period. So we are comfortable with that.”
— Gonzalo Luchetti, CFO

Assessment: A cleaner answer than in July. Management now owns both the near-term number and the 2027-2028 band, and it volunteered the Q4 arithmetic rather than leaving the Street to back into it. The 11-13% recommitment matters more than the Q4 detail, because that band is the basis of our valuation.

Whether the Implied Fourth Quarter Signals Something Worse

A blunt follow-up: the Q4 return implied by the guide looks poor, so is management expecting a shock, or just being cautious? Management chose caution and pointed out that the guide had already been raised twice.

Q: “But the return for the fourth quarter if you back into it, looks really lousy to get to 11%.”
— Mike Mayo, Wells Fargo

A: “So we are guiding to above 11% which is again an improvement from the a bit over 11% that we talked about 30 days ago. And of course, there is a range. We could do better but it is hard to predict early in the quarter… And, yes, abundance of caution.”
— Gonzalo Luchetti, CFO

Assessment: “Abundance of caution” is the most useful phrase on the call, because it is management conceding that the guide is a floor. This is the same pattern as July (guide low, beat, raise), and it is why the full year will likely land well above 11%.

Whether the Buyback Pace Can Hold

With CET1 flat, AOCI under pressure and the minimum rising, the question was whether the current repurchase pace is sustainable while the bank also deploys capital into growth. Management answered with priorities rather than a number, and investment came first.

Q: “And so I guess, just wanted to confirm that you can continue this pace of buyback at around these levels, even with higher rates.”
— Erika Najarian, UBS

A: “So as Gonzalo said, once we are comfortable around the resiliency of our capital base, which we feel very good about, we see a lot of opportunities to deploy capital into high returning opportunities… We know buybacks are important.”
— Jane Fraser, CEO

Assessment: Not a commitment, and that is the point. The answer extends the July message (opportunities get funded first) from expenses to capital. A later questioner summarized the takeaway as possible tempering in the pace of buybacks, and management did not push back. We now model about $3B a quarter, not the $4-6B of the first half.

Whether the Markets Seasonality Is Conservatism

A peer had said it expected little Q4 seasonality given activity levels, so the question was whether Citi's 20% decline guide reflects caution or a mix issue in an equities business still being built. The CFO dismissed both readings in three sentences.

Q: “Is that conservatism, in terms of the typical seasonality, or is that more of a comment on mix in terms of you guys are still building out the equities business?”
— Erika Najarian, UBS

A: “I think you are reading too much into it. This is just the normal seasonality that we have seen over many, many years between a strong third quarter and the typical fourth quarter, it is no more than that. Do not read too much into it.”
— Gonzalo Luchetti, CFO

Assessment: The brevity is credible. Last year's Q3-to-Q4 decline was almost exactly 20% ($5,745M to $4,609M), so the guide is a historical base rate. If a peer's view of a stronger Q4 wallet proves right, Citi has room to beat it.

What the Pull-Forward Is and How Much Landed

The July call left investors unable to size the second-half investment. This question asked for the amount spent in Q3, the purpose, and whether the $5B Investor Day envelope was growing. It got specific answers on all three.

Q: “How much of that has been spent in the third quarter? Why did you pull it forward? What are you spending it on?”
— Mike Mayo, Wells Fargo

A: “You can think of it as about half of it landed in this quarter. You can see about $200 million of severance in the quarter. So that is a good portion of it. And then there is a few investments extra around technology, around marketing, around client experience that we also we also support.”
— Gonzalo Luchetti, CFO

Assessment: The disclosure the July call lacked, and it confirms our July estimate that the ramp was modest. About $250M a quarter, inside the existing $5B, mostly severance that takes cost out later. The market sold Q2 on an unsized spending story; the sized version is small.

Why a 30% Services Business Should Earn Mid-20s

A sharp question on the Services target: with the business near 30%, what brings it back to 25% if rates stay put? The CEO did not offer a mechanism for decline. She described the target as through-the-cycle and made the competitive case instead.

Q: “Assuming rates do not do anything, like, why should that go from 30% to, let's say, 25%?”
— Ebrahim Poonawala, Bank of America

A: “The target we have is a through the cycle target… And that is giving us a model that is pretty almost impossible to replicate. It is a true moat.”
— Jane Fraser, CEO

Assessment: The non-answer is informative. Nothing in the business is expected to push the return down. The mid-20s figure is a cycle-average assumption, and in a higher-rate world it looks conservative. That supports our choice to treat Services over-earning as upside to fair value rather than a risk to it.

The Capital Glide Path Under Rules Not Yet Written

Two questioners pressed on whether the 13.1% capital target could be revisited once Basel III, G-SIB and stress-test changes are finalized. Management confirmed the glide path reflects current rules only and that the outcome will not be clear for months.

Q: “Once we get finalization and, maybe as you go through the stress test and you see the results of that, revisiting that 13.1% is still on the table?”
— Manan Gosalia, Morgan Stanley

A: “To the first part of your question, yes. The 1 which anchors to a minimum requirement of 12.1%, into January 2028, anchors to current rules. So it does not include anything related to any of the rules that are in play that some of which have been, released and others have not.”
— Gonzalo Luchetti, CFO

Assessment: The honest version of the capital story. Under today's rules Citi must build capital into 2028, which constrains buybacks. The pending rule changes look more likely to bring relief than to add to the requirement, none is assumed, and the CFO expects clarity around mid-2027. This is the clearest asymmetric option in the name.

How Far Along the Cards Contra-Revenue Drag Is

The only line of questioning on cards economics asked how much further the contra-revenue drag from the general purpose push has to run. The answer explained the mechanics and pointed to driver growth, but it did not address the gap to the July guide or give a duration.

Q: “How far along are you in kind of the drag of those initiatives as you think about the revenues.”
— Matt O'Connor, Deutsche Bank

A: “So the fact that losses are performing quite well that some of that gets shared. Still accretive for us because we do not share 100%, but that is what has a little bit of an impact on the revenue.”
— Gonzalo Luchetti, CFO

Assessment: The question asked how far along the drag is, and the answer did not say. Profit-sharing explains part of the overrun and is a by-product of good credit. But the July guide was specific, the miss was $99M, and no one on either side mentioned it. We treat cards NIR as an open item for Q4.

What They're NOT Saying

1. Why cards NIR missed its own guide. Management guided in July that cards non-interest revenue would hold at Q2's ($659M) in Q3 and Q4. It printed ($758M). The prepared remarks did not mention the guide, no analyst asked, and no Q4 number replaced it.
2. A buyback number for 2027. Asked directly whether the current pace can continue, the CEO answered with capital priorities. With the minimum rising 50 bps under current rules and buybacks down in each of the last two quarters, the absence of even a directional figure is the main thing the stock is discounting.
3. The consent orders. In July management said a large body of consent-order work had passed internal audit validation. This quarter the orders were not mentioned once on the call. That probably means nothing has changed. It also means the last undated structural catalyst received no update.
4. The 2027 expense jump-off in numbers. Asked how the run-rate exits 2026, the CEO said investors should “feel good about how we exit this year” and cited stranded and transformation cost roll-off. With more than $1B of severance this year and a $5B investment envelope still ahead, a dollar figure would have settled a debate that a reassurance cannot.
5. Why the All Other tax benefit shrank. Before tax, the All Other loss narrowed year over year. After tax, it widened because the tax benefit fell from $277M to $165M. Neither the tax line nor the $201M rise in mostly Mexico-driven provisions came up on the call.
6. A dollar figure for the AI benefit. “Significantly” outweighing the investment is a ratio without a base. The 30% cut in vendor spend is the closest the call came to a number we can put in a model.

Market Reaction

• Pre-print setup: C closed at $131.92 on October 12, up 13.1% year to date against the S&P 500's 14.0% and up 37.3% over twelve months. It was down 5.0% over the trailing 30 days and 9.4% below its $145.67 52-week closing high, so it entered the print already lagging.
• Reaction-day move: The shares opened down 0.4% at $131.35, traded as high as $133.50 (+1.2%), then sold off steadily to close at $124.53, down 5.6% ($7.39), just above the session low of $124.44. The S&P 500 was flat on the day.
• Volume: 20.4M shares against a 9.4M 30-day average, or 2.2x normal.
• Peers: Among banks reporting the same morning, JPMorgan fell 0.9%, Wells Fargo 3.1% and Goldman Sachs 0.2%. Bank of America fell 4.8% the following session.
• Follow-through: C fell another 1.8% to $122.27 on October 14, 1.20x tangible book, then recovered to $124.26 by October 16 while the S&P 500 fell 1.9% over the same three sessions.

The reaction rhymes with July, but the cause is different. In July the stock sold off during the guidance discussion because management would not commit above 11%. This time management did commit above 11%, and the stock still fell. The selling pressure appears to have come from the shape of the quarter rather than the outlook: revenue and earnings down sequentially, Markets down 13%, Investment Banking down 13% sequentially, a Q4 guided lower again, and a buyback that stepped down. For a stock that had rallied hard into June, that read as a peak.

Our read is that the market is now pricing Citi as though Q3 were the cyclical top of a business that will drift back toward 10%. The data says the opposite: the cyclical layer has already come off and the base is 11.5%. Two days after the print, the stock was holding its level while the broader market fell, which suggests the forced selling was done. A 2.2x-volume flush to 1.22x book, on a quarter that raised the guide, is the kind of dislocation we said in July we would want to act on.

Street Perspective

Debate: Was Q3 the peak quarter for this cycle?

Bull view: No. The cyclical part of the return has already reverted: Markets is at its own target and the firm still earned 11.5%. Services and Wealth, the durable engines, both grew 15%, and management raised four guides. A peak quarter does not come with an upward revision to the full year.

Bear view: Yes. Revenue, net income and return all fell sequentially, Investment Banking fell 13% sequentially, Q4 is guided lower again, and the stock's second straight selloff on a beat shows the market sees earnings momentum rolling over after a strong spring rally.

Our take: The bulls have the better argument, because the bears are describing seasonality. A 13% sequential decline in Markets that leaves it exactly at target is the normal pattern, not a top. The quarter that matters for the “peak” question is the full-year 2027 return against the 11-13% band, and nothing in Q3 argues for the bottom of it.

Debate: Does the capital priority break the buyback support?

Bull view: No. $13.3B repurchased in nine months is already more than all of 2025, the payout ratio is 85%, and pending rule changes are more likely to cut Citi's requirement than to raise it. At 1.22x book, buybacks are also more accretive per dollar than they were in the spring.

Bear view: Yes. Quarterly buybacks have fallen from $6.3B to $3.0B, the minimum rises 50 bps by 2028 under current rules, AOCI is moving against the bank as rates rise, and the CEO answered a direct buyback question by talking about investment. The capital return that supported the multiple is shrinking.

Our take: The bears have the direction right and the magnitude wrong. A $3B quarterly pace is still about 6% of the market capitalization a year. The step-down is the main reason the stock trades at 1.22x rather than 1.35x, and we think that discount is adequate. The regulatory option is free at this price.

Debate: Is 1.22x tangible book cheap or a fair price for an 11% bank?

Bull view: Cheap. The durable return is now evidenced at 11-12%, tangible book compounds at 6%, management recommitted to 11-13% for 2027-2028. A bank earning above its cost of equity, with a credible path to 13%, deserves more than 1.22x book.

Bear view: Fair. An 11% return against a 10-11% cost of equity justifies little more than 1.1-1.2x book. Cards is losing ground, All Other is still widening, and management will keep spending upside rather than returning it. The discount to peers reflects a lower return, not a mispricing.

Our take: We side with the bulls, narrowly, and the margin is the 2027-2028 band. Priced on today's 11.5% alone, 1.22x is close to fair. But the evidence that the base return holds when trading normalizes lowers the risk on the path to 12-13%. On that path, 1.33-1.45x is reasonable. We are being paid about 13% plus a dividend to wait for it.

Debate: Is cards an investment or an erosion?

Bull view: An investment. General purpose spend is up 13% and loans 9%, losses are running at the low end of guidance, the book is prime-heavy, and the return is still above 21%. Contra-revenue is the accounting form of acquisition spend, and profit-sharing rises because credit is good.

Bear view: An erosion. Net income fell 8%, the return improved only because capital was cut by about a quarter, new account acquisitions fell despite more marketing, and management missed its own NIR guide by $99M within a quarter of setting it.

Our take: The bears have the better quarter and the bulls have the better franchise. Cards is about 16% of the firm's earnings, so even a sustained erosion moves the firm return by tens of basis points, not hundreds. We would treat a second consecutive NIR overrun in Q4 as evidence the investment is costlier than disclosed. That would weigh on conviction, but on its own it would not reverse the rating.

Model Implications

A correction to our own numbers. Our July FY2026 EPS range of $11.75-12.50 was inconsistent with the ~11-11.5% full-year return we published alongside it, which implied a figure nearer $11. We now carry $11.35-11.55, reconciled to a ~11.7% return. The change is in our arithmetic, not in Citi's performance, and it does not touch fair value, which is anchored to tangible book and the durable return.

Valuation framing. At $124.26 the stock trades at 1.22x tangible book of $101.59, about 10x our FY2027 EPS estimate before the Banamex accounting charge, and with a dividend yield near 2%. Our fair value of $135-147 applies the same 1.33-1.45x multiple we used in July to a slightly higher book. The multiple is consistent with a return on tangible equity of 11.5-12% today and a credible path to 12-13% in 2027-2028, against a cost of equity near 10%. At the reaction close, fair value of $141 at the midpoint implies about 13% upside before dividends.

The structural point about buybacks has flipped since July. At 1.32x book, the $30B program was a return of capital that diluted tangible book per share. At 1.22x it is less dilutive, and the slower pace matters less per dollar. Tangible book per share has compounded 6% over the year and should keep doing so at a ~85% payout. Even with no re-rating, that plus the dividend is close to a market return. The re-rating is the excess.

Thesis Scorecard: Post-Earnings

Graded against the standing thesis as last argued at Q2 2026.

Overall: strengthened on the bull side, mixed on the bear side, and decisively changed on price. The pillar that mattered most, sustainable double-digit returns, got its strongest evidence yet: 11.5% in a quarter with normal trading and extra costs, and a full-year guide raised above the range management would not leave in July. The two bear points that drove the Hold (valuation and the flattered 13%) have both resolved, one by price and one by the print. Two other bear points sharpened. Cards missed its own guide, and the “spend, not bank” posture now reaches the buyback. We think both are reflected in a 1.22x multiple.

Action: Upgrading to Outperform from Hold. We downgraded in April at 1.31x book and asked for a meaningful pullback. In July we made the routes back specific. One has been met: the second half is proving the waypoint was conservatism. The other is within about 2%: the stock is at 1.22x against a 1.15-1.20x trigger zone, and it closed at 1.20x on October 14. Fair value is $135-147, with the midpoint about 13% above the reaction close, plus a ~2% dividend. The risks we are accepting are a slower buyback, a cards business that costs more than guided, and an All Other line that is not yet shrinking. We would revisit the rating if Q4 falls below the high-single-digit floor, if buybacks step down again without a regulatory reason, or if cards NIR overruns again. Outperform.