Rocket Companies
America's biggest home lender and the company that collects payments on about one mortgage in seven, plus the Redfin home-search site.
The scores
Growth2/4 ↓
Revenue doubled by buying Redfin and Mr. Cooper, but Rocket's own loan locks fell 15% and the market is forecast to shrink in 2027.
Business quality3/4 →
The biggest servicing book feeds its next loans at fat direct margins, but brokers' loans earn little and borrowers who shop save.
Financial position3/4 →
Notes cost 5.43% and almost none fall due before 2029, but book value moves with servicing-rights marks.
Execution record3/4 ↑
Merger savings are ahead of plan and margins rose four quarters running, though the record is short and most of it is two deals.
Per share capture2/4 →
The share count rose by about a third to pay for the deals, and pay rewards adjusted totals rather than return per share.
Potential2/4 →
A $2 trillion lending market falling toward $1.9 trillion, where Rocket already makes about one loan in twelve; upside needs lower rates.
Risk profile3/4 →
One founder controls the votes and a big lock-up ends in mid-2027, but the business is spread across millions of borrowers.
Valuation2/4 →
At 3.09 times tangible book the price needs returns near 26% for years, about twice the 13.6% earned in the last four quarters.
Evidence quality3/4 →
Most claims are filed figures; the moat and the market rest on agency data and a rival's 10-K, and the return needed is our arithmetic.
The Verdict
Neutral, with medium conviction. The business is Mixed: servicing fees now pass a billion dollars a quarter and merger savings are lifting profit, but Rocket's own lending is shrinking. The price is Full. At 3.09 times tangible book, which is what the company owns minus what it owes and minus goodwill from its deals, the price needs a lasting return near 26% on that equity, about twice the 13.6% it earned. Fannie Mae sees the lending market shrinking again in 2027, so the recovery the price pays for is not yet in sight.
The long caseWhen rates fall, millions of borrowers refinance, and the biggest servicer is first in line to make those loans.
The evidence under this - 6 claims
- The savings are management's own count and unaudited.
- Part of the Q3 cost fall is volume-driven pay that returns when lending recovers.
Servicing rights are carried at fair value and moved by $710M in one quarter, so book value swings with rates.
Q4 2025 was the first quarter with Mr. Cooper, so the year mixes integration costs and a market that is now smaller.
- Goodwill from two stock deals is excluded, so on full book of 1.49 times the needed return is near 14%.
- A mortgage lender's return swings with the market; 2027 could be lower and 2029 far higher.
- The MBA's October forecast is higher, near $1.99 trillion for 2027.
- Fannie Mae notes there is no universal source for market-wide originations data, so these are its estimates.
The company
Rocket lends money to buy or refinance a home, directly through its own app and website and through independent brokers who sell its loans. It then services loans: it collects the payments, handles escrow and talks to the borrower for the life of the loan.
It services $2,007B of loans for 9.1 million borrowers. It also owns Redfin, a home-search site and brokerage, and uses AI to approve loans and answer calls with fewer staff. Without it a borrower goes to a bank or another lender.
The evidence under this - 3 claims
None of these is audited, and Rocket gives no cost per loan to show the saving.
How it makes money
Two streams. Rocket earns a gain when it sells each new loan to Fannie Mae, Freddie Mac or Ginnie Mae: 4.31% of the loan's value when it found the borrower itself, 0.70% when a broker did. Then it earns a yearly servicing fee on each loan it keeps servicing, $1,032M in Q3 2026. Mortgages made a $1,138M contribution in the quarter; Redfin and the rest $189M.
The evidence under this - 4 claims
Where the money comes from, stream by stream
| Stream | Share of sales (nine months to 2026-09-30) | How it earns it |
|---|---|---|
| Gain on sale of loans | 40% | The profit on making and selling loans, including the value of the servicing right kept; $3,609M. |
| Servicing fee income | 35% | Fees for collecting payments on $2.0 trillion of loans; $3,181M. |
| Interest income | 18% | Interest on loans before they are sold and on borrowers' escrow deposits; $1,640M. |
| Other income | 17% | Title and closing fees, Redfin commissions, Rocket Money subscriptions and Rocket Loans; $1,524M. |
| Change in value of servicing rights | -10% | The fall in value as loans pay off, net of hedges and rate moves; −$945M. |
Who runs it
Varun Krishna has run Rocket since 2023, after Intuit, and led both deals. Founder Dan Gilbert controls 57.2% of the votes, so outside holders cannot change the board. The chief executive was paid $52.9M for 2025, mostly in stock; the bonus rides on adjusted revenue, profit and market share, and none of it on return on equity. Insiders sold 555,000 shares in a year, all under pre-set plans, and bought none.
The evidence under this - 3 claims
Why it wins, and who could beat it
Why customers come back
Size of the servicing book is the edge. Rocket holds the payment relationship with about 15% of all US mortgage debt, so when one of those borrowers wants a new loan, Rocket knows first and can call first. Management says recapture hit a record, and 98% of clients stayed over twelve months.
That is why its direct loans earn so much more than brokers' loans. The limit is shopping: Freddie Mac found a borrower who asks four lenders can save over $1,200 a year.
The evidence under this - 5 claims
- Rocket gives no recapture rate in the 10-Q, only a net client retention rate.
- The 65% goal is Rocket's own and has a date more than a year away.
The New York Fed figure includes home equity lines and loans Rocket could never service, and is dated June rather than September.
The threat
PennyMac's own annual report names Rocket, UWM, Rithm and Freedom as its rivals. UWM fights for the brokers' loans, where Rocket's locks fell 28% in a year. Banks still make over a third of loans, and AI shopping agents could one day pick the cheapest lender for the borrower.
The evidence under this - 4 claims
None of these is audited, and Rocket gives no cost per loan to show the saving.
The rivals: UWMC, PFSI, RITM, LDI, and 4 more that are private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| UWMC | United Wholesale MortgagePriced against it: Yes | Mortgage: origination and servicing | On sale | Yes |
| PFSI | PennyMacPriced against it: Yes | Mortgage: origination and servicing | On sale | Yes |
| RITM | Rithm Capital (Newrez)Priced against it: Not checked | Mortgage: origination and servicing | On sale | Not checked |
| Private | Freedom MortgagePriced against it: Private | Mortgage: origination and servicing | On sale | Private |
| Not listed | Banks and credit unionsPriced against it: Not listed | Mortgage: origination and servicing | On sale | Not listed |
| Not listed | Consumers' own AI shopping agentsPriced against it: Not listed | Mortgage: origination and servicing | Announced | Not listed |
| Not listed | Other real estate websites and brokeragesPriced against it: Not listed | All Other: Redfin, Rocket Money and Rocket Loans | On sale | Not listed |
| LDI | loanDepotPriced against it: Yes | - | On sale | Yes |
How big it could get
How big this could get
Americans borrow about $2.1 trillion a year to buy or refinance homes, and Fannie Mae expects $1.9 trillion in 2027. Rocket made about one loan in twelve last quarter. The stock of debt to service is $13.1 trillion, of which Rocket holds about 15%. Lower rates would grow the first figure fast, since refinancing is what falls hardest when rates are high.
The evidence under this - 4 claims
- The MBA's October forecast is higher, near $1.99 trillion for 2027.
- Fannie Mae notes there is no universal source for market-wide originations data, so these are its estimates.
Closed volume includes correspondent loans bought from other lenders, which Rocket did not originate.
The New York Fed figure includes home equity lines and loans Rocket could never service, and is dated June rather than September.
Who has signed
The agencies hire Rocket to service $2,007B of loans for as long as each loan lives, though they can end that at any time. Compass signed a three-year alliance to send more listings to Rocket's buyers, with no volume promised.
The evidence under this - 3 claims
The results
The numbers
Sales grew 31% in the year to Dec 2025, about the same pace as the 34% in the year before.
It lost 1.0% of sales.
Yearly sales growth has ranged from −55% to +34% over the last 5 years, with sales falling in 3 of them.
Shareholders' money lost 0.4% a year.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-31. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | |
|---|---|---|---|---|---|
| Size and growth | |||||
| Sales | $12.9B | $5.84B | $3.80B | $5.10B | $6.70B |
| Sales growth | −17% | −55% | −35% | +34% | +31% |
| Profit | |||||
| Net margin | 2.4% | 0.8% | −0.4% | 0.6% | −1.0% |
| What growth costs, per sales dollar | |||||
| General and admin | 9.2% | 16% | 21% | 18% | 21% |
| Pay in shares | 1.3% | 3.7% | 4.7% | 2.8% | 5.1% |
| Capital spending | 0.9% | 1.6% | 1.6% | 1.3% | 1.4% |
| Cash | |||||
| Cash from operations | $7.74B | $10.8B | $111M | −$2.63B | −$3.93B |
| Free cash flow | $7.63B | $10.7B | $51.0M | −$2.70B | −$4.02B |
| Balance sheet | |||||
| Cash and short-term investments | $2.13B | $722M | $1.11B | $1.27B | $2.70B |
| Debt | - | - | - | - | $10.4B |
| Net cash | - | - | - | - | −$7.73B |
| Returns and shares | |||||
| Shares (diluted) | - | - | - | 141M | 1.32B |
What growth costs
Growing has been bought, not earned: the deals doubled revenue and added amortisation and stock pay. Since then cost savings have made each new dollar more profitable, with margins rising four quarters running. Office and pay costs rose while revenue first collapsed and then recovered.
The evidence under this - 3 claims
- The savings are management's own count and unaudited.
- Part of the Q3 cost fall is volume-driven pay that returns when lending recovers.
The last report (29 Oct)
Quarter ended September 2026, filed figures.
In the quarter to September 2026 adjusted revenue rose to $2,574M and adjusted profit to $419M, against $158M a year earlier, on a servicing fee line 2.5 times as large after Mr. Cooper. Net income of $709M looks better still, but $710M of it was a rise in the value of servicing rights, which the adjusted figures take out.
Underneath, lending shrank: rate locks fell 15%. Cost savings carried the profit, and the guide for Q4 points to adjusted revenue of $2.1B to $2.3B, lower than Q3.
What it did to the thesisIt proved the servicing fees and the merger savings are real and growing, which supports the business case. It did not show lending recovering, and the weaker guide leaves the return the price needs further away, so the view stays Neutral.
The quarter in detail
Sales growth from the previous quarter, filed figures. The company has restated its own history, so a quarter either side of the restatement may describe a different business; n/c marks one that does not compare.
Earnings per share $0.15 against $0.14 expected; $0.25 as filed, the gap being what the company adds back.
The evidence under this - 6 claims
- The savings are management's own count and unaudited.
- Part of the Q3 cost fall is volume-driven pay that returns when lending recovers.
- The finance chief expects the refinance market to fall by almost half from Q3, so the guide rests on share gains.
- Q4 is seasonally weak, 15% to 18% below Q3 in a normal year, management says.
What management did with the money
Rocket paid for Redfin and Mr. Cooper with new shares, which is why the count rose by about a third. It sells part of its servicing rights for cash when the price is good, keeping the work and the right to the next loan. It has bought back no stock.
Insiders in 2026Q3: 2 purchases and 13 sales, from their Form 4 filings.
The evidence under this - 3 claims
The price, and where we disagree
In short. Full: the price pays for a lending recovery and a return about twice today's, which no forecast shows yet.
The shares fell by more than half from January to October as forecasts for rates rose, then rose 7.4% the day after the Q3 print. Two dates matter next. The Q4 results in late February will show whether revenue lands inside the guide. And on 30 June 2027 about 924 million locked-up shares become free to sell. Under both, Fannie Mae's forecast says the loan market shrinks again next year.
The evidence under this - 4 claims
- The finance chief expects the refinance market to fall by almost half from Q3, so the guide rests on share gains.
- Q4 is seasonally weak, 15% to 18% below Q3 in a normal year, management says.
- The MBA's October forecast is higher, near $1.99 trillion for 2027.
- Fannie Mae notes there is no universal source for market-wide originations data, so these are its estimates.
How we read the multiple
To be worth this price, Rocket must earn nearly twice its present return for good. On full book, counting goodwill, it looks fairer beside PennyMac and loanDepot, but Rocket paid for that goodwill in its own shares.
- Take tangible equity of $11.7B, what Rocket owns minus what it owes and minus goodwill.
- The market pays 3.09 times that.
- With an 11% required return and 4% growth, that multiple needs about 26% earned on it each year.
- Rocket earned 13.6% over the last four quarters.
Against its own history. Cheaper than its own normal on full book, but book now carries goodwill from two stock deals, so the old normal priced a smaller company with almost none.
The evidence under this - 4 claims
Servicing rights are carried at fair value and moved by $710M in one quarter, so book value swings with rates.
Q4 2025 was the first quarter with Mr. Cooper, so the year mixes integration costs and a market that is now smaller.
- Goodwill from two stock deals is excluded, so on full book of 1.49 times the needed return is near 14%.
- A mortgage lender's return swings with the market; 2027 could be lower and 2029 far higher.
The vendor's tangible book multiple is unavailable for every peer, so the comparison is on full book, which flatters a serial stock acquirer.
Priced against its rivals
The competitors its own annual report names, each confirmed against the rival's filing. Measured on price for each $1 of net worth, cheapest first:
| Ticker | Company | price for each $1 of net worth |
|---|---|---|
| LDI | loanDepot | 0.7× |
| PFSI | PennyMac Financial Services | 0.8× |
| RKT | Rocket Companies | 1.5× |
| UWMC | UWM Holdings | 2.5× |
The middle of the other 3 is 0.8. Rocket Companies, Inc. is 2nd dearest of 4.
Rivals' figures come from a data provider, so read the spread of the group rather than any single pair.
Rocket sits above the lenders that look most like it and below one, so the premium is its own: the servicing book and the direct channel, not the group.
Where we disagree with the market
A lender hostage to rates: shares sold by more than half this year, most analysts at hold, and over a quarter of the float sold short.
Steadier than the market thinks, since fees and savings keep profit up, but still priced for a recovery nobody forecasts for 2027.
The selling looks overdone on the business: profit rose every quarter while the shares fell, and short sellers hold 27.8% of the float. Against us, the price on tangible equity already pays for a strong recovery, and analysts have grown more positive, not less.
What analysts say, and the evidence
The evidence under this - 5 claims
- Goodwill from two stock deals is excluded, so on full book of 1.49 times the needed return is near 14%.
- A mortgage lender's return swings with the market; 2027 could be lower and 2029 far higher.
What could go wrong, and how we would know
The risks
- Lending keeps shrinking. Fannie Mae sees $1.9 trillion of loans in 2027, and Rocket's own locks already fell 15%.
- Brokers' loans slip away. Rocket Pro locks fell 28% in a year and earn 0.70% of loan value.
- A wave of shares to sell. About 924 million Class L shares can be sold from 30 June 2027.
- The agencies hold the contracts. Fannie, Freddie and Ginnie set servicing fees and can end Rocket's servicing at any time.
- Borrowers shop around. A borrower who asks four lenders can save over $1,200 a year, which caps direct margins.
- Book value moves with rates. Servicing rights are marked to market and moved $710M in one quarter.
The evidence under this - 6 claims
- The MBA's October forecast is higher, near $1.99 trillion for 2027.
- Fannie Mae notes there is no universal source for market-wide originations data, so these are its estimates.
What we still do not know
- What share of borrowers leaving Rocket's servicing book take their next loan from Rocket, against other servicers?
- How much revenue and profit Redfin makes on its own inside All Other.
- Rocket's share of all home loans measured by someone outside the company, not by its own estimate.
- How long mortgage rates stay above 7%.
The evidence under this - 4 claims
- Rocket gives no recapture rate in the 10-Q, only a net client retention rate.
- The 65% goal is Rocket's own and has a date more than a year away.
Closed volume includes correspondent loans bought from other lenders, which Rocket did not originate.
- The MBA's October forecast is higher, near $1.99 trillion for 2027.
- Fannie Mae notes there is no universal source for market-wide originations data, so these are its estimates.
What would change our view
- Neutral→AttractiveThe price asks for a return in reach
It needs:
- ○The price falls to about 2.0 times tangible equity, where it needs an 18% lasting return rather than 26%.checked daily
- Neutral→AttractiveThe return shows up in the filings
It needs all of:
- ○Adjusted return on tangible equity reaches 20% a year on a quarter's pace, against about 14% in Q3 2026.checked at the next results, due by 25 Feb
- ○The price is at or below about 2.3 times tangible equity, where it needs a 20% lasting return.checked daily
We would also be wrong if
- Q4 2026 adjusted revenue below the $2.1B low end of guidance.
- Servicing fee income below $900M in a quarter.
- Mr. Cooper savings short of the $400M yearly pace by early 2027.
What is coming
- 25 February 2027 Q4 2026 results: Adjusted revenue against the $2.1B to $2.3B guide, whether costs fall to about $1.85B, and whether the Mr. Cooper savings are complete.
Every dated event ahead
- 25 February 2027 Q4 2026 results: Adjusted revenue against the $2.1B to $2.3B guide, whether costs fall to about $1.85B, and whether the Mr. Cooper savings are complete.
- 30 June 2027 Class L-2 share lock-up ends: Whether Dan Gilbert or other former holders sell any of the 924 million shares that become free to sell.
Since we wrote this
- 31 OctThesis writtenNeutral, medium convictionA well-placed lender and servicer earning at the bottom of its market, at a price that already needs about twice today's return.
Sources 14
Sources - 14
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-03-02business, competition classes, acquisitions and their terms, agency dependence, governance and control, risk factorsbacksC01C02C21C37C39C42C45
- S02primarycompany filing2026-10-29Q3 statements and reconciliations, volumes and margins, liquidity, MSR sales, the Q4 guide, segment results
Generated Q3'26 total revenue, net of $3.28 billion and adjusted revenue of $2.57 billion. Reported Q3'26 GAAP net income of $709 million and adjusted net income of $419 million.
backsC03C04C05C13C16C18C34C35C36C40C44 - S03primarytranscript2026-10-29Rocket Companies Q3 2026 earnings call transcript (FMP)data/cache/transcripts/RKT-2026Q3.jsonmarket share estimates, cost savings, the Q4 guide and costs, recapture, AI and Redfin claims; questions from Zelman, Goldman Sachs, Morgan Stanley, BTIG, StoneX, Wells Fargo and KBW
Based on our estimates, purchase market share approached 6.5% and refinance market share was 15.5%. ... we expect adjusted revenue to be between $2.1 billion and $2.3 billion.
backsC08C13C15C16C34C35C43 - S04primarycompany filing2026-10-30share count, borrowings and maturities, segment note, key indicators, servicing data, tax receivable agreementbacksC01C03C05C06C07C14C18C19C20C36C42
- S05primarycompany filing2026-04-29ownership and voting power, the Class L lock-ups, executive pay and how it is earnedbacksC21C22
- S06primarycompany filing2026-02-20PennyMac Financial Services, Inc. Form 10-K for the fiscal year ended December 31, 2025 PennyMac Financial Servicesa rival naming Rocket Mortgage among the non-bank producers and servicers it competes withbacksC28
- S07primaryindustry data2026-10-12single-family originations by quarter and year, purchase and refinance, and the 30-year rate path, 2025 to 2027
Single-Family Mortgage Originations 379 517 511 585 571 553 513 448 377 507 501 480 1,992 2,085 1,865
backsC09 - S08primaryregulator2026-08-11Federal Reserve Bank of New York, Household Debt and Credit Report, Q2 2026 Federal Reserve Bank of New YorkUS mortgage balances outstanding and new originations, from the central bank's credit panel
Mortgage balances declined by $74 billion in the second quarter and totaled $13.1 trillion at the end of June.
backsC11 - S09primaryregulator2024-10-07Consumer Financial Protection Bureau, 'CFPB Report Finds Significant Drop in Annual Mortgage Applications and Originations in 2023' Consumer Financial Protection Bureauwhat a mortgage costs the borrower and the non-bank share of lending, from HMDA data
The median total loan costs for home purchase loans was about $6,700 in 2023, while for refinance loans the loan cost was over $7,300. ... independent mortgage companies originated nearly 62% of all closed-end home purchase loans and over 64% of refinance loans.
backsC29 - S10primaryindustry data2023-02-16what shopping between lenders is worth to a borrower
homebuyers can potentially save $600-$1,200 annually by applying for mortgages from multiple lenders, according to new research by Freddie Mac.
backsC30 - S11secondaryvendor data2026-10-31Research pack (FMP market tier and FINRA short interest), frozen as content/RKT/research-pack.jsoncontent/RKT/research-pack.jsonprice and returns, the analyst split and its history, short interest, the filed cohort's multiples and marginsbacksC27C31C32C33
- S12primarycompany filing2026-10-30every open-market trade in the year, each with its Rule 10b5-1 flagbacksC23
- S13primarycompany filing2026-10-31six years of filed revenue, costs, stock pay and capital spendingbacksC38
- S14primarycompany filing2026-02-26Q4 and full-year 2025 results and reconciliations, the Compass alliance
Announced three-year strategic alliance with Compass International Holdings to expand housing inventory
backsC37C41