Coinbase (COIN) — Leaner at the Trough. The Price Needs a Low Hurdle for a High-Beta Stock.
Coinbase is worth $71.81 a share on our model, against a $160.59 close. The third quarter itself moved value little: transaction revenue hit a two-year low, but lower costs, prediction markets and stablecoin offset most of it, and on July's method fair value moves about 4 dollars once two double counts are removed. The gap to the price comes down to two questions: what return an owner should demand from a stock with a measured beta above 3, and how likely a new crypto cycle is. Our rates are equivalent to a flat 13.4%; the price needs about 8.4%.
In July we called Coinbase the right company at the wrong price. The franchise was gaining share, and the price assumed that boom-year earnings were the normal level. Two quarters later, both halves are sharper. Coinbase's share of crypto trading volume rose for a fourth quarter, to 10.7%, and average USDC held in its products set a record at $22 billion. In the same quarter, transaction revenue fell to $590 million, 44% below a year earlier, as crypto volatility hit multi-year lows. Management cut costs again; the reported $719 million of net income is almost entirely a mark on the bitcoin and ether Coinbase holds. This note asks what a share is worth at $160.59 after the print, and what the price needs to be true.
What this note answers
- What did the quarter earn once the crypto marks are taken out, and which earnings figure is the $0.27 the brokerage feed shows?
- Is the trough-revenue floor July named ($5.5 billion) holding?
- What are the three lines that moved (prediction markets, stablecoin, the cost base) worth, and what is the legal exposure of the first?
- What is Coinbase worth after the print, and why does the number move?
- What does the $160.59 price require?
- What would change the view, and when do we find out?
Terms used throughout, defined once. Adjusted net income: Coinbase's measure, net income excluding gains and losses on crypto held for investment and on other investments, after their tax effect. Transaction revenue: fees on crypto trading (consumer, institutional) and other transactions, including prediction markets. Subscription and services (S&S): stablecoin revenue, blockchain rewards (staking), interest and finance fees, Coinbase One and other services. Stablecoin revenue: Coinbase's share of the interest earned on the reserves behind USDC, the dollar token issued by Circle. Prediction markets: exchange-listed yes/no contracts on events (sports, economics, politics, crypto prices) that pay $1 if the event occurs, offered to Coinbase customers through a regulated exchange. FCFE: free cash flow to equity, net income less the reinvestment growth needs. Cost of equity (CoE): the return shareholders require, from the capital asset pricing model (CAPM): risk-free rate + beta × equity risk premium. Beta: how much the stock moves with the market. Here Coinbase's measured beta is blended with the median of its closest listed analogs, weighting each by its precision (a Vasicek blend), so a noisy estimate leans on its analogs. ROIC: operating profit after tax over the capital invested in the business. WACC: the weighted cost of equity and debt. SBC: stock-based compensation. Normalized earnings: what a year of mid-cycle activity earns, the basis of July's method.
The case for the price, in the company's terms
The strongest reading of this quarter comes from the filings and the deck. It goes like this.
- The franchise keeps gaining in a down market. Coinbase's share of crypto trading volume rose for a fourth straight quarter, to 10.7%, from 5.6% a year earlier, with gains in both spot and derivatives. Assets on platform are $377 billion, back to a peak 12.5% of crypto market value. These are the measures that decide who earns the next cycle's volume.
- The business earns money at the bottom. Adjusted EBITDA rose 60% from the second quarter to $333 million, the fifteenth positive quarter in a row, in a quarter when crypto volatility hit multi-year lows. Adjusted expenses came in below the outlook, and the full-year guide was cut a second time.
- The revenue mix is no longer a trading bet. Subscription and services were 49% of net revenue. Stablecoin revenue rose 5% on the quarter to $307 million, on record USDC balances and higher rates, and the Circle arrangement renewed on the same terms.
- A new line is scaling fast. Prediction markets passed $200 million of annualized revenue, compounding 82% a quarter since launch, with daily traders up more than 60% on the quarter. Users who trade predictions also trade more spot crypto.
- The balance sheet is a strength. Coinbase holds $8.6 billion of cash and $11 billion of available resources. It repaid $1.3 billion of convertible notes in June from cash, and it added about 2,600 bitcoin to its holdings this year.
- Each trough is higher than the last. Trailing revenue at this trough is about $5.6 billion, against $3.1 billion for the 2023 trough year. On that view, today's run-rate is the floor of the next cycle, not its average.
- A five-year beta overstates the risk. The window covers the 2022 crypto collapse. A diversified owner who also holds crypto, or who holds Coinbase as the regulated way to own the industry, would not demand 15.8% a year from it.
What this note tests, and what it cannot. It tests claims 1 to 4 against the filings, in §1 and §2, and claim 6 against the revenue floor July set, in §1 and §4. Claim 5 is counted in the balance-sheet layer at its marks. The note adopts part of claim 7: it blends Coinbase's beta with its closest analogs by precision, and discounts the mature business at a lower 11.0% (+$16.77). It does not adopt the rest: §3 argues that the measured beta is priced risk. Our rates equal a flat 13.4%, against the proponent's 12% ($71.81 against $85.22); the larger part of the remaining gap to the price is how much weight the bull case deserves. The note cannot test the deepest claim: that crypto becomes the rail for payments, tokenized securities and automated agents, and that Coinbase earns a toll on all of it. No filing sizes that yet. The note runs the proponent's inputs in §3: on the bull case at an explicit-years rate of 12%, a share is worth $220.42, which is above the price. The disagreement is about how likely that world is and what return an owner should demand while waiting for it.
What the quarter earned
Total revenue was $1,215.6 million, flat on the second quarter and down 35% on a year earlier. Transaction revenue fell 1% on the quarter to $590.2 million (−44% on the year): consumer $450.1 million, institutional $91.7 million, other $48.5 million. Subscription and services rose 3% to $570.2 million: stablecoin $306.9 million, blockchain rewards $85.7 million, interest and finance fees $68.6 million, other $109.0 million. Corporate interest and other income was $55.1 million, down 48% on the year, which the 10-Q attributes to a 97 basis point lower average rate earned and to lower average corporate cash balances.
Exhibit 1 — Thirteen quarters of revenue: transaction against the rest
How to read it. Each bar is one quarter's total revenue in millions of dollars, Q3-2023 to Q3-2026. The light gold part is transaction revenue; the light grey part on top is subscription and services plus corporate interest and other income. The latest quarter is drawn dark. The number above each bar is total revenue. Compare the gold parts to see the trading cycle and the grey parts to see how much of revenue no longer depends on it.
The market explains the trading line. On Coinbase's own figures, total market spot volume fell 6% on the quarter and derivatives volume 11%, while volatility fell 15% to multi-year lows. Consumer spot volume on Coinbase was $23.4 billion, against $57.4 billion a year earlier. Total crypto market value rose 37% on the quarter, most of it in September. Higher prices did not bring trading back.
The $0.27 is adjusted, not reported. Net income was $718.7 million, $2.51 a diluted share. Of that, $715.0 million is the gain on the crypto Coinbase holds for investment, and $39.2 million is gains on other investments; after their $113.3 million tax effect, adjusted net income was $77.7 million. Divided by 287.98 million diluted weighted shares, that is $0.27, the figure the brokerage feed shows as the quarter's earnings per share. The reported $2.51 is mostly the prices of bitcoin and ether on 30 September.
Costs fell on the quarter while revenue did not. Adjusted EBITDA rose 60% from the second quarter to $333 million, about 40% of that from lower operating costs. Adjusted expenses (technology and development, general and administrative, sales and marketing, excluding amortization) were $975 million, below the $980 million to $1,080 million outlook and 10% below a year earlier. General and administrative fell 11% on the quarter on lower professional services spend, part of it moved into the fourth quarter, for which the adjusted-expense outlook is $1,007 million to $1,157 million. Headcount was 4,204, from 4,988 in March. Operating income was $61.7 million, helped by $49.4 million of items that do not recur (a crypto-for-operations gain, a restructuring reversal and other operating income). Transaction expense rose to 16.8% of net revenue, which the deck attributes to exchange fees on prediction markets. The full-year adjusted-expense guide is now $4,150 million to $4,300 million, from $4,250 million to $4,600 million at the start of the year.
The balance sheet. Cash was $8.6 billion. Long-term debt is $5.95 billion after Coinbase repaid $1.3 billion of 2026 convertible notes in June. Coinbase held 17,954 bitcoin and 152,649 ether for investment, up from 15,389 bitcoin in December. Buybacks stopped: 38,000 shares ($5.3 million) in the quarter, against $1.19 billion over nine months, with $2.0 billion of authorization left. One counterparty accounted for 27% of total revenue, from 19% a year earlier; the 10-Q does not name it. Our reading is that it is Circle: the deck's stablecoin revenue plus revenue on corporate USDC balances is $328 million, 27.0% of total revenue (19.0% a year earlier).
The trough floor July named holds today; the first read on the fourth quarter points below it. July said the compounding-platform view gains weight if trailing revenue holds above $5.5 billion with services still growing. Trailing revenue is $5.63 billion today and services grew 3% on the quarter, so the test is met for now. But nine-month revenue is $3.85 billion, so the fourth quarter must bring $1.65 billion for 2026 to clear $5.5 billion. October's first read points well below that. Transaction revenue was about $165 million through 25 October. At that pace the quarter brings about $607M of transaction revenue and, with the S&S outlook midpoint and corporate interest at the third quarter's level, about $1.25 billion in total, which puts 2026 near $5.1bn. The trigger is due at the 10-K; §4 records it as unresolved, with the first read adverse.
Three lines that moved
Prediction markets. The line passed $200 million of annualized revenue in the quarter, from about $100 million in June on the deck's thresholds, so the third quarter brought at least $50 million, roughly a twelfth of transaction revenue. The CFO said predictions and derivatives together were 25% of transaction revenue. The economics are thinner than spot trading. The fees Coinbase pays the exchange that lists the contracts pushed transaction expense up 32 basis points of net revenue in a quarter when spot volumes fell, and the 10-Q shows the size: "other" transaction expense, which it says rose largely on these fees, was $45.0 million against $9.6 million a year earlier. Against at least $50 million of revenue, the fees take up to about 71%. We model 70%, near the top of that range, because it is the only figure the filings give; revenue above the floor, or fees that are only part of the rise, would put the share lower, and a 50% share adds +$0.50 a share. In the company's transcript of its prepared remarks, the CEO said Coinbase now has CFTC approval to list its own contracts, "which will improve our economics". The CFTC's registrant listings show the parts of that: Coinbase Derivatives, LLC has been a designated contract market since 2020, and on 28 September 2026 the CFTC registered Coinbase Clearing LLC as a clearinghouse, limited to fully collateralized contracts. They list no separate order on prediction-market listings, and a designated market can list contracts by certifying them itself. The bull case cuts the fee share to 45%, and the base case fades it to 50% by 2031.
The legal ground. One question decides the line, and Robinhood faces it too: whether federal commodities law displaces state gaming law for sports contracts. The Third Circuit ruled for Kalshi at the preliminary-injunction stage (New Jersey, Pennsylvania, Delaware). The Ninth Circuit held on 28 August that federal law does not displace state law (KalshiEX v. Assad), binding nine states including California and Nevada; Kalshi's petition for rehearing by the full court is pending. On 25 September the Sixth Circuit ruled against Kalshi in Ohio and Tennessee, affirming the denial of its preliminary injunction in Ohio and vacating the one it had won in Tennessee. New Jersey's petition asking the Supreme Court to decide (Flaherty v. KalshiEX, No. 26-299) is pending; on 29 October Kalshi asked for more time to respond. Kalshi matters because Coinbase told the Nevada court that it offers the contracts through a partnership with Kalshi, acting as a CFTC-registered intermediary between its customers and Kalshi's exchange; it has not named the exchange in an SEC filing. Coinbase Financial Markets is also a party itself, and the rulings in its own cases have gone against it. In Nevada, the Gaming Control Board sued on 2 February, and on 26 March the court issued a preliminary injunction barring sports, election and entertainment contracts in the state, concluding, at that preliminary stage, that federal commodities law does not displace the state's gaming law. In Connecticut, a federal judge denied Coinbase's own request for a preliminary injunction against the state's gaming regulators on 10 August. New York's Attorney General filed a petition on 21 April, and Coinbase said it had removed the case to federal court. The 10-Q discloses no prediction-market litigation as material: it names no exchange, reports no change in risk factors, and mentions neither action. Coinbase does not report the line by state, so we carry the exposure as a grid. Population stands in for customers, and the sports share is 50%, from the CFO's statement that more than half of the quarter's contracts were not sports. Coinbase discloses no state where it has stopped offering the contracts; apart from Nevada, the grid assumes every state is in today's run-rate.
Exhibit 2 — Prediction-market revenue at risk, by legal outcome
How to read it. Each bar is one legal outcome. Its length is the prediction-market revenue that outcome removes, in millions of dollars a year on the third-quarter run-rate (the $200 million floor). The label gives the share of the line. Revenue lost = the outcome's share of the US population × 50%, the sports share (the CFO said more than half of the quarter's contracts were not sports). Population stands in for customers because Coinbase does not report revenue by state. The gold bar is the base case. Preemption removes nothing from the run-rate, so its bar sits at zero.
The base case assumes the split stands: the states bound by the Ninth and Sixth Circuit rulings are lost, 15% of the line. The bear case assumes the Supreme Court sides with the states in mid-2027 and sports contracts end everywhere (50% of the line from then). The bull case assumes preemption, which keeps the whole line. Applied to the base case, the full legal swing moves fair value by about a dollar a share (−$1.00 for the states, +$0.43 for preemption). The line matters more for what it says about Coinbase's ability to add products than for its present value.
Why the New York and Nevada actions are not priced. Neither filing states a total. Nevada asks for declarations, an injunction, and its attorneys' fees and costs. New York asks for an injunction, an accounting, restitution to customers, disgorgement, damages, a penalty of three times Coinbase's gain, $100,000 for each offering of sports wagering, prejudgment interest and $2,000 in costs; press reports of a $2.2 billion demand do not match the petition. Disgorgement and the treble penalty scale with what Coinbase earned in New York. Taking the gain as gross prediction-market revenue before the exchange's fees, New York's share as its 5.8% share of the US population (a proxy), and revenue since launch at about $87 million (built from the company's run-rate statements), they come to about $20 million, 7 cents a share, or 22 cents at three times that share. Restitution, measured by customers' losses, and the per-offering penalty cannot be sized from the record; each thousand offerings counted would add $100 million, 36 cents a share. The larger measurable risk is losing the state: an injunction removing the whole line in New York costs about 36 cents a share across the three cases. For scale, the bear case already removes 25% of the line from 2027, worth 14 cents a share at its 25% weight, and the Supreme Court siding with the states, applied to the base case, costs $1.00. If Coinbase has complied in Nevada, which it does not report, the base and bear cases count Nevada's loss twice, an overlap worth 1.5 cents a share. Each measurable amount from the two actions is under half a dollar on a fair value of $71.81, so none gets a bridge row. Restitution on a scale the petition does not state, a court counting penalties by the thousands of offerings, or a Supreme Court ruling for the states would change that.
Stablecoin. Revenue was $306.9 million, up 5% on the quarter and down 5% on the year, an annualized $1.23 billion on the current definition ($1.31 billion on July's, before revenue on corporate USDC balances was reclassified) against the $1.6 billion trigger July set. It is the product of two things. Balances are at records: average USDC held in Coinbase products was $22 billion, from $15 billion a year earlier, about 30% of the $73 billion average USDC in circulation. The rate is the other factor. The Fed raised rates in September, and the effective fed funds rate averaged 3.67% in the quarter. It held at 3.75% to 4.00% on 28 October, with two dissents in favor of another increase. The Fed's September projections put the policy rate at 4.1% at the end of 2026 and 2027, then 3.9% in 2028, 3.6% in 2029 and 3.2% in the longer run. We run stablecoin revenue on that path: each year's revenue is balances times the projected rate. The base case grows balances 10% into 2027 and 15% a year to 2030, giving $1,509M in 2027, and lets the rate path take back about a fifth of the yield by 2030. The bear case uses the low end of the Fed's projections and a 25% cut in Coinbase's share at the next Circle renewal in 2029. The line is most exposed to rate cuts: the bear-case rate path on the base case costs −$2.87 a share.
The cost base. July set each scenario's fixed costs against a run-rate of $4.9 billion. The same measure today, built from the full-year guide, is $4,384M, about 10.5% lower. On July's method we keep July's ratios: the base case's 2027 fixed costs fall to $4,205M from $4,700 million, the bear's to $3,847M. On the new model the base case takes the FY26 cost base itself, without amortization ($4,239M), since the May cuts are already in it; the bear's further cut stays as a scenario assumption. The bull keeps July's $5,400 million, because a boom year would bring the spending back. The cut came as revenue fell. It is also real. Headcount is down 16% since March, and the guide was lowered twice.
What Coinbase is worth after the print, and why the number moved
July reproduced. July's engine, run unchanged, gives $102.25. It values each scenario's normalized 2027 earnings at 25 times (base) and 28 times (bull), the bear at 1.5 times revenue, and adds net corporate cash and the investment portfolio at market value.
Correction: cash counted once. July added net corporate cash at face value, and its 2027 earnings also included the interest that cash earns (corporate interest and other income, $280 million in the base case, less $50 million of interest expense), capitalized at 25 to 28 times. The same dollars were counted twice. Taking corporate interest out of capitalized earnings, and keeping the cash at face, moves July's value to $88.12 (−$14.13). This is an error in July's arithmetic, not a change of view. Correction: convertible notes counted once. July's 288 million diluted shares included about 11 million shares the convertible notes would convert into (the third-quarter earnings-per-share note counts 10.9 million), while its balance sheet also deducted the notes in full as debt. At $160 the notes are counted as debt, so the shares come out: +$3.46, to $91.59.
The quarter's data on July's method. Exhibit 3 re-runs July's engine with one input at a time moved to today's value. Shares and the 30 September balance sheet, with crypto marked to the 29 October close (bitcoin $80,743), add +$0.36. Re-basing the bear and base cases' trading lines to the third-quarter trough takes away $30.26: July's base case put consumer trading revenue about 23% above the first-quarter run-rate; the same recovery now starts from a consumer run-rate 28% lower once prediction-market revenue is set aside (reported consumer revenue is 21% lower, $450 million against $567 million in the first quarter), and an institutional run-rate 32% lower. Staking takes away $1.95. Prediction markets add +$5.64, stablecoin on the Fed's path adds +$4.18, and the lower cost base adds +$17.87. On July's method the quarter moves fair value to $87.42, $4.17 below the corrected July number.
Exhibit 3 — From July's $102 to today's value, one factor at a time
How to read it. The first bar is July's fair value a share; the last is today's. Each bar between is the change in dollars a share from moving one factor to today's value, in the order shown, on the same engine: gold adds, grey subtracts. The two light grey bars are the corrections (cash and convertible notes each counted once). The bars between the dotted lines are the quarter's data on July's method, which end at the value printed above them. The pale dark bar is the change of method; the bars after it are the five calls on the new model. The dashed line is the price.
| Step | Kind | $ a share | Fair value after |
|---|---|---|---|
| Correction: cash counted once (corporate interest no longer capitalized beside the cash) | correction | -14.13 | 88.12 |
| Correction: convertible notes counted once (if-converted shares out of the count; the notes stay deducted as debt) | correction | +3.46 | 91.59 |
| Shares: Q3 diluted weighted, less if-converted (277.09M) | data | +0.01 | 91.59 |
| Balance sheet at 30 Sep, crypto marked to 29 Oct (convertible repaid; BTC 17,954 at $80,743) | data | +0.35 | 91.94 |
| Transaction lines re-based to the Q3 trough (bear, base) | data | -30.26 | 61.69 |
| Staking (blockchain rewards) re-based to the Q3 trough (bear, base) | data | -1.95 | 59.74 |
| Prediction markets added as a leg (Q3 run-rate, FY27 growth, legal grid, 70% exchange fees) | data | +5.64 | 65.37 |
| Stablecoin on the Fed's projected rate path and USDC balances | data | +4.18 | 69.55 |
| Fixed costs re-based to the FY26 expense guide (bear, base) | data | +17.87 | 87.42 |
| Method: normalized-earnings multiple -> multi-stage FCFE (stages argued, GDP terminal, CAPM 16.2%) | method | -29.24 | 58.18 |
| Regulatory net capital deducted: $1,538M required (CB Inc. 1,185, CCTC 222, other 131), not excess cash | house call | -5.55 | 52.63 |
| Acquired-intangible amortization ($145M, non-cash, finite acquisition schedules) out of the perpetual fixed-cost base | house call | +2.96 | 55.59 |
| Base 2027 fixed costs = the FY26 cost base, flat nominal (July's 0.959 ratio not re-applied; the May cuts are already in it) | house call | -1.87 | 53.72 |
| Base other services re-based to the Q3 run-rate ($68.6M + $109.0M) x 4 | house call | +0.66 | 54.39 |
| Base prediction-market fee share fades 70% -> 50% over 2027-31 (own listings under the new CFTC approval) | house call | +0.66 | 55.04 |
| Cost of equity: beta 2.395 (own 3.218 blended toward Robinhood and Interactive Brokers) -> 15.8% explicit, converging to 11.0% for the terminal value (mature-platform beta 1.327; from a flat 16.2%) | cost of equity | +16.77 | 71.81 |
| Fair value 102.25 → 71.81 | -30.44 | 71.81 |
Each row re-runs the same engine with one more factor at today's value, in the order shown. The July engine is run unchanged ($102.25; July printed each scenario to the cent). Rows sum to the move to within a cent of display rounding. Kinds: correction (two errors in July's arithmetic: cash and convertible notes each counted twice), house call (five input changes on the new model), cost of equity (blended beta for the explicit years, stable rate for the terminal value), data (filed figures and 29 October marks, on July's 25 / 28× multiples and the bear case's 1.5× revenue), and method (the change of valuation model and its cost of equity; the only row that is not data or a correction). The workbook's Attribution sheet carries every row.
The method change, and why it earns its place. July capitalized normalized earnings at 25 and 28 times. We now require each model's long-run growth to land on the economy's rate (4.0%, from the Fed's longer-run projections of 2.0% real growth and 2.0% inflation; 3.0% in the bear case, below the bottom of the Fed's longer-run range of 1.7% real growth plus 2.0% inflation) and every stage of faster growth to be argued from what drives it. A multiple does neither: at the 11.0% stable rate, 25 times earnings assumes about 6.7% growth a year forever, and 28 times about 7.2%; at the 15.8% rate of the explicit years, 11.3% and 11.8%. The lines also run on different clocks: prediction markets on a court calendar, stablecoin on the rate path and the 2029 Circle term, trading on the crypto cycle. One multiple cannot carry three clocks. July chose 25 to 28 times deliberately, above the mature-exchange multiples it compared against (CME 19.6 times, Cboe 21.3 times), to credit the stablecoin and services annuity, and it said those multiples were already generous against a cash-flow perpetuity. That premium now has its own lines, so keeping the premium multiple as well would count the annuity twice. We model each line year by year from 2027 to 2040, discount free cash flow to equity at the middle of each year, and expense SBC in the cost base. Buybacks have offset most of the share issuance since late 2024, so the share count is held at the third quarter's diluted 277.09 million, without the convertible shares, since the notes are deducted as debt. Reinvestment is growth divided by a 30% return on new capital, inside the 18% to 58% range Coinbase earned on capital excluding goodwill across 2024 and 2025.
The model choice changes almost nothing by itself. At 12.0%, the stage model (before the calls below) returns $87.42, the same value as July's multiples on today's data. The whole method step, −$29.24, is the cost of equity, taken at a flat 16.2% on Coinbase's own Blume-adjusted beta. July's multiples valued Coinbase as if owners needed about 12.0% a year. July's multiples are kept as a sensitivity and as an unweighted cross-check.
The cost of equity. The rate is a judgment about Coinbase's risk, argued from the evidence, with the standard reads shown beside it. Coinbase's own five-year monthly beta against the S&P 500 is 3.22, with a correlation of 0.60 and a standard error of 0.56. We read it as priced risk, not noise: it is more than five standard errors from zero, it holds across windows (the last three years 4.01, the full listing 3.19), and it has a cause, since Coinbase's revenue is a levered claim on crypto trading, which rises and falls with risk appetite. We shrink it toward its closest listed analogs, the trading platforms Robinhood (2.17) and Interactive Brokers (1.33), weighting each estimate by its precision: the blend is 2.40, and with the 10-year Treasury at 5.05% and a 4.5% equity risk premium the explicit-year rate is 15.8%. Schwab (0.77) is left out. About a third of Coinbase's revenue is interest earned on stablecoin reserves, loans and corporate cash, which is the case for keeping it, but most of that income rides on USDC balances that grow and shrink with crypto activity, not on a bank's deposit franchise; with Schwab in the blend the value is $80.12. The exchanges are left out because their revenue rests on hedging, clearing and data rather than retail speculation: CME (0.29) and Cboe (0.43) are derivatives marketplaces built on risk management, and ICE (0.95) earns more than half its revenue from derivatives and listings exchanges, the rest from data and mortgage technology, while Coinbase's retail volume falls with risk appetite. The terminal value describes the mature business, which in the model still earns about 60% of its revenue from crypto trading and staking, so it stays cyclical. Its analog is a mature trading platform: Interactive Brokers' beta of 1.33, just above the 0.8 to 1.2 range usually used for mature companies, gives 11.0%. Each case moves from 15.8% to 11.0% in a straight line across its convergence stage (2036 to 2040; the bear 2032 to 2034). The change from the flat 16.2% of the method row is its own row: +$16.77.
| Read | Beta | Explicit years | Terminal | Value a share |
|---|---|---|---|---|
| Own beta, unadjusted | 3.22 | 19.5% | 11.0% | $58.15 |
| Own beta, Blume-adjusted | 2.49 | 16.2% | 11.0% | $70.10 |
| Chosen: blend toward Robinhood and Interactive Brokers | 2.40 | 15.8% | 11.0% (beta 1.33) | $71.81 |
| Blend toward Robinhood, Interactive Brokers and Schwab | 2.16 | 14.8% | 10.4% (beta 1.2) | $80.12 |
| Blend toward CME, ICE and Cboe | 1.66 | 12.5% | 8.6% (beta 0.8) | $114.39 |
| Last three years, unadjusted | 4.01 | 23.1% | 11.0% | $48.56 |
| Method row: one flat rate | 2.49 | 16.2% | 16.2% | $55.04 |
Five calls on the new model. Five inputs are set where the evidence puts them on the new model, each its own row in the attribution. (a) Required regulatory capital is not excess cash. Coinbase's regulated subsidiaries must hold $1,538 million of net capital (Coinbase, Inc. $1,185 million, its trust company $222 million, others $131 million, per the 10-Q), so it comes out of the cash counted at face: −$5.55. The requirement grows with the crypto Coinbase holds for customers, so it is also a reinvestment need the model does not charge. Treating it instead as the present value of the interest it forgoes gives $75.82 (+$4.01). (b) Amortization of acquired intangibles, about $145 million a year (quarterly amortization rose from about $5 million to $34–36 million after the 2025 acquisitions, Deribit the largest), leaves the perpetual cost base in every case: it is non-cash and finite, and growth reinvestment is charged separately (+$2.96). (c) The base case takes today's cost base as its 2027 level, without re-applying July's 4% discount to it (−$1.87). (d) The base case's other services start at the third quarter's run-rate, interest and finance fees plus other services times four (+$0.66). (e) In the base case the prediction-market fee share falls from 70% to 50% by 2031, as Coinbase lists its own contracts (+$0.66). Together they net −$3.13, to $55.04.
How long each line grows faster than the economy. The stage table is the model's spine. Each length is set by an observable, and the convergence to 4.0% is a straight line. Coinbase's own history is the check on the trading stage. Transaction revenue grew about 19% a year from the 2023 trough to this one (2023: $1.52 billion; 2026: about $2.55 billion, nine months reported plus the fourth quarter at October's pace, which management cautions against extrapolating), but that includes the Deribit acquisition and the new prediction-market line. Without predictions and derivatives (25% of the third quarter's transaction revenue, on the CFO's figure), the rate is about 8% a year, which is why the base case takes 7% for two halving cycles; trading at 19% a year for eight years in the base case gives $112.65. July said that if the trough floor held, the compounding view would carry fair value toward $130 to $160. On this model the same evidence, four more years of the trading stage, is worth +$9.77. Most of the difference is the cost of equity, not the evidence: at July's implied 12.0%, the floor holding takes the value from $85.76 to $97.66.
| Business | Stage | Driver | What sets the length | Evidence | Years | Growth bear / base / bull |
|---|---|---|---|---|---|---|
| Crypto trading and staking | Adoption | Crypto trading activity beyond the economy, plus share gains | Two four-year bitcoin cycles (the 2028 and 2032 halvings); bear one cycle | Volume share 5.6% → 10.7% in a year; transaction revenue ~19% a year, 2023 to 2026 at October's pace (~8% without predictions and derivatives) | 8 (bear 4) | 4% / 7% / 12% |
| Crypto trading and staking | Convergence | Share gains run out | Straight line to terminal by 2040 (bear 2034) | Above-median growth shows little persistence beyond chance, and very high long-run growth is rare (Chan, Karceski and Lakonishok, 2003) | 5 (bear 3) | to 3% / 4% / 4% |
| Stablecoin | Balance growth | USDC balances under the GENIUS Act; on balances held in its products Coinbase keeps the reserve income (per Circle's filings, as described in our July note) | To 2030: Treasury Secretary Bessent's $3 trillion stablecoin figure for 2030 (remarks of 12 November 2025), and the August 2029 end of the renewed Circle term (three-year terms, per Circle's filings as described in our July note) | USDC in Coinbase products $22bn from $15bn in a year; USDC in circulation $73bn, down from $77bn in Q2 | 3 (bear 2) | 5% / 15% / 30% a year on balances; rate on the Fed's path |
| Prediction markets | Scale-up | Users and categories, inside each case's legal outcome | The Supreme Court's decision on preemption, due by mid-2027 or mid-2028 | Over $200M annualized; 82% compound quarterly growth; daily traders +60% in the quarter | 1 (bear and bull 2) | 5% / 25% / 40% after 2027 |
| Prediction markets | Convergence | The category matures with a settled rulebook | Straight line to terminal by 2032 (bull 2035) | Exchange fees take up to about 70% of the revenue on the 10-Q's figures; the base case fades that to 50% by 2031 and the bull to 45%, with own listings and in-house clearing | 4–5 | to 3% / 4% / 4% |
| Other services | Platform growth | Coinbase One subscribers and the lending book | The subscriber base's present pace of growth | Borrow and lend balances $1.9bn from $0.6bn in a year; record paid subscribers | 4 (bull 6) | 3% / 6% / 9% |
Fixed costs grow 4.5% a year in the base case through 2035 and 11% in the bull, where a boom's revenue brings marketing and hiring back with it. That puts terminal operating margins at 24% and 41%, against 20% in 2025 and 35% in 2024. In the bear case, costs fall 18% and then 12%, a scenario assumption (headcount is already down 16% since March), and the margin settles at 10%.
| Bear | Base | Bull | |
|---|---|---|---|
| Weight | 25% | 50% | 25% |
| 2027 revenue ($M) | 3,588 | 5,599 | 9,313 |
| 2027 net income, ex corporate interest ($M) | −$536M | $389M | $1,931M |
| Terminal operating margin | 10% | 24% | 41% |
| Terminal growth | 3.0% | 4.0% | 4.0% |
| Terminal value, share of value | 47% | 48% | 54% |
| Value a share (incl. $14.56 of cash and holdings, net of required capital) | $21.48 | $49.77 | $166.24 |
| July's method on today's data | $39.54 | $53.29 | $203.57 |
Why 25 / 50 / 25. The cases stand for a range of outcomes, not three exact futures. Decision analysts commonly weight outcomes at the 10th, 50th and 90th percentiles of a range at 25 / 50 / 25 or 30 / 40 / 30, so the test is whether the cases sit near those points. On transaction revenue, the line that most separates them, they roughly do. Before prediction markets, 2027 transaction revenue is $1.65 billion in the bear case, $2.73 billion in the base and $5.45 billion in the bull. From 2021, the year of the listing, annual transaction revenue has been $6.84 billion, $2.36 billion, $1.52 billion, $3.99 billion, $4.06 billion and about $2.55 billion for 2026 at October's pace: one year of six above the bull's level, one below the bear's, and three on each side of the base. Prediction markets and Deribit are in the later years, but taking them out (the $152 million of derivatives growth in 2025; predictions and derivatives, 25% of the third quarter, from 2026) leaves each year in the same band. Two adjustments could shift the count. Adding 2020 ($1.10 billion, before the listing and at a smaller scale) makes two years of seven below the bear's level. Scaling earlier years up for today's volume share, which has nearly doubled in a year, cuts the other way: a factor of about 1.4 lifts 2024 and 2025 above the bull's level and takes 2023 above the bear's. We do not scale them, because share counts all volume, derivatives included, and revenue has not followed it: the consumer line is about three quarters of transaction revenue, and its spot volume fell 59% in the year share nearly doubled. A down market explains part of that, so the question stays open. Six years is a check, not a measurement, and today's evidence does not separate the outer cases either: the record volume share favors the bull case, the third quarter's transaction revenue and October's first read favor the bear, and our regime composite, a blend of bitcoin market signals, reads +0.10, neutral. So we weight them equally. Two years of six fell outside the outer cases, more than the one in five those percentiles imply, which if anything leans toward the wider 30 / 40 / 30; six years cannot settle that, so 25 / 50 / 25 is a judgment. The weights matter, because fixed costs make earnings swing harder than revenue and the bull case supplies $41.56 of the $71.81: 30 / 50 / 20 gives $64.58, 20 / 50 / 30 gives $79.05 and 30 / 40 / 30 gives $76.22.
Why the bull case keeps July's trading levels. Re-basing to the third quarter moves a recovery that starts from today's run-rate, so it applies to the bear and base cases. The bull case describes a new crypto cycle, and a trough says little about how high such a cycle goes. Transaction revenue has turned quickly before: about 3.7 times in the two quarters after the third quarter of 2023 and about 2.7 times in the quarter after the third quarter of 2024. It has not always held: the turn after the second quarter of 2025 reached about 1.4 times, and revenue has since fallen four quarters in a row, from $1,046 million to $590 million (the gold bars in Exhibit 1). The bull's $5.45 billion is below 2021's $6.84 billion and a little below the $5.64 billion annualized pace of the fourth quarter of 2024 and the first of 2025, but no year since 2021 has reached it. With staking, its trading line starts 2027 at $6.2 billion and compounds 12% a year for eight years, to about $15.4 billion in 2035, so the choice carries through the whole path. Re-based, the bull's 2027 transaction revenue would be about $3.94 billion, which three years of six exceeded, two of them by less than 3%, too often for a 90th-percentile case. Weak quarters bear on how likely that cycle is, which is the weight, not on its size. For the reader who disagrees, the same ratios applied to its revenue alone give $116.58 for that case and a fair value of $59.40. The bull also keeps July's fixed costs, less the same $145 million of amortization; cutting them in proportion as well gives $135.56 and $64.14. On either, even the re-based bull case sits below the $160.59 price.
Sensitivities. Holding the stable rate, an explicit rate one point lower or higher gives $76.38 or $67.66; holding the explicit rate, a stable rate one point lower or higher gives $77.88 or $67.25. At a flat 12% the weighted value is $85.22; on the last three years' unshrunk beta (4.01), $48.56. A reader who sees Coinbase simply as a broker would use the brokers' own median beta (1.33): 11.0% for the explicit years and a value of $98.27. We keep Coinbase's own beta in the blend because, at a standard error of 0.56, it still carries information about Coinbase (44% of the weight). The crypto it holds does not inflate that beta: the $2.2 billion held for investment is about 5% of its market value. Discounting the terminal value at the top of the usual mature range (beta 1.2, 10.4%) instead of 11.0% gives $75.23. If new investment in the terminal value earns only the stable 11.0% rather than 30%, the value is $63.98 (−$7.83). Terminal growth one point lower or higher moves the value by under 50 cents. Four more years of the trading stage give $81.58 (+$9.77); four fewer give $61.80. A flat rate path from 2027 gives $82.15. Removing prediction markets gives $65.61; doubling them, $77.98. A 15 / 50 / 35 weighting, the kind of shift a turn in the cycle would justify, gives $86.29. Leaving the bear and base trading lines at July's normalized levels instead of re-basing them to the trough gives $89.14.
Relative cross-check (unweighted). Listed exchanges (CME, ICE, Cboe, Nasdaq) trade at a median 24.5 times trailing earnings; brokers (Robinhood, Schwab, Interactive Brokers) at 32.0. On the base case's 2027 earnings per share of $1.40 plus cash and holdings, those give $49.00 and $59.44. At the stable rate those multiples would need 7.2% and 8.1% growth a year forever, against 4.0% for the economy; they hold for Coinbase only if its growth stays well above the economy's for long. The price is 104 times the base case's 2027 earnings, after cash and holdings.
Exhibit 4 — Return on invested capital against the cost of capital, twelve quarters
How to read it. The gold line is Coinbase's trailing-twelve-month return on invested capital each quarter from December 2023 to September 2026: operating income after the trailing effective tax rate (21% where that rate is not meaningful) over average equity plus debt minus cash. The grey dashed line removes goodwill and intangible assets from capital. The dashed horizontal line is the weighted cost of capital, 14.6%, held flat; the shaded area below it is where the business earns less than its capital costs. The label on each point is the return.
The return on capital follows the cycle. It rose from below zero at the end of 2023 to 43% a year later, on the trading boom that followed the bitcoin ETFs, and has fallen every quarter since, to 1.5% at September. Quarterly transaction revenue fell from $1.56 billion (Q4-2024) to $590 million, while goodwill and intangibles reached $5.4 billion after the Deribit acquisition; excluding them, the return is 2.9%. The regime is weakness. The May cost cuts reach adjusted EBITDA this quarter but not yet a trailing return; that waits for revenue to turn.
What the price requires. The market values Coinbase's operations, after cash and holdings, at $40.5 billion. On July's frame that is normalized net income of $1.62 billion at 25 times, without corporate interest, against $1.26 billion in 2025 and $105M of adjusted net income over the last four quarters. On this note's model, three readings get there. The weighted value equals the price at a flat cost of equity of about 8.4%. The bull case alone equals it at 13.7%. At our rate, the base case gets there only if its trading lines compound about 22% a year for eight years, taking 2036 net income to about $9.5 billion (Exhibit 5).
Exhibit 5 — The base case's earnings and the earnings the price needs
How to read it. Both lines are Coinbase net income in billions of dollars, 2027 to 2036, before corporate interest. The gold line is the base case. The grey dashed line is the same case with the trading lines compounding 22% a year for eight years instead of 7%: the path on which the base case is worth $160.59 at our rates (15.8% converging to 11.0%). The labels at the right give each line's 2036 level. The gap between the lines is the growth the price assumes beyond the base case.
Exhibit 6 — Fair value against the cost of equity
How to read it. The horizontal axis is the cost of equity, 7% to 19%; the vertical axis is fair value in dollars a share. The gold line is the probability-weighted value (25 / 50 / 25). The grey line is the bull case alone. The dashed line is the $160.59 price. Where a line crosses the price is the cost of equity at which that case is worth the price. The curves use one flat rate for every year; the gold dot is our value (15.8% converging to 11.0%), placed at the flat rate that gives the same value (13.4%). Readers can find their own rate on the axis and read the value.
The receipt: the triggers July left armed
July left six triggers. None fired. One, the Circle renewal, had already resolved on the benign branch it named, and the 10-Q now confirms it. The receipt reads each against the filings.
Closed at $160.59 on 29 October; the 52-week low was $139.11 on 31 July. The trigger never fired. It is retired with this note: at a fair value of $71.81, $120 is no longer the boundary of the analysis, and the reversal condition in this note replaces it.
Q3 stablecoin revenue $306.9M, an annualized $1.23bn, up from $1.17bn in Q2 (+5% on the quarter) on record USDC balances ($22bn average in Coinbase products) and a higher policy rate. On July's definition, before revenue on corporate USDC was reclassified, $1.31bn. Still $0.29bn short. Re-armed as cq3-stablecoin with balances and the rate read separately.
Trailing revenue $5.63bn is above $5.5bn today and S&S grew 3% on the quarter, but nine-month revenue is $3.85bn, so Q4 must bring $1.65bn. October's transaction revenue (~$165M through 25 Oct) paces Q4 at about $1.25bn with the S&S outlook midpoint, FY2026 about $5.10bn. Due with the 10-K (28 Feb 2027); on the first read the compounding-platform condition July named is on course to be missed.
Our regime composite read +0.097 (Neutral) on 2026-10-29: neither winter nor a bullish turn. The trigger asks for a flip into bullish; no change to the 25% bull weight. Due 31 Dec; carried to that date.
H.R. 3633 is not law. The Senate Banking Committee reported it on 1 June; on 15 September cloture on the motion to proceed failed 49-50 (Record Vote 234), and Senator Tillis entered a motion to reconsider (congress.gov actions, read 29 October). No further action is recorded. Carried, armed.
Recorded on 31 August as a three-year renewal on unchanged terms. The Q3 10-Q now says so in Coinbase's own words: it "renewed our arrangement with Circle Internet Financial, LLC on the same terms". No 8-K amended the agreement. The next renewal point is August 2029; the bear case in this note cuts Coinbase's share of the line by a quarter there.
What carries the story to the next update
- COIN Q4-26 transaction revenue (by 2027-02-11). The base and bear cases re-base trading to the Q3 trough ($590M). October ran about $165M through the 25th, a ~$607M quarter. Reads: Q4-26 transaction revenue >= $740M (25% above Q3) or <= $530M (10% below). Silence: Between the two = no change. Prediction-market revenue inside the line is read separately if disclosed; a rise carried by it alone is read under cq3-prediction-markets, not here. Model consequence: At or above $740M: bear and base trading lines up 25%, FV +10.64/share. At or below $530M: down 10%, FV -4.26/share.
- COIN FY2026 revenue against the $5.5bn floor (by 2027-02-28). Continues coin-trough-revenue-floor on the 10-K: FY2026 needs Q4 >= $1.65bn; October's pace gives about $5.10bn. Reads: FY2026 total revenue >= $5.5bn with S&S growing (holds) or < $5.5bn (fails). Silence: If the 10-K is late, the Q4 8-K deck is read. A miss with S&S growing is recorded as a fail: the floor was named on total revenue. Model consequence: Holds: the compounding view gains weight; trading stage +4 years in all cases, FV +9.77/share. Fails: no change (the trough is already in the base).
- COIN stablecoin revenue: balances and the rate (by 2027-07-30). Stablecoin revenue = balances x rate. Base: average USDC in Coinbase products grows ~15% a year from $22bn; the policy rate follows the Fed's September path (4.1% end-2026 and 2027). Reads: average USDC in Coinbase products >= $28bn or <= $18bn by the Q2-27 print; or effective fed funds <= 3.1% / >= 4.4%. Silence: Silence = between the bands: no change. A balance rise carried by one partner's deposits (Hyperliquid's in June) is read as one event, not a trend. Model consequence: Balances strong: base stablecoin +15%, FV +3.53/share; weak: -15%, FV -3.53/share. Rates cut to the SEP low path: FV -2.87/share; held high: FV +2.51/share.
- COIN prediction-market revenue and its legal ground (by 2027-06-30). Q3 run-rate >$200M (floor used: $200M). Base FY2027 after the circuit split: ~$272M. The Supreme Court petition on state preemption (Flaherty v. KalshiEX, No. 26-299) is pending. Reads: annualized prediction-market revenue >= $400M or <= $150M on company disclosure; or a Supreme Court grant, denial or decision on 26-299. Silence: No disclosure = no change. A grant changes nothing until the decision; a denial confirms the base case (the split stands). Model consequence: Revenue strong (x1.6): FV +1.46/share; weak (x0.6): FV -0.97/share. Decision for the states: FV -1.00/share; for preemption: FV +0.43/share.
- COIN FY2027 adjusted-expense outlook (by 2027-02-11). The base case's 2027 fixed costs are $4,205M (July's ratio on the FY26 guide midpoint, $4,225M adjusted). Reads: FY2027 adjusted-expense guide midpoint <= $3.95bn or >= $4.55bn. Silence: No full-year guide = read the Q1-27 quarterly outlook times four. A rise that management ties to revenue-generating spend (USDC rewards) is read with cq3-stablecoin. Model consequence: Guide low: bear and base fixed costs -$300M, FV +5.48/share. Guide high: +$300M, FV -5.49/share.
- COIN five-year beta (the cost of equity) (by 2027-10-31). House CoE 15.8% explicit = DGS10 5.05% + beta 2.395 (own 3.218, SE 0.563, blended toward the HOOD/IBKR median 1.748) x 4.5%; 11.0% (IBKR beta 1.327) for the terminal value. The value is more sensitive to this than to any operating input. Reads: re-measured raw 60-month beta <= 2.72 or >= 3.72 at the next re-measure (explicit ke about 1pp lower / higher, peers unchanged). Silence: The beta and the peers' betas are re-measured at each update and the rate re-argued; a move inside the band is applied without comment. Model consequence: Explicit ke -1pp: FV +4.56/share. +1pp: FV -4.16/share.
- COIN 2027 earnings before marks (the reversal condition) (by 2028-02-28). Base-case 2027 net income ex corporate interest $389M ($1.40/share); bull $1,931M; bear -536M. Reads: toward: FY2027 adjusted net income less corporate interest after tax >= $1.93bn, or the regime composite bullish with Q4-26 transaction revenue >= $740M; away: FY2027 <= the bear case (-0.54bn), or the composite bearish with Q4-26 transaction revenue <= $530M. Silence: Between the two = no change. Read on Coinbase's adjusted figure less corporate interest after tax, never reported net income (it carries the bitcoin mark). Model consequence: Toward: half the weight to the bull case, FV +36.19/share; with a measured beta near 1.3 (CoE ~12%) as well, FV +69.25/share. Away: half the weight to the bear case, FV -36.19/share.
- COIN: the Circle arrangement at its next renewal (by 2029-08-31). Renewed in August 2026 on the same terms (10-Q); the next renewal point is August 2029. The bear case cuts Coinbase's share of stablecoin revenue by 25% from 2030. Reads: an amendment to the revenue split, or Coinbase acquiring Circle. Silence: Silence through August 2029 = a further renewal on the same terms: the bear case's cut is removed. Model consequence: A 25% cut in the base case: FV -4.63/share. An acquisition is a new valuation, not a trigger consequence.
What would change our view
Toward the price. Either of two observable conditions. Coinbase's 2027 earnings before marks reach the bull case's level ($1,931M of net income, about $6.97 a share); or our regime composite turns bullish and fourth-quarter transaction revenue reaches $740 million. Then half the weight moves to the bull case, and fair value rises +$36.19 to $108.01. If, on top of that, Coinbase's measured five-year beta falls to about 1.3, so that the rule gives about 12% for the explicit years, the same half-weight value is $141.06, which is 12% below the price, just outside the 10% band this note calls fair. If 2028 earnings before marks also reach the bull case's level, the bear case stays out and 52% of the weight goes to the bull case. That is the condition under which we would call the price fair.
Away from it. Either of two observable conditions, the mirror of the above. Coinbase's 2027 earnings before marks fall to the bear case's level (a loss of $536M or worse); or the regime composite turns bearish and fourth-quarter transaction revenue falls to $530 million or less. The fourth-quarter bands differ in percent because $740 million is roughly the base case's 2027 quarterly pace arriving early, while $530 million would be a new low below the trough. Then half the weight moves to the bear case, and fair value falls $36.19 to $35.62.
What fair price means here. The note calls the price fair when the probability-weighted value is within 10% of it. On today's weights, that needs a flat cost of equity near 8.4%. In observable terms: 2027 and 2028 earnings before marks both at the bull case's level, plus a 60-month beta at or below 1.3 (an explicit-year rate near 12%). At that rate the bear case drops out, and 52% of the weight on the bull case, with the rest on the base case, puts the value within 10% of the price; the toward branch's 50% falls just short ($141.06).
The call
The quarter sharpened both halves of July's call. The company is leaner, its share of trading is at a record and it has a new product that is scaling. The trading revenue that pays for the platform is at its lowest in two years, and October has started at about the third quarter's pace: $165 million of transaction revenue through 25 October, against $590 million for all of the third quarter. On July's own method, once its two double counts are removed, all of that nets to a few dollars: $87.42 against $91.59.
Valued line by line, with each growth stage argued and discounted at 15.8% while it grows faster than the economy and 11.0% once it is mature, Coinbase is worth $71.81 a share (bear $21.48, base $49.77, bull $166.24), against $160.59. The return we require, a flat 13.4% equivalent, is 1.4 points above the 12.0% July's multiples implied; what separates the value from the price is mostly how likely the bull case is. The price is fair if owners need a flat 8.4% a year, or on the bull case alone at 13.7%. Exhibit 6 shows what each reader's own hurdle rate implies. The Fund holds no position.
Sources & Method
Filings. Coinbase Global 8-K of 29 October 2026, Exhibit 99.1 shareholder deck (accession 0001679788-26-000121): revenue, expense and non-GAAP reconciliations, operating metrics, outlook. Form 10-Q for the quarter ended 30 September 2026 (0001679788-26-000122): revenue disaggregation (Note 5), crypto held for investment (Note 9), debt, equity statement, earnings per share, the Circle renewal and the counterparty concentration. Q2 8-K deck and 10-Q (0001679788-26-000087, -000088) for comparisons. 8-Ks of 2 September and 2 October 2026 (Items 5.02). The July model (research/coin-valuation/model.py) and the July 30 addendum.
Call. Coinbase's own transcript of its pre-recorded prepared remarks (investor.coinbase.com, 29 October 2026), the only text of that presentation; the live question-and-answer session on X in two independent transcriptions (Investing.com and Benzinga, 29 October), checked against a third (gloom.sh). Figures from the live Q&A are used only where both transcriptions agree. The non-sports share of contracts and the CEO's remark on CFTC approval come from the company's own written transcript of the prepared remarks, which neither the deck nor the 10-Q repeats.
Weights. Annual transaction revenue from Coinbase's Form 10-Ks: 2020 to 2022 from the 2022 filing (0001679788-23-000031), 2023 from the 2023 filing (0001679788-24-000022), 2024 from the 2024 filing (0001679788-25-000022) and 2025 from the 2025 filing (0001679788-26-000015); 2026 from the third-quarter 10-Q plus October's pace. The three-point weighting convention as summarized by Strategic Decisions Group from Hammond and Bickel, "Reexamining Discrete Approximations to Continuous Distributions," Decision Analysis, 2013.
Courts and Congress. Congress.gov actions for H.R. 3633 (the CLARITY Act), read 29 October 2026. Supreme Court dockets 26-299 (Flaherty v. KalshiEX) and 26-338 (Robinhood Derivatives v. Dreitzer), read 29 October. The Sixth Circuit's opinion in KalshiEX v. Schuler, Nos. 26-3196/5235 (25 September 2026), and the District of Connecticut's order in Coinbase Financial Markets v. Tong, No. 3:25-cv-02121 (10 August 2026), both from govinfo.gov; the Third and Ninth Circuit rulings as described in Robinhood's third-quarter 10-Q litigation note. Coinbase's 10-Q legal note (Note 20) and Item 1. The New York Attorney General's verified petition (Supreme Court, New York County, Index No. 451604/2026, filed 21 April 2026) and press release. The Nevada Gaming Control Board's complaint (First Judicial District Court, Carson City, No. 26 OC 00030 1B, 2 February 2026), restraining order (5 February) and preliminary-injunction order (26 March), from the Board's site. The CFTC's registrant listings for designated contract markets and clearing organizations, read 1 November 2026; self-certification of new contracts under 17 CFR 40.2(a). Coinbase's statement on removing the New York case, as reported by Cointelegraph, April 2026.
Market and macro. Settled closes, monthly bars (Coinbase, the S&P 500 ETF, Robinhood, Interactive Brokers, Schwab, CME, ICE, Cboe) and peer trailing price-to-earnings multiples from Robinhood market data, 29 October 2026; the after-hours print (about $151 in the evening of 29 October) is a data point, not an input. Bitcoin and ether at the 4:00 pm ET close from Coinbase Exchange hourly candles. FRED: DGS10 (10-year Treasury, 5.05% on 28 October, the latest posted) and DFF (effective fed funds, monthly). Federal Reserve Summary of Economic Projections, 16 September 2026 (longer-run growth and inflation; federal funds medians, central tendencies and ranges). Census Bureau state population estimates, vintage 2024 (the legal grid's proxy). TON618's BTC regime composite (Signal Ledger), read 29 October.
Method. Multi-stage FCFE, 2027–2040, by business line, mid-year discounting from 29 October 2026; terminal growth 4.0% (3.0% bear, below the Fed's longer-run range) on the following year's earnings after reinvestment at a 30% return on new capital. SBC is expensed inside fixed costs; the share count is held at the third quarter's 277.09 million diluted weighted shares less the convertible shares (the notes are deducted as debt), since buybacks have offset most issuance. Cash and holdings are counted once, net of the $1,538 million of net capital the regulated subsidiaries must hold (10-Q, regulatory capital requirements), at 30 September balances with crypto marked to 29 October (other crypto moved with bitcoin and ether; other crypto and strategic investments at a 15% haircut, as in July). Corporate interest is excluded from earnings, and amortization of acquired intangibles (FY2026 guide $140–150 million) from the perpetual cost base. Cost of equity, argued in §3 with the reference reads beside it: 15.8% for the explicit years (a precision-weighted blend of Coinbase's 60-month beta with the median of Robinhood and Interactive Brokers), converging across each case's convergence stage to 11.0% (Interactive Brokers' beta, 1.33) for the terminal value; WACC 14.6% for the return-on-capital exhibit (debt at its $5.4 billion fair value, at the 10-year rate plus one point, after a 20% tax). Fourth-quarter 2026 cash flow is not counted. Every figure is computed by the scripts in data/ and carried into the model workbook.
Limitations. Prediction-market revenue is not disclosed; the run-rate uses the deck's $200 million floor, and the fee share is our estimate from the 10-Q's "other" transaction expense. The New York and Nevada actions are sized in the text but not priced; New York's restitution claim and per-offering penalty cannot be sized from the petition. Population is a proxy for customers by state, and the sports share is a contract share from the CFO's remarks, used as a revenue share. The trough re-basing assumes July's recoveries from the first-quarter run-rate apply to the third quarter's. Beta is measured on 60 months that include the 2022 crypto collapse; it is blended by precision with the median of Robinhood and Interactive Brokers, and the other reads are shown in §3. The terminal rate is a judgment about how cyclical the mature business stays. The Circle renewal cut in the bear case and the rate paths are scenarios, not forecasts. No bitcoin price return is assumed: holdings are valued at the 29 October close.
Disclosures
Information only. TON618 Capital. This report is for information purposes only. Nothing here is an offer to sell or a solicitation of an offer to buy any security, fund interest, or digital asset, and nothing here is personalized investment advice or a recommendation regarding any instrument.
Publisher's exclusion. All research is published solely as general, impersonal information of regular circulation. It is not tailored to the objectives or circumstances of any individual and is not issued in connection with compensation from any client. The Fund has no clients and distributes all research free of charge. On that basis it publishes in reliance on the publisher's exclusion from the definition of "investment adviser" under the Investment Advisers Act of 1940 (§202(a)(11)(D); cf. Lowe v. SEC, 472 U.S. 181 (1985)).
Registration & conflicts. TON618 Capital is not registered as an investment adviser or broker-dealer in any capacity. The Fund is a Bitcoin fund and may hold or transact in the securities or digital assets it discusses; material conflicts are disclosed where they exist. Ownership: the Fund holds no position, long or short, in COIN or in any other security named in this report as of the report date, and has no economic interest in their price. Compensation: the Fund received no compensation from any party in connection with this report and charges nothing for it.
Use of AI. Artificial intelligence is used in the creation of this research. All methodology and data integrity are reviewed and approved before publication by TON618 Capital's Chief Investment Officer, Keyth Beck; errors may nonetheless occur, and readers should verify independently.
CFA. This report was prepared to align with CFA Institute analytical standards (methodology only). CFA® and Chartered Financial Analyst® are registered trademarks owned by CFA Institute. That reference describes the analytical framework applied; it does not imply the report was prepared, reviewed, or authored by a CFA charterholder, and the report is not issued, reviewed, endorsed, certified, or approved by — nor affiliated with — CFA Institute.
Risk & feedback. Past performance is not indicative of future results. Digital assets and equities are volatile and may result in total loss of capital. Corrections and feedback are welcome — please direct them to CIO Keyth Beck at keyth@ton618capital.com. Version 1.0.