COIN — Lean bearish · edge 5/10

Thesis

Today's drop is crypto-beta, not mispricing: COIN trades as a levered Bitcoin proxy (3.4-3.8 equity beta, 72% idiosyncratic share vs SPY, only 0.18 correlation to XLF), and Bitcoin sold off sharply below ~$77k as the post-election rally faded (T4 sourcing caveat applies). Beneath the daily noise, fundamentals are deteriorating — transaction revenue -43% YoY, MTUs down to 7.4M from 9.3M, 2026 consensus now a loss, 2027 EPS cut ~23-28% over 90 days across vendor snapshots — while the stock still commands ~55x 2027 consensus EPS. Down 54% is not the same as cheap.

Bull case

15 consecutive quarters of positive adjusted EBITDA ($333M, +60% QoQ); opex guidance lowered to $4.15-4.30B; subscription/services at 49% of net revenue with record Coinbase One subscribers; prediction markets >$200M annualized and derivatives ~25% of transaction revenue post-Deribit; Binance's hack could migrate trust and volume to the regulated U.S. incumbent; the Q4 run-rate is holding at Q3 levels rather than deteriorating further; any crypto recovery gives 2x upside beta.

Bear case

Even after -54%, the stock is ~55x a 2027 EPS estimate cut ~23-28% in 90 days with no run-rate evidence of the recovery consensus embeds; 2026 consensus is now a loss; the core trading business shrank -43% YoY with users leaving; the Binance hack is a sector-wide trust shock; the most prominent new target ($120) sits below the current price.

What would change our mind

If the Q4 transaction-revenue pace inflects clearly above ~$250M/month (a sustained step above the Q3 ~$200M/month pace) while crypto stabilizes, the 'no recovery' leg fails — note the current ~$200M/month disclosed pace sits at the boundary, already consistent with Q3, so this threshold requires a genuine inflection
If 2027 consensus EPS stops falling and the stock holds below ~$100, the 'still expensive' claim weakens materially
If subscription/services revenue demonstrably decouples earnings from spot volumes (e.g., a down-BTC quarter with rising EBITDA), the proxy framing must be revised

Since publication vs SPY · closed horizons are frozen

No price data after the entry date yet.

The full research everything the view rests on

Direct answer to the question asked

The market appears to be reading today correctly: COIN's -10.9% is a levered reaction to a sharp Bitcoin selloff below ~$77k (per T4 market coverage — no higher-tier source in the bundle confirms the day's price action), not idiosyncratic news; its correlation to financials (XLF) is only 0.18, so 'moving alone vs. financials' is the wrong benchmark. The real problem is that even after -54% in a year, the stock trades at ~55x a 2027 consensus EPS of $2.47 that has been cut ~23-28% in 90 days, above Morgan Stanley's new $120 target. The one number to check: Q4 transaction revenue run-rate, disclosed at ~$165M through October 25 — that pace decides whether 2027 estimates survive or get cut again.

Where the edge would come from

What the price assumes

At ~$134.75 and ~55x 2027 consensus EPS of $2.47, the price still assumes a meaningful 2027 crypto-volume recovery materializes — consensus $6.56B revenue for 2027 versus a Q4 transaction-revenue run-rate (~$165M through October 25, ≈$200M/month) that is only flat versus the depressed Q3 pace and implies no recovery at all.

Why the market may be wrong

Consensus may still be too high rather than too low: the disclosed Q4 pace (~$200M/month) is roughly in line with Q3's $590M quarterly print but shows no inflection toward the levels embedded in $6.56B 2027 consensus revenue, and Morgan Stanley just cut 2027 revenue 28% and EBITDA 47%. The opposite risk: the market may be underpricing the diversification story (subscription/services at 49% of net revenue, prediction markets >$200M annualized) and potential share gains from Binance's $6.3B hack driving flow to the regulated U.S. incumbent.

The one thing to watch

Stop benchmarking COIN against financials — with 0.18 correlation to XLF and ~2x leverage to Bitcoin, the correct daily comp set is BTC, and the correct fundamental conversion metric is the monthly transaction-revenue run-rate ($165M through Oct 25, ≈$200M/month). That number is currently flat vs. Q3 — not collapsing — which means the bear case rests not on deterioration today but on the absence of the recovery that $6.56B 2027 consensus revenue and $2.47 EPS require; watch whether the run-rate inflects above or breaks below the ~$200M/month line.

Key evidence (8)

Q3 2026 (Oct 29, T1 IR/T2 Reuters): revenue $1.216B, slightly below ~$1.233B consensus; transaction revenue -43% YoY to $590M; MTUs 7.4M vs 9.3M
Q3 headline EPS $2.51 was cosmetic: ~$715M crypto-asset gains vs only ~$62M operating income (T1/T4)
Q4 transaction revenue ~$165M through October 25 (T1 Q3 disclosure) — ≈$6.6M/day or ~$200M/month, roughly flat vs Q3's $590M quarterly pace but implying no recovery vs 2027 consensus
Morgan Stanley downgrade to Underweight, PT $120 from $258, 2027 revenue cut 28%, EBITDA cut 47% (CNBC, Nov 2, T3)
2026 consensus EPS now -$0.36 (loss); 2027 EPS cut to $2.47, down ~23-28% over 90 days across vendor snapshots ($2.47 vs $3.19; $3.02 vs $4.21) (T3/T4)
Bitcoin fell sharply below ~$77k on Nov 5 and crypto-linked equities (HOOD, MSTR, Bullish) fell ≥9% — per T4 Yahoo/aggregated market coverage only; no T1-T3 source in the bundle corroborates the day's price action
$6.3B Binance hack confirmed by Reuters/Bloomberg, attributed by FBI to North Korea's TraderTraitor; Binance covered ~95% from company funds (T2, Oct 29-Nov 3)
Quant: 60d correlation 0.18 to XLF vs 0.53 to SPY; beta 3.4-3.8; stock -54% 1y, ~10% above 52-week low

Catalysts and risks

Catalysts

Q4 2026 earnings (likely February 2027, unconfirmed)
Monthly/QTD transaction revenue updates and Bitcoin price/volatility trajectory
Post-election crypto regulatory and market regime clarification
Further analyst estimate revisions following Morgan Stanley's cut
Binance hack fallout: fund flows, regulatory response, potential share migration to Coinbase

Risks

High volatility (70% annualized) makes position sizing hazardous in either direction
Crypto gains can again mask weak operating income, complicating earnings reads
A crypto rebound would squeeze shorts rapidly given 2x beta
Estimate data is inconsistent across vendors (-$0.36 to -$0.50 for 2026 EPS), adding uncertainty to the 'expensive' claim
The same-day catalyst explanation rests on T4 coverage; if the day's driver was something else (e.g., an unflagged company-specific item), the 'market read it right' verdict weakens

What to watch

The one number: Q4 transaction revenue run-rate vs the $165M-through-Oct-25 (≈$200M/month) disclosure — inflection above ~$250M/month = recovery; break below = fresh cuts
Bitcoin price around $77k — the real-time tell for COIN's earnings power
2027 consensus EPS ($2.47) revision direction over the next 90 days
Subscription and services revenue vs the $550-630M Q4 guide
Evidence of volume/share migration from Binance post-hack
T1-T2 confirmation (or contradiction) of the Nov 5 crypto-selloff narrative

Market snapshot when the call was made 2026-11-04

Price

Return 1w	-6.7%
Return 1m	-18.4%
Return 3m	+4.2%
Return 6m	-23.5%
Return 1y	-54.1%
From 52-week high	-54.1%
Realized vol, 20d	69.9%
SPY 3m / 1y	-0.2% / +12.2%
QQQ 3m / 1y	-3.2% / +9.5%

Technicals

COIN technicals as of 2026-11-04: last close $151.50; 20d SMA 165.76 (bearish); 50d SMA 177.37; 12d EMA 156.82; 26d EMA 165.77; RSI 14 30.4 (balanced); 20d vol 68.0%; volume x1.0 vs 20d avg

Verifier notes REVISE · opinion revised once

Unsupported claims

'Q4 transaction revenue ~$165M through October 25... implying a pace well below Q3's $590M quarterly print' — the arithmetic doesn't support 'well below': $165M over 25 days is ~$6.6M/day (~$198M/month), which extrapolates to ~$600M for a 92-day quarter, i.e. roughly in line with Q3's $590M. The opinion's own invalidation condition (~$200M/month pace) confirms this run-rate ≈ Q3, not far below it. The figure is genuinely weak versus the recovery implied by 2027 consensus, but not versus Q3.
'2027 EPS cut ~25-40% in 90 days' — the evidence shows $2.47 vs $3.19 (-23%) in one snapshot and $3.02 vs $4.21 (-28%) in another. The ~40% figure is only obtainable by comparing $4.21 from one vendor's old snapshot against $2.47 from a different vendor's current snapshot — a cross-vendor mix the evidence itself warns about. The defensible range is ~23-28%.

Weakly sourced

The central same-day driver — 'Bitcoin falling 5-6% below $77k' and 'crypto-linked equities (HOOD, MSTR, Bullish) all fell ≥9%' — rests entirely on [T4] Yahoo/aggregated coverage. This is the load-bearing explanation for the 'market is reading it right' verdict, and no T1-T3 source in the bundle corroborates the day's price action.

Internal inconsistencies

The 'one number' framing is internally muddled: the thesis says the $165M-through-Oct-25 pace is 'well below Q3's $590M' (bearish deterioration), while invalidation_condition_1 says the bear leg fails only if the pace inflects 'above ~$200M/month' — but $165M/25 days already equals ~$198-200M/month. By the opinion's own threshold, the current disclosed pace is at the invalidation line, not below it.

Corrections applied

Restate the run-rate comparison: '~$165M through October 25 (~$200M/month) is roughly flat versus Q3's $590M quarterly pace, but implies no recovery toward the levels embedded in $6.56B 2027 consensus revenue / $2.47 EPS' — drop 'well below Q3' or re-anchor it against consensus expectations instead.
Change 'cut ~25-40% in 90 days' to 'cut roughly 23-28% over 90 days across vendor snapshots', matching the evidence.
Reconcile invalidation_condition_1 with the current data: either raise the falsification threshold (e.g., 'sustained pace above ~$250M/month, i.e. a clear inflection above Q3') or acknowledge the current pace sits at the boundary.
Soften or hedge the intraday-driver claim ('Bitcoin fell 5-6% below $77k') to note it rests on T4 market coverage, or drop the precise % and state the direction with the sourcing caveat.

Sources (16 · T1 = primary filings)

Earlier and later calls on COIN

Nov 3, 2026Partially — and it won't last without BTC.No clear edgeview changed
Oct 28, 2026Noise — the event is tomorrow.No clear edgeview changed

Yes — and it's still not cheap.