Damage Math

Damage Math

The market erased about $4.33 billion of Lyft's equity value from the November 2025 peak, a 42% cut. Over the same window consensus forward EPS rose roughly 14% and forward revenue held flat, so the fundamental numerator behind the move is close to nil. A transparent two-scenario DCF sizes plausible value destruction at $0.3 billion if the hit is temporary and about $4.33 billion if permanent; the trial weights those at a probability the impairment is temporary of 0.62, leaving a residual gap near $2.5 billion that is priced multiple, not lost cash.

The near-term hit — the numerator

The trigger was the February 10, 2026 fourth-quarter print. Reported Q4 2025 revenue was $1,593 million against consensus of $1,755 million — a 9.2% shortfall [1]. The entire shortfall is one line: the full-year filing states revenue was reduced by a "$168 million impact from certain legal, tax, and regulatory reserve changes and settlements, which were recorded as a reduction to revenue" [2]. Add that $168 million back and Q4 revenue is $1,761 million — essentially the consensus number. On the same print, normalized EPS beat, $0.366 against $0.316, a 16% upside surprise.

Gross bookings tell the same story from the volume side: Q4 2025 bookings were $5,074 million, up 19% year over year and accelerating [3]. Nothing about demand broke; a one-time reserve was recorded against the revenue line.

The clearest test of whether earning power was impaired is what the analysts who follow the name did to their forward numbers. They raised them. Six months before today — a snapshot dated January 25, 2026, before the trough — consensus FY2027 normalized EPS was $1.83; today it is $2.09, up 13.9%. FY2028 EPS moved $1.86 to $2.12, up 13.6%. Forward revenue barely moved: FY2027 slipped 0.7% and FY2028 slipped 1.3%.

Source: consensus estimate revisions, S&P Capital IQ estimate momentum (data/sp/estimates.json), 180-day snapshot vs current; EPS in $, revenue in $M.

The near-term hit that drove a 42% de-rating, measured in the currency the whole tab rests on — dollars of forward cash flow — is not a cut. The forward EPS stream is higher now than before the sell-off began. The only genuinely soft print since was Q1 2026, where normalized EPS came in $0.285 against $0.292 consensus, a 2.3% miss, while revenue beat by 1.1% — a rounding-scale variance, not a demand event.

The price and EV change — the denominator

Against a numerator of roughly nil, the price moved a great deal. The shares fell from $24.57 on November 12, 2025 to a $12.65 trough on March 30, 2026 — a 48.5% peak-to-trough drawdown over 138 days — and sit at $14.20 today [fit_features:capitulation_gauge]. On 417.659 million shares that is a market-cap move from $10.26 billion at the peak to $5.93 billion now, an erasure of $4.33 billion, or 42% [fit_features:market_cap]. With net cash of $79 million the enterprise value is fractionally lower than the market cap at every point, so the EV move is the same $4.33 billion [fit_features:balance_sheet_class].

Source: consensus revisions (data/sp/estimates.json); price and market-cap moves derived from fit_features.json (capitulation_gauge, market_cap), price as reported. The 1.35x volume spike on the drawdown marks capitulation, not a structural repricing [fit_features:capitulation_gauge].

The move is inverted in signature: the estimates that price is supposed to discount rose while the price fell by nearly half. That does not by itself prove a mispricing — the market may be discounting an impairment the analysts have not yet cut for — but it does establish that the fall was not a response to a fall in near-term earnings. It is a change in the multiple applied to those earnings.

The NPV arithmetic — two scenarios, workings visible

A de-rating is only a mispricing if the value of the forward cash flows exceeds the price by more than the plausible damage. Here is a deliberately simple discounted-cash-flow on consensus mean free cash flow, so every number is on the page. Discount rate 10%; terminal growth 3% applied to the final explicit year; net cash added back at the end.

Source: consensus mean free cash flow FY2026–FY2029 (data/sp/estimates.json, mirrored in fit_features.consensus_forward_yield); present values computed at r = 10%.

The four explicit years present-value to $3.91 billion. The terminal value at the end of FY2029 is $1,247.9M × 1.03 / (0.10 − 0.03) = $18.36 billion, worth $12.54 billion today. Enterprise value is $3.91B + $12.54B = $16.45 billion; adding $79 million of net cash gives equity of about $16.53 billion, or roughly $39.6 per share against the current $14.20. Hold the growth assumption and lift the discount rate to 12% and the enterprise value still computes to about $12.8 billion — more than twice the current $5.93 billion market cap. On consensus figures the post-crash price already embeds a wide margin of safety.

That intrinsic estimate is the temporary reading: consensus cash flows, intact and rising. The damage question is how much of that value the near-term hit plausibly destroys under each diagnosis.

Source: price damage = market-cap decline peak-to-current, derived from fit_features.json; NPV damage from the two-scenario DCF above; trial weights from ruchir/trial/tally.json.

The gap is the whole point of the tab. If the impairment is temporary, fundamental damage explains about $0.3 billion of a $4.33 billion price move — roughly 93% of the fall is multiple, not cash. If it is permanent, the price move is warranted and there is no gap. Weighting the two scenarios by the trial's probability the impairment is temporary of 0.62 gives plausible NPV damage of 0.62 × $0.3B + 0.38 × $4.33B = $1.83 billion, and a residual gap of $4.33B − $1.83B = $2.50 billion — about 58% of the price move that neither scenario's fundamental damage accounts for. That residual is the price of autonomous-vehicle displacement fear, expressed as a lower multiple rather than a lower forecast.

The trial — temporary versus permanent, at strength

The temporary-versus-permanent question was argued by two opposing corpus-cited briefs and ruled on by three independent judges. Both cases are real; the strongest evidence on each side follows.

The panel ruled the probability the impairment is temporary at 0.62. The three seats landed at 0.62, 0.43 and 0.64 — a range of 0.43 to 0.64 and a mean of 0.56, with a small 0.085 gap between the temporary-first and permanent-first reading orders; the tally is not flagged contested [tally:p_temporary]. One permanent-first judge sat below even odds, so the reading is a lean, not a consensus. That 0.62 is the probability the report carries; this tab's arithmetic is weighted by it and does not override it. What would move it, per the ruling: gross bookings decelerating below ~10% for two consecutive quarters, insurance cost-per-mile continuing to compress revenue margin while the EBITDA-margin path stalls, or documented U.S. rideshare share loss to Uber or AV entrants.

Which line broke — and whether it self-corrects

The driver behind the hit is the revenue margin — Lyft's take rate on gross bookings. In Q4 2025 revenue was 31.4% of gross bookings ($1,593M / $5,074M), down from 36.2% in Q4 2024 ($1,550M / $4,279M); stripping the $168 million reserve lifts the Q4 2025 figure to 34.7%, still below the prior year [13]. For the full year revenue was 34.1% of the $18,507 million of gross bookings [14].

Source: quarterly revenue and gross bookings, Q4 FY2025 earnings presentation p.10 [15]; take rate = revenue / gross bookings, derived.

The compression has a repricing mechanism working in its favor. The squeeze came from insurance cost per mile, and Lyft expects total insurance costs to "continue to increase at a lower rate than they have historically as a result of California's rideshare insurance reform bill, SB 371," which lowered cost per mile in its largest state [16]. If insurance reprices and incentive intensity normalizes, the take rate recovers and the temporary reading holds.

The structural counter is that insurance is non-discretionary — cost of revenue "primarily includes insurance costs" — so each incremental ride carries a regulated cost that scales with volume, and the Q1 2026 doubling of incentives is not a one-quarter accounting event [17] [18]. Set against Uber's scale in the same market, that is the case for a permanent haircut rather than a passing squeeze. The trial weighed both and landed on the lean above; the arithmetic on this page — a residual gap near $2.5 billion at that weighting — is what the diagnosis has to be right about for the price to be right.

The forward cash yield that this de-rating created is treated in the Yield tab; the anatomy of the drawdown itself in Dislocation.