Options Meerkat

Case study

Calls bought into a crash: which expiry, which strike, and how bad was buying early?

NVIDIA (NVDA) fell 22.5% into its close of April 4, 2025, then rallied. Anyone who bought calls on the way down had two choices to make: how far out to buy, and how far from the price. Here is one path through that question in Options Meerkat: sixteen calls bought on four days around the low, from ten sessions before it to three after, what each did, what buying too early felt like, and whether the same held for Tesla and AMD.

The data: each option's daily closing prices from March to October 2025, and the chains at the close of Monday, October 26, 2026, for today's step. The low is picked with hindsight, as the lowest close between March 1 and May 15, 2025: the study asks what buying early or late against it did, not how to find it. The answers the dashboard plays with each step were recorded on October 27, 2026; the dashboard around them is live. Entry studies are part of Pro.

Seven steps is a lot of asking. One question runs most of this path at once, and the steps show what it skips.

Recorded answers from October 27, 2026, on option closes from 2025, played on the live dashboard.

  1. Step 1 of 7 · The price of a call

    See what options cost on the way down

    “NVDA fell into early April 2025. Around its low that spring, how expensive were its options: the at-the-money IV of each expiry from a week to a year out, against the stock's realized volatility, 10 sessions before the low and on it?”

    SessionNVDA~1 week~1 month~6 monthsRealized
    Mar 21 −10$117.7038.1%42.1%48.5%68.0%
    Mar 28 −5$109.6751.4%46.8%50.2%61.8%
    Apr 4 low$94.3188.3%70.4%56.8%60.1%
    Apr 9 +3$114.3364.5%52.2%49.4%84.9%

    Ten sessions before the low, calls were cheap against the stock: every expiry's implied volatility sat well under the 68% NVDA had been moving over the 21 sessions before. Then the short end rose fastest. On the low, a one-week option was priced at 88% against 60% realized, an inverted curve, while six months out was 57%. Three sessions later, after a 21% rise in a week, the front had fallen back to 64%. Buying at the bottom meant paying the most for the shortest options. (Implied against realized volatility, the term structure)

    Next A call's price is only half the question. Buy sixteen of them, four expiries by four strikes, on each of the four days, and see what each was worth later.

  2. Step 2 of 7 · Sixteen choices, four entries

    Rank the expiries and strikes

    “Study NVDA calls bought around its low between March 1 and May 15, 2025: 10 and 5 sessions before the low, on it, and 3 after, at 14, 30, 90 and 180 days to expiry and 30, 50, 70 and 85 delta, held 5, 21 and 63 sessions. Which expiry and strike did best, and which did worst?”

    Each call is bought at the day's close and held 63 sessions, about three months, or to expiry if that comes first. The median of the four entries' returns on the premium, and how many of the four made money:

    Expiry30 delta50 delta70 delta85 delta
    14 days−100% 0/4−89% 1/4−43% 1/4−16% 2/4
    30 days−100% 1/4−95% 1/4−20% 1/40% 2/4
    90 days+390% 4/4+290% 4/4+198% 4/4+131% 4/4
    180 days+309% 4/4+244% 4/4+182% 4/4+125% 4/4

    The expiry decided almost everything. The 14- and 30-day calls needed the rebound before they expired; bought ahead of the low, they ran out first, and the cheap 30-delta ones mostly went to zero. Every 90- and 180-day call made money from every entry. The strike came second: further out of the money paid more when it worked, deeper in the money lost less when it didn't. Held only 21 sessions, though, every 90- and 180-day choice lost money in the median: only the call bought on the low itself was reliably ahead a month later. The longer expiry bought time to be wrong about the timing, and needed it. (The Greeks, backtesting options)

    Where the recorded answer differs Meerkat's answer calls the entry three sessions after the low the weakest of the four, and buying late into the bounce worse than buying early into the fall. For the 90-day 30-delta call it was the other way round: bought on April 9 it made +508%, bought ten sessions early +28% (step 3).

    Next A median of four entries hides the order they came in. Take the best choice, the 90-day 30-delta call, and follow each entry day by day.

  3. Step 3 of 7 · Buying too early

    Follow the early entries through the drop

    “If I had bought NVDA's 90-day 30-delta and 50-delta calls 10 sessions before its low between March 1 and May 15, 2025, how bad was holding them for 63 sessions: the deepest loss, when it came, and how many sessions they spent at a loss?”

    BoughtCallPaidDeepestUnder water, in a rowAt the end
    Mar 21 −10Jun 20 138$4.58−85%33+28%
    Mar 28 −5Jun 20 128$4.25−70%17+273%
    Apr 4 lowJun 20 113$4.40−20%2+601%
    Apr 9 +3Jul 18 135$4.94−74%22+508%

    Bought ten sessions early, the call lost half its value in a week and 85% by April 21, eleven sessions after the low, not on it. It then spent 33 sessions in a row at a loss, a month and a half of a position worth a fraction of what was paid, and finished 28% up only because it was held to its June expiry. Bought three sessions after the low, it fared little better on the way: NVDA had jumped, the second leg down to April 21 took 74% of it, and it was under water for 22 sessions before it paid. All four deepest losses fell on the same day. Only the call bought on the very close of the low, which nobody knew was the low, had an easy ride.

    The 180-day 50-delta call, bought ten sessions early, was the gentler version of the same trade: −66% at its worst, also 33 sessions in a row under water, and +78% after 63 sessions. Further out and nearer the money gave up some of the gain to lose less of the premium.

    Next Calls bought on the low paid the most implied volatility of any day. Split their profit into what the stock did, what time took, and what the IV's fall took.

  4. Step 4 of 7 · What the IV cost

    Split the profit into its causes

    “For NVDA calls bought on its low between March 1 and May 15, 2025, at 14, 30, 90 and 180 days to expiry and 50 delta, held 21 sessions: how much of the profit was the stock's move, how much time, and how much the fall in implied volatility?”

    One 50-delta call bought at the close of April 4, in dollars, the profit split by Black-Scholes at the entry's own implied volatility: the stock's move first, then the time passed, then the rest, which is the change in IV.

    ExpiryIV paidStock's moveTimeIV's fallProfit
    14 days expired81.1%+$438−$339—+$98
    30 days expired71.0%+$1,405−$231—+$1,173
    90 days63.9%+$1,211−$270−$86+$854
    180 days56.9%+$1,139−$177−$107+$854

    The 90-day call's IV fell from 63.9% to 56.6% over the month, and cost it $86, a tenth of what the stock's rise made. Paying a panic's IV was a small cost against a 21% rebound in a week. It would not have been against a slow recovery: the IV's fall comes either way, the rise does not. The two short calls expired inside the month, so the IV's part of theirs is in their time (an expired option has no IV to change), and the 14-day one's time took three-quarters of what the stock gave it. (IV crush)

    Next One stock's sell-off is one story. Run the same study on two others that fell in the same weeks.

  5. Step 5 of 7 · Not just NVDA

    Run it on Tesla and AMD

    “Run the same study on TSLA and AMD: calls bought 10 and 5 sessions before their low between March 1 and May 15, 2025, on it, and 3 after, at 30, 90 and 180 days to expiry and 30 and 50 delta, held 21 and 63 sessions. Does the long expiry still win?”

    Both bottomed at the close of April 8, 2025: TSLA at $221.86, 23.0% below its high, AMD at $78.21, 31.9% below. The median return held 63 sessions, and how many of the four entries made money:

    ChoiceTSLAAMD
    30 days, 30 delta−52% 1/4−100% 1/4
    30 days, 50 delta+24% 2/4−46% 2/4
    90 days, 30 delta−92% 1/4+549% 3/4
    90 days, 50 delta+37% 3/4+370% 4/4
    180 days, 30 delta−8% 2/4+439% 4/4
    180 days, 50 delta+23% 4/4+277% 4/4

    The 30-day calls lost again. Past that, the answer changed with the stock. AMD rebounded hard and the cheap 90-day 30-delta call led, though one of its four entries went to zero. On TSLA the same call lost 92% in the median: the two bought before the low, struck at 360 and 335, were up 81% and 194% at their best, then expired worthless on June 20 with TSLA at $322. An out-of-the-money call has to get there by its date. The one choice that made money from every entry on both stocks was the 180-day 50-delta call, the least exciting of the six. Every 90- and 180-day choice still went at least 67% under water at its worst.

    Where the recorded answer differs Meerkat's answer ranks by median and finds 90 days ahead of 180 in three of the four columns. That is right, and it is the median; counted by how many entries lost, 180 days was the steadier choice.

    Next The past sell-off is settled. See what the same calls cost on NVDA today, before the next one.

  6. Step 6 of 7 · From the study to today

    Price today's calls the same way

    “What do NVDA's calls cost today: the at-the-money IV from a week to six months out, against the stock's realized volatility and its IV percentile?”

    NVDA
    $216.50 the close of Oct 26, 2026
    30-day IV
    41.4% 60th percentile of its year
    Realized
    29.6% 20 sessions
    The curve
    33–43% a week to six months
    Reports
    Nov 17 inside the Nov 20 expiry

    Today is the opposite of March 21, 2025. Then, options were priced well under what the stock was moving; now they are priced above it, with the Nov 20 expiry lifted by the report inside it. That is what a calm market and a report coming look like, not a sell-off. Step 1's table is the one to read when one comes: whether the short end is running away from the long end, and how far IV sits above what the stock is actually doing.

    Next The study said a lot about one spring. Ask what it cannot say about the next.

  7. Step 7 of 7 · The edge of the data

    Name what one sell-off cannot tell you

    “What can't a study of NVDA calls bought around its April 2025 low tell me about buying calls in the next sell-off?”

    • One rebound. April 2025 ended in one of the fastest recoveries on record. In a sell-off that keeps going, every call here loses, and the long ones lose the most dollars.
    • Hindsight. The low is the lowest close, found afterwards. The days around it are what you could have bought on, not what you would have.
    • The fills. Prices are each contract's daily close, 6% of them modeled where a contract did not trade; in a panic the spread you pay is wider than a close shows.
    • Four entries. A median of four is a description, not an expected return.
    • Your situation. Whether you could hold a position down 85% for a month and a half, and how much of your account it was.

  8. What it adds up to

    From a hunch to a narrower question

    You started with “Buy calls when it crashes.”

    You ended with “In that one sell-off, expiry mattered more than strike, the early buyers sat through an 85% loss for a month and a half, and the IV was a small cost against a fast rebound.”

    The checks, to use again

    1. Read the term structure against realized volatility. On the low, a week's IV was 88% against 60% moved.
    2. Choose the expiry before the strike. The 14- and 30-day calls mostly lost; every 90- and 180-day call paid over 63 sessions.
    3. Price being early in depth and in time. Ten sessions early: −85% at the worst, 33 sessions in a row at a loss.
    4. Split the profit. The IV's fall cost the 90-day call $86 against $1,211 from the stock.
    5. Run it on another stock, and count the losers. On TSLA the best NVDA choice lost 92% in the median; the 180-day 50-delta call made money every time.

    Judge a crash trade by how it did when it was early, not by how it did from the bottom. What to do with that is a decision the data does not make for you.

  9. One question · The path at once

    Or ask for most of it at once

    “In a crash like April 2025, which NVDA calls held up best: compare 30, 90 and 180 days to expiry at 30 and 50 delta, bought 10 sessions before the low through 3 after, held 21 and 63 sessions, with how deep and how long they were under water.”

    One entry study runs steps 2 and 3 together: the ranking, and the depth and length of the losses on the way. Meerkat finds the low from the dates it is given, buys on each day around it, and ranks the choices by their median for each holding period, naming the entry each one's deepest loss came from.

    Asked this way, Meerkat bought on three days, ten sessions before the low, on it, and three after, not step 2's four, and its medians move with that: the 90-day 30-delta call's 63-session median is +508% here, +390% there. Held 21 sessions, every long call was still behind in the median; the 180-day 50-delta was nearest even, at −6%, and the shallowest at its worst, −66%.

    What it leaves out is step 1's term structure against realized volatility, step 5's second and third stock, and step 7: that one sell-off is one draw.

  10. Your turn

    Ask your own questions

    Entry studies run on any symbol of the S&P 500 or the Nasdaq-100, over any dates since the option history begins, with Pro. Click one to put it in Meerkat's chat, in the dashboard, then press Enter. Asking needs a sign-in, and entry studies Pro.

    Sign in and ask Open the dashboard

    More case studies: the wheel, backtested · an earnings report · a high IV percentile · dealer gamma · deep put blocks

    Click a question to put it in the dashboard's chat. Nothing on this page is investment advice; past results do not promise future ones, and options can lose their whole value quickly.