This study measures how much five of the largest US companies, Apple, Microsoft, Nvidia, Alphabet and Tesla, have moved from one daily close to the next, and sets them beside two index funds: SPY, which tracks the S&P 500, and QQQ, which tracks the Nasdaq-100. For each it reports annualized volatility over one, three and five years, beta against SPY, the deepest fall from a high, the worst single day and how often a day's move reached 3%.
The figures are recalculated from stored daily closing prices rather than frozen in a snapshot, and every window ends at SPY's latest close, the date printed under each table. When that history is not available, the tables say so rather than show estimates, and the sections below explain each measure with made-up numbers so they read the same either way. For more on owning shares, start at the stocks hub.
What to look for in the tables
Each column answers a different question about the same prices. These rules of thumb, with invented numbers, show how to read them.
- Volatility. A higher percentage means wider swings in both directions. Dividing by about 15.87 gives a typical day: 24% a year is about 1.5% a day.
- Beta. A beta of 1 means the stock moved one for one with SPY on average; 1.4 means 1.4% for each 1% move in SPY, and below 1 means less.
- Maximum drawdown. The deepest fall from a previous high over five years. A figure of −35% means $10,000 at the peak was worth $6,500 at the low.
- Worst day. The largest one-day fall in five years, with its date, which helps tie the fall to the news behind it.
- Days moving 3% or more. A share of 5% means about one trading day in 20, roughly one a month, moved that much up or down.
- Volatility by year. Reading down a column shows whether a stock's swings came in one rough year or were spread across several.
How is each measure calculated?
Every measure starts from one closing price per trading day, turned into daily returns and summarized over windows that end at SPY's latest close.
| Measure | How it is calculated | Window |
|---|---|---|
| Daily return | The day's close divided by the previous trading day's close, minus 1 | Each trading day |
| Volatility | Standard deviation of daily log returns, the natural log of each close divided by the one before, multiplied by √252 and shown in percent; needs more than 20 closes | 1, 3 and 5 years |
| Beta vs SPY | Covariance of the symbol's daily returns with SPY's, divided by the variance of SPY's daily returns, using days both traded; needs at least 60 such days; SPY is 1 by definition | 3 years |
| Maximum drawdown | The largest percentage fall from the highest close so far to any later close | 5 years |
| Worst day | The lowest daily return, dated by the close that ended it | 5 years |
| Share of days moving 3% or more | Daily returns of 3% or more up or down, divided by all daily returns | Past year |
| Volatility by calendar year | The volatility measure for each calendar year in the stored history, with the current year running to the latest close; needs more than 120 closes | Each year |
The closes come from Twelve Data, a market-data provider, and are stored by the site. Twelve Data's documentation says its daily prices are adjusted for stock splits by default; dividends are not added back, so every figure is a price-only return. GOOG is Alphabet's Class C share; its Class A share, GOOGL, trades separately and is not included. A measure appears only when a symbol's history starts no more than a week after its window begins, so a history that is too short shows n/a rather than a figure for a shorter period. The stored history reaches back a little over five years, enough for the longest window. The glossary entry on standard deviation covers the statistic itself.
Log returns barely differ from percentage changes on ordinary days: a 2% rise is a log return of 1.98%. They differ more on big days, where a 10% fall is a log return of −10.54%, and they add up from one day to the next, which suits a measure of spread.
How much do the biggest stocks swing in a typical year?
Annualized volatility answers that by scaling the typical daily move up to a yearly figure, shown here over one, three and five years.
| Asset | Symbol | Volatility, 1 year | Volatility, 3 years | Volatility, 5 years | Beta vs SPY, 3 years | Maximum drawdown, 5 years | Worst day, 5 years | Date | Share of days moving 3% or more, past year |
|---|---|---|---|---|---|---|---|---|---|
| Apple | AAPL | 24.2% | 26.7% | 27.9% | 1.04 | -33.4% | -9.2% | Apr 3, 2025 | 5.6% |
| Microsoft | MSFT | 32.5% | 25.9% | 28.0% | 0.98 | -37.6% | -10.0% | Jan 29, 2026 | 10.0% |
| Nvidia | NVDA | 37.3% | 46.5% | 51.5% | 2.05 | -66.4% | -17.0% | Jan 27, 2025 | 18.3% |
| Alphabet (Google) | GOOG | 33.3% | 30.6% | 32.4% | 1.13 | -44.6% | -12.3% | Oct 29, 2026 | 11.2% |
| Tesla | TSLA | 45.3% | 57.4% | 59.5% | 2.25 | -73.6% | -15.4% | Mar 10, 2025 | 29.5% |
| SPDR S&P 500 ETF | SPY | 12.8% | 15.2% | 17.2% | 1.00 | -25.4% | -5.9% | Apr 4, 2025 | 0.0% |
| Invesco QQQ (Nasdaq 100) | QQQ | 20.3% | 20.5% | 23.1% | 1.27 | -35.6% | -6.2% | Apr 4, 2025 | 3.2% |
The main table holds every measure for all seven symbols; this section reads its three volatility columns, and the sections after it take the rest in turn. Rises count as much as falls, so a stock that climbed steeply can score as high as one that slumped. The windows answer different questions. The one-year figure reflects the past twelve months alone, so one rough quarter can lift it, while the five-year figure blends calm and stormy stretches. A one-year figure well above the five-year one means the past year was rougher than the stock's longer record; the reverse means it has been calmer lately.
Worked example: From a yearly figure back to a typical day
Annualized volatility is the daily figure multiplied by √252, about 15.87, so dividing by 15.87 undoes it. The numbers are invented.
- A stock with 32% annualized volatility: 32% ÷ 15.87 ≈ 2.0% on a typical day, about $200 on a $10,000 holding.
- A fund with 16%: 16% ÷ 15.87 ≈ 1.0%, about $100 on the same amount.
If daily returns followed a bell curve, about two days in three would stay within one typical move, and about one in 20 would go more than twice as far: beyond 4.0% for the stock and 2.0% for the fund. Real daily returns have more extreme days than a bell curve allows, which is why the table also counts large moves directly.
Why 252 is used for stocks and 365 for crypto is set out in the bitcoin vs S&P 500 study, which applies both.
What does beta against SPY show?
Beta shows how strongly a stock has moved with the market: the average move it made for each 1% move in SPY over the past three years.
A beta of 1 means one for one, above 1 means more than the market and below 1 less. SPY's own beta is 1 by definition, and QQQ is measured against SPY in the same way as the stocks. Beta is not a measure of total swings: FINRA notes that a stock can have high volatility but a low beta if its moves do not track the market's. The covariance formula in the method table can be rewritten to show why.
Beta = correlation with SPY × (the stock's volatility ÷ SPY's volatility)
Worked example: Two stocks with the same beta
Suppose SPY's volatility is 15%. The other figures are invented too.
- Stock A: volatility 45%, correlation with SPY 0.5. Beta = 0.5 × (45 ÷ 15) = 1.5.
- Stock B: volatility 25%, correlation 0.9. Beta = 0.9 × (25 ÷ 15) = 1.5.
On days SPY fell 2%, both would have fallen about 3% on average. Stock A's actual days scatter much more widely around that average, because most of its movement comes from its own news, and its typical daily swing is 45 ÷ 25 = 1.8 times as large as B's.
This study's five companies add a twist: they are a large part of SPY itself. On October 29, 2026, State Street listed Nvidia at 7.77% of the fund, Apple at 7.65%, Microsoft at 5.92%, Alphabet's Class A and Class C shares at 2.50% and 2.02%, and Tesla at 1.71%, together 27.57% and all among its 10 largest holdings. Part of each stock's beta is therefore measured against its own price, which pulls its correlation with SPY up. What is the S&P 500 explains how those weights are set, and the SPY page shows the fund's latest price. A three-year beta is also an average over about 750 trading days, so another window, or a change in the business, can give a different figure. The beta glossary entry has the general definition.
How far have these stocks fallen from a high?
The maximum drawdown column gives the deepest fall from a previous high in the past five years, measured on closing prices.
The calculation starts at the window's first close and tracks the highest close so far; each later close is compared with that running high, and the widest gap is the drawdown. The figure does not say whether the price has since recovered, and a fall that began before the five-year window opened is counted only from the window's first close. The drawdown glossary entry covers the general term.
Worked example: A drawdown measured from the running high
An invented stock opens the window at $100, climbs to $160 and then falls to $96.
- Drawdown at the low: $96 ÷ $160 − 1 = −40.0%, even though the price is only 4% below where the window began.
- Gain needed to get back to $160: $160 ÷ $96 − 1 = 66.7%.
- If it then recovers to $150, its current drawdown is $150 ÷ $160 − 1 = −6.25%, but the maximum drawdown stays at −40.0% unless a deeper fall replaces it.
For the S&P 500's own falls since 1946, and how long its recoveries took, see bear markets and corrections.
What was each stock's worst day?
The worst-day column shows the largest one-day fall in the past five years, from one close to the next, dated by the second close.
A close-to-close change folds in everything between the two closes, including news released after the bell and trading before the next open, so a worst day dated on a Monday covers the whole weekend. For a single company the largest falls can follow its own announcements, such as quarterly results; for SPY and QQQ, a worst day reflects selling across many stocks at once.
Worked example: Reading a one-day fall
An invented stock closes at $250 on a Thursday. The company reports weak results after the close, and on Friday the stock closes at $220.
- Worst day: $220 ÷ $250 − 1 = −12.0%, dated Friday.
- On a $10,000 holding: $8,800 at Friday's close, $1,200 less than a day earlier.
- Had the stock opened Friday at $205 and climbed back to $220 by the close, the table would still show −12.0%; the 18% low during the day never appears.
For scale, the market-wide circuit breakers that pause all US stock trading are set on the S&P 500, not on single stocks: a 7% (Level 1) or 13% (Level 2) fall in the S&P 500 from the prior day's close halts trading for 15 minutes if it happens before 3:25 p.m. ET, and a 20% fall (Level 3) ends trading for the rest of the day. A single stock can fall far more than 7% in a day without setting them off.
How often do they move 3% or more in a day?
The last column gives the share of trading days in the past year on which the price rose or fell by 3% or more from the previous close.
It turns volatility into something easier to picture: how often a holder saw a large move. Rises and falls count alike, and a move of exactly 3% counts. Two stocks with the same volatility can score differently, because one may have many medium-sized days and the other a few very large ones.
Worked example: Counting large-move days
An invented stock has 250 daily returns in the past year. It rose 3% or more on 14 of those days and fell 3% or more on 16.
- Share of days moving 3% or more: (14 + 16) ÷ 250 = 12.0%.
- That is about one trading day in eight, or two to three a month, since a month has about 21 trading days.
Has volatility changed from year to year?
The yearly table repeats the volatility calculation for each calendar year, so a single rough year stands out instead of being averaged into five.
| Year end | AAPL | MSFT | NVDA | GOOG | TSLA | SPY | QQQ |
|---|---|---|---|---|---|---|---|
| Dec 31, 2022 | 35.6% | 35.4% | 63.2% | 38.8% | 67.1% | 24.3% | 32.2% |
| Dec 31, 2023 | 19.9% | 25.0% | 46.8% | 30.7% | 52.4% | 13.2% | 17.8% |
| Dec 31, 2024 | 22.3% | 20.0% | 52.1% | 27.7% | 62.6% | 12.6% | 18.0% |
| Dec 31, 2025 | 32.2% | 24.0% | 49.8% | 31.8% | 63.2% | 19.3% | 23.4% |
| Oct 30, 2026 | 25.8% | 34.5% | 37.5% | 33.9% | 45.4% | 13.0% | 20.8% |
- SPY
- QQQ
- AAPL
- NVDA
- TSLA
Each row uses one calendar year's closes, and the last row runs only to the latest close, so it covers part of a year and shifts as more days arrive. A year needs more than 120 closes, about six months of trading, to get a figure. Reading across a row compares the seven symbols in the same market conditions; reading down a column shows how one symbol changed. The chart draws five of the seven columns to keep the lines readable, and the table has all seven.
What can't these measures tell you about risk?
They describe how prices moved, not why, and they leave out several risks that matter to a holder.
- Hindsight in the list. The five companies are among today's largest, so by construction their shares rose enough to get there. Companies that shrank or failed are absent, a form of survivorship bias.
- Closing prices only. Moves during the day and between the close and the next open are folded into one close-to-close change, so intraday lows never appear.
- No dividends. Price-only returns make drawdowns look slightly deeper than they would with dividends reinvested.
- Split adjustments. Twelve Data adjusts daily prices for splits when they are downloaded, but the site's daily update rewrites only the latest closes. A split after the history was loaded would show up as a false one-day fall until the full history is reloaded. Nvidia's 10-for-1 split, for example, began trading on a split-adjusted basis on June 10, 2024, so a worst day dated on a split day deserves a check; stock splits explained covers how splits work.
- Short, overlapping windows. One, three and five years are brief next to a market's history, and the shorter windows sit inside the longer ones, so the columns are not independent readings.
- Rises and falls together. Volatility treats a 3% rise like a 3% fall; drawdown and the worst day look only at falls, which is why the table shows both kinds.
- What prices cannot show. A company's debts, its competition, how much of a portfolio sits in one stock and when the money is needed are risks no price statistic captures.
What does this mean for someone who owns these stocks or funds?
It shows how widely these prices have swung and how far they have fallen, not what they will do next: past performance never predicts future results.
Read together, the columns separate two things an owner feels. Volatility and the count of 3% days describe the ride from day to day; the drawdown and the worst day describe how deep the worst stretches went. A fund holding hundreds of stocks spreads one company's bad news across the rest, which is why SPY and QQQ sit beside the single stocks. Index funds explained covers how that spreading works, and what is the Nasdaq-100 explains the index behind QQQ.
Owning SPY already means owning these companies: at State Street's October 29, 2026 weights, $10,000 in SPY held about $2,757 of them, and holding the shares directly as well adds to that overlap. Asset allocation by age looks at how stocks fit beside bonds and cash, and the Nvidia page shows one of the five companies' latest prices.
How to cite or reuse the volatility data
The data is free to reuse under CC BY 4.0, with credit and a link to this page.
- Get the CSV file, one row per symbol as in the main table. It is rebuilt from the latest stored closes, and while the history is still loading it returns a short unavailable message instead.
- Quote each figure with its measure, window and as-of date, such as a three-year beta against SPY to the date printed under the tables, because the numbers change as new closes arrive.
- Mention that the figures are price-only returns from daily closes, without dividends or intraday moves.
- Suggested credit: Source: Sonic Wallet analysis of Twelve Data daily closing prices, data through the as-of date shown with the tables.
How the site sources, checks and corrects its studies is set out in the editorial policy.