The S&P 500 after 21 oil shocks
An oil shock here is WTI crude up 25% or more within about two months. This page shows what the S&P 500 did 1, 3, 6 and 12 months after each of the 21 completed shocks from 1974 to 2021, with the worst case beside the best and the sample size on every figure.
· study run September 14, 2026 · research, not advice
// The short answer
Twelve months after the 21 completed oil shocks from 1974 to 2021, the S&P 500 was up a median 14.7%, and up in 16 of the 21. The worst twelve months was -37.5% (the 2007 episode); the best was +50.2%. One month after the trigger, the median was 0.0% and 10 of 21 were up.
- Median+14.7%12 months after, n = 21
- Worst-37.5%12 months after, 2007
- Best+50.2%12 months after
- Up16 of 2112 months after
That is the record of 21 cases, and a year out they ranged from -37.5% to +50.2%. It reads history; it does not forecast what follows the next shock.
// The S&P 500 after the trigger
All 21 completed episodes, measured from the month each shock triggered. S&P 500 price index at monthly closes, gross: no dividends and no costs.
| Statistic | 1 month | 3 months | 6 months | 12 months |
|---|---|---|---|---|
| Median | 0.0% | +2.3% | +5.1% | +14.7% |
| Worst | -6.1% | -13.7% | -28.9% | -37.5% |
| Best | +8.5% | +25.0% | +38.8% | +50.2% |
| Mean | +0.1% | +2.2% | +4.0% | +11.6% |
| Up | 10/21 | 12/21 | 13/21 | 16/21 |
Median Worst and best Range of the 21 episodes
The range widens with the horizon. One month out, the worst and best episodes sat 6.1% below and 8.5% above the start; twelve months out, 37.5% below and 50.2% above. The 2007 episode's -37.5% stays in every table it belongs to; no episode was dropped.
// Every episode the study names
The study reports its 21 completed episodes as a group, not one row per episode. These are the episodes it names, in date order, with every interval it reports for each. Where it reports none, the table says so.
| Episode | What the study reports | S&P 500 after |
|---|---|---|
| 1973-74the oil embargo | One of the two episodes whose oil speed reached +100%. | Not reported for this episode alone. |
| 1979 | Named with 2007 for oil up more than 50% on the year, a line the 2026 episode crossed on September 10, 2026. | Not reported for this episode alone. |
| 1990Kuwait | Its oil speed peaked faster than the 2026 episode's +84%, one of two episodes since 1986 to do so. | Not reported for this episode alone. |
| 2003-01 | A second trigger within 15 months of an earlier one (the re-shock group, n = 3). | +32.2% twelve months after the second trigger. |
| 2007 | The worst twelve months of the 21. Also named with 1979 for oil up more than 50% on the year. | -37.5% at twelve months. |
| 2011-11 | A second trigger within 15 months (the re-shock group, n = 3). | +13.6% twelve months after the second trigger. |
| 2016-03 | A second trigger within 15 months (the re-shock group, n = 3). | +14.7% twelve months after the second trigger. |
| 2020the restart | The other episode whose oil speed reached +100%, and the other one since 1986 whose speed peaked faster than the 2026 episode's. | Not reported for this episode alone. |
| 2026in progress, not in the 21 | Triggered March 3, 2026. Oil speed peaked at +84% on April 2, the third fastest since 1986. A second leg reached +29.3% on September 14, an intraday print on a spliced spot and futures series. | +10.4% at 1 month, +14.9% at 3 months and +17.3% at 6 months, as measured on September 14, 2026. 12 months: not yet. |
Re-shocks, n = 3. Three episodes had a second trigger within 15 months of the first: 2003-01, 2011-11 and 2016-03. Twelve months after the second trigger the S&P 500 was up 32.2%, 13.6% and 14.7%. Three cases is a small sample, and the study labels it that way everywhere it appears.
// Faster shocks: the ladder
Each rung counts the episodes whose oil speed reached it, with the S&P 500 measured from the day that rung first broke. The +25% rung is the full sample, so its row matches the table above.
| Rung | Episodes | 12-month median | Up at 12 months | Worst 6 months | Recession within 12 months |
|---|---|---|---|---|---|
| +25% | 21 | +14.7% | 16/21 | -28.9% | 6/21 |
| +35% | 14 | +13.3% | 10/14 | -28.9% | 4/14 |
| +50% | 7 | +12.1% | 5/7 | -20.9% | 3/7 |
| +75% | 3 | +22.6% | 2/3 | -17.9% | 2/3 |
| +100% | 2 | +8.9% | 1/2 | -17.9% | 1/2 |
The +100% rung holds two episodes, 1973-74 and 2020. The 2026 episode crossed every rung through +75% in the spring of 2026. The top two rows rest on 3 and 2 episodes. A recession here means the NBER recession indicator within 12 months of the trigger.
// Sectors: held back
The study also averaged each sector's stocks six months after the trigger. Those averages are held back from this page and the CSV. They are built on QuantGPT's own stock data, whose member lists leave out companies that later delisted (a fix is in progress), so they lean toward the survivors. They also carried no worst case and no count of losing episodes. They will be published once the fix lands and the study is re-run. Every S&P 500 figure on this page uses the index level itself and is not affected.
// Method and data
- Oil price. WTI crude from FRED: the daily spot series DCOILWTICO from 1986, spliced with front-month WTI futures (CL) after the last daily spot print, and the monthly series WTISPLC for 1946 to 1985. Data through September 14, 2026; that last bar is an intraday print.
- Oil speed. The change in WTI over 42 trading sessions, or over two months in the monthly data before 1986.
- Episode. An episode triggers when oil speed reaches +25%. Days that hit the line within 180 days of each other join one episode, and each episode's peak speed is recorded.
- Sample. 21 completed episodes, triggered from 1974 to 2021. The 2026 episode (triggered March 3, 2026) is in progress and sits only in its own row, outside every average.
- Stock returns. The S&P 500 price index at monthly closes, measured from the trigger month; the ladder measures from the day each rung first broke. Gross: no dividends and no costs.
- Recessions. The NBER recession indicator (FRED series USREC) within 12 months of the trigger.
- Rerun. The first run used oil prices that ended September 3, 2026, because the commodities data had not been refreshed. The study was rerun the same day after a refresh: the episodes and the ladder did not change, and the 2026 figures moved.
- Where it ran. Every input query ran over QuantGPT's MCP (the
query_warehousetool) on September 14, 2026.
// Run your own
- Your own dates. QuantGPT's
event_studytool runs this kind of study in one call over MCP: give it event dates and a ticker, and it anchors each event on the last close before the event date, measures forward returns at 21 trading sessions a month, and sets them beside the same horizons over every session in its price history. It measures from daily closes, so its figures will differ a little from this study's monthly closes.
Connect Claude · Connect ChatGPT - Your own holdings. A futuretest runs a book at today's weights through every past episode of one
kind of event, an oil supply shock among them, 63 trading sessions each. That scenario dates its episodes
by event (the 1973 embargo, Iraq's invasion of Kuwait, the Abqaiq attack, chokepoint restrictions), not by
this study's price trigger, so its episodes differ from the 21 here. Futuretests read history; they do not
forecast.
What futuretesting is - The full study. Members can read the study's thread in the QuantGPT forum, every chart
included.
The oil pack in the forum - The data. Every figure on this page, one row each, with its sample size and a source line.
Download the CSV - New to QuantGPT? 5 backtests and 5 AI calls, no card.
Start free
// Questions
How did the stock market react after past oil shocks?
Across the 21 completed oil shocks from 1974 to 2021, the S&P 500 was flat at the median one month after the trigger (0.0%, up in 10 of 21), up a median 2.3% after three months (12 of 21 up), 5.1% after six months (13 of 21 up) and 14.7% after twelve months (16 of 21 up). The range was wide: twelve-month results ran from -37.5% to +50.2%. These are S&P 500 price returns from monthly closes, with no dividends and no costs.
What was the worst case?
The worst twelve months after a trigger was -37.5%, in the 2007 episode. The worst results at the shorter horizons were -6.1% at one month, -13.7% at three months and -28.9% at six months; the study names the episode only for the twelve-month worst. For comparison, the best twelve months was +50.2%.
How many episodes are in the sample?
21 completed episodes, with triggers from 1974 to 2021, found on a WTI crude price series that starts in 1946. An episode starts when WTI rises 25% or more over 42 trading sessions (over two months in the monthly data before 1986), and days that hit the line within 180 days of each other count as one episode. Faster shocks are rarer: 14 episodes reached +35%, 7 reached +50%, 3 reached +75% and 2 reached +100%. The 2026 episode, triggered March 3, 2026, was still in progress when the study ran and is not one of the 21.
Does this forecast what stocks will do after the 2026 oil shock?
No. It is the record of 21 past episodes, and twelve months out they ranged from -37.5% to +50.2%. It reads history; it does not forecast. QuantGPT is a research tool, not investment advice.
Where does the data come from?
WTI crude comes from FRED: the daily spot series DCOILWTICO from 1986, spliced with front-month WTI futures after the last daily spot print, and the monthly series WTISPLC for 1946 to 1985. Stock returns use the S&P 500 price index at monthly closes from QuantGPT's warehouse. Recessions use the NBER indicator (FRED series USREC). The study ran on September 14, 2026 over QuantGPT's MCP, and every figure on this page is in the free CSV.