Does weather affect the stock market? We tested sunny mornings in 10 cities

What the study found

Does the weather affect the stock market? In 2003 the economists David Hirshleifer and Tyler Shumway said it does, in “Good Day Sunshine: Stock Returns and the Weather”, published in the Journal of Finance. At 26 stock exchanges from 1982 to 1997, an index tended to do better on days when its city was sunny than when it was overcast. They measured cloud in oktas, eighths of the sky, from 0 for a clear sky to 8 for a fully overcast one. Each okta more cloud took about 1.1 basis points off the day's return (a basis point is 0.01%), so a clear day beat an overcast one by about 9 basis points. Cloud went with lower returns in 17 of the 26 cities.

Their explanation was mood: sunshine puts people in a better frame of mind, and people in a good mood judge the future more kindly, shares included. They also tried a trade: each morning, buy the indices of the cities with a sunnier sky than usual and sell those under more cloud than usual. It paid before costs, but because it trades every day, they wrote, fairly modest costs would eliminate the gains.

How we tested it

We paired ten stock indices with their exchange cities and took every trading day from 2004, years after the study's data ended: the S&P 500 with New York, the FTSE 100 with London, the DAX with Frankfurt, the CAC 40 with Paris, the AEX with Amsterdam, the IBEX 35 with Madrid, the SMI with Zurich, the Nikkei 225 with Tokyo, the Hang Seng with Hong Kong and the ASX 200 with Sydney. The weather came from the hourly reports of each city's weather station, mostly at the airport, kept by the US National Oceanic and Atmospheric Administration (NOAA). Those records run to the end of 2025, so that is where the test ends.

As the study did, we averaged the cloud over the early morning, 5 to 8 am local time, and over the working day, 6 am to 4 pm, and compared each day with that city's normal for the same week of the year. So a sunny morning means less cloud than usual for the season: a bright winter morning in London counts as sunny even if Madrid's sky was clearer still.

Before any price was looked at, the weather records had to agree with what everyone knows. Madrid's sky was clearer than London's (1.79 oktas of cloud on an average day against 3.53), winter was cloudier than summer in London, Frankfurt, Paris and Amsterdam, and the warmest hour of the day fell in the afternoon in every city. All of it held. The rules for a pass were written down before the run as well.

To measure luck, we kept each date's market returns but gave it the weather of the same date in another year, shuffling the years 2,000 times, the same way for every city. That keeps each city's seasons and the way weather moves across a continent, but cuts any real link between a morning's sky and that day's market.

What we found

The study's own trade failed. On the 5,512 days with at least one sunny and one cloudy city, the indices under a sunnier morning than usual did 0.91 basis points a day worse than those under more cloud, not better. They were worse in both halves: by 0.55 from 2004 to 2014 and by 1.28 from 2015 to 2025. And 80.9% of the shuffled calendars fitted the belief at least as well as the real weather. To beat 1 in 20 of them, the sunny side would have needed a lead of +1.18 basis points a day.

City by city

Over the whole working day, more cloud did go with slightly lower returns in 7 of the 10 cities, as the study found. But the effect was small. Pooled across the cities, each okta more cloud took 0.24 basis points off the day, about a fifth of the study's 1.1, and its standard error, 0.25, is as large as the effect itself: it cannot be told apart from zero. In Hong Kong, more cloud went with better days, by 1.79 basis points an okta.

Split simply into mornings sunnier and cloudier than usual, the cities divide evenly. In five of them, New York, Frankfurt, Paris, Madrid and Zurich, the sunny mornings were followed by the better days. In the other five the cloudy mornings were, and in London, Tokyo, Hong Kong and Sydney by a wide margin.

One difference from the study: airport reports leave some high cloud out. A report of a clear sky can mean only that there is no cloud below 5,000 feet, or below 12,000 feet at automated US stations, while the study's observers counted the whole sky. The limit is the same in every year and city, but it makes our measure of cloud a little coarser than theirs.

Could you trade it?

No. A trader only knows the morning's weather by the time the market opens, so the fair version of the study's trade buys the sunny cities' indices and sells the cloudy ones at the open and closes both at the close. We tested it on the days with a real opening price in our data: 5,446 days with a sunny and a cloudy city. Before any costs it lost 2.37% a year. After a broker's spread of 1.5 basis points on each side, it lost 9.33% a year, and it lost money after costs in both halves.

Before costs the trade made a little money only in the first half, +0.48% a year from 2004 to 2014, and lost 5.04% a year after that. With no profit before costs, there is no break-even cost to give: the weather trade did not pay even when trading was free.

Small mood effects are hard to see in daily returns, which swing far more from news than from the sky. The study saw one in 1982 to 1997. In the years since, the sunny mornings brought no better days on average, and the trade built on them lost money.

Questions people ask