From 29 January 1993 to 26 October 2026, SPY (the S&P 500 exchange-traded fund, or ETF) closed higher on 54.05% of trading days. It finished higher over 81.83% of 1-year stretches, with dividends counted. Each share is a base rate: how often an outcome happened across all comparable cases.
A forecaster's hit rate is the share of calls that proved right. A base rate sets its bar. Calling a rise for every 1-year stretch would have scored 81.83%. A lower hit rate does worse than always saying up.
What is a base rate in investing? A base rate is the share of all comparable past cases in which an outcome happened. In SPY since 1993, the base rate of a positive return rose from 54.05% over one day to 81.83% over one year.
The share positive was 65.73% over a month. It was 92.03% over 10 years, on only 2 or 3 separate windows. It was 100% over 15 and 20 years, on one or two.
A base rate also measures loss. SPY lost more than 10% in 11.33% of 1-year windows. The worst 1-year window fell 47.36%.
The 1-year base rate depended on the start date: 58.65% for windows starting in 2000 to 2009, against 93.48% for 2010 to 2019.
What is a base rate in investing?
A base rate is the share of all comparable past cases in which an outcome happened. Here the outcome is a positive return. A case is one holding period, such as one day or one year.
The base rate changes with the length of the holding period. It also changes with the sample.
How did we measure the base rates?
We used SPY's daily closing prices from 29 January 1993 to 26 October 2026, adjusted for dividends and splits. That is 8,493 prices over 33.7 years. Every return in this post counts dividends.
A window is a start day and an end day a fixed number of trading days later. We used every start day that fits, so the windows overlap.
We tested ten horizons, with trading days in brackets. They are 1 day (1), 1 week (5), 1 month (21) and 3 months (63). The rest are 1 year (252), 3 years (756), 5 years (1,260), 10 years (2,520), 15 years (3,780) and 20 years (5,040).
A window counts as positive when its end price is above its start price. The share of positive windows is the base rate. The median is the middle return once we sort the windows by return.
How often does SPY rise over each horizon?
SPY rose on 4,590 of 8,492 trading days, which is 54.05%. The share climbed at every step up to 10 years. It was 58.67% for 1 week, 65.73% for 1 month, 72.02% for 3 months, 81.83% for 1 year and 92.03% for 10 years. Chart 1 shows all ten horizons.
Source: YX Insights price data
The median return was 0.07% over a day and 14.45% over a year. The table gives the median, the worst window and the share of windows that lost more than 10% at each horizon.
Horizon | Windows | Share positive | Median return | Worst return | Share that lost more than 10% |
|---|---|---|---|---|---|
1 day | 8,492 | 54.05% | 0.07% | −10.94% | 0.01% |
1 week | 8,488 | 58.67% | 0.37% | −19.79% | 0.33% |
1 month | 8,472 | 65.73% | 1.44% | −32.73% | 1.86% |
3 months | 8,430 | 72.02% | 3.73% | −41.48% | 5.47% |
1 year | 8,241 | 81.83% | 14.45% | −47.36% | 11.33% |
3 years | 7,737 | 83.64% | 41.22% | −43.48% | 13.16% |
5 years | 7,233 | 84.85% | 72.06% | −35.54% | 5.68% |
10 years | 5,973 | 92.03% | 131.93% | −34.86% | 4.15% |
15 years | 4,713 | 100.00% | 234.56% | +66.05% | 0.00% |
20 years | 3,453 | 100.00% | 441.92% | +131.68% | 0.00% |
SPY with dividends counted, every window from 29 January 1993 to 26 October 2026. The windows overlap. Source: YX Insights price data
All 4,713 windows of 15 years and all 3,453 windows of 20 years were positive.
How large were the losses?
The worst return at each horizon is the lowest return across every window of that length in the file, from 29 January 1993 to 26 October 2026. Chart 2 shows it. A bar to the left is a loss.
Source: YX Insights price data
The worst 1-day return was a fall of 10.94%, from the close on 13 March 2020 to the close on 16 March 2020. The worst month fell 32.73%, from 21 February to 23 March 2020. The worst year fell 47.36%, from 5 March 2008 to 5 March 2009.
The worst return stopped deepening after one year. It was a loss of 43.48% over 3 years, 35.54% over 5 years and 34.86% over 10 years. The worst 10-year window ran from 23 February 1999 to 2 March 2009.
At 15 years the worst window gained 66.05%, from 14 August 2000 to 25 August 2015. At 20 years it gained 131.68%.
A 1-year window lost more than 10% in 934 of 8,241 cases (11.33%). It lost more than 20% in 406 (4.93%). A 3-year window lost more than 10% in 13.16% of cases, which is more often than a 1-year window.
Do overlapping windows overstate the evidence?
Yes, for the longer horizons. The 8,241 windows of one year come from 33.7 years of prices. A window shares 251 of its 252 daily returns with the window that starts the next day.
We repeated the 1-year test with windows that do not overlap. We tried each of the 252 possible first start days. Each run held 32 or 33 windows. The share positive ran from 71.9% to 87.9%, so the full-sample 81.83% sits inside that range.
Calendar years give a plainer count. From the last close of 1993 to the last close of 2025, SPY rose in 26 of 32 calendar years (81.25%). It fell in 2000, 2001, 2002, 2008, 2018 and 2022.
At 10 years, 5,973 windows exist, but only 2 or 3 fit without overlapping, so the 92.03% is also thin. Only 1 or 2 non-overlapping 15-year windows fit. Only 1 of 20 years fits. The 100% shares at 15 and 20 years therefore describe one or two stretches of history.
The long windows also start early. The last 20-year window began on 11 October 2006. The last 15-year window began on 12 October 2011.
Does the base rate depend on when the window starts?
Yes. We grouped every 1-year window by the year it started. The share positive was 97.37% for starts in 1993 to 1999, 58.65% for 2000 to 2009, 93.48% for 2010 to 2019 and 83.09% for 2020 to 2025.
Source: YX Insights price data
The 2000 to 2009 group holds the worst 1-year window, the 47.36% fall that ended in March 2009. Each group covers 7, 10, 10 and 6 calendar years of start dates. The windows within a group overlap. The last start in the 2020 to 2025 group is 23 October 2025.
The full-sample figure of 81.83% sits between the groups. A base rate taken from a different 10 years of SPY would have given a different answer.
How does a base rate help judge a signal?
A signal that says "up" every day scores 54.05% on daily direction. The same signal scores 81.83% on 1-year direction. The two figures describe the same fund, but they set different bars.
A base rate does not say whether the next case will rise. In 1,497 of the 8,241 1-year windows (18.17%), SPY ended lower.
Our post on time in the market shows how a few days drive long-run returns. SPY is one of the names covered in our Multi-model Signals. This post is not one of its results.
What does this test leave out?
The test covers one fund over one window. The file begins on 29 January 1993, so earlier decades are absent. Another index or start date would give different shares.
All returns are nominal, which means before inflation, trading costs and tax. A year here means 252 trading days, which can differ from a calendar year by a few days.
We ran no significance tests. Our guide to stock market recovery time covers how long falls took to repair.
For the general problem of judging a pattern found in past prices, see what is backtesting and how to read a backtest.
In SPY since 1993, the positive share rose from 54.05% over one day to 81.83% over one year. The long-horizon shares rest on few separate stretches of history.
Get the next explainer
The Academy is free to read, here on the web. The Long Trend, our free weekly letter, comes by email.
Learn more with YX Insights
This explainer is part of the YX Insights Academy. Each one takes a single idea and checks it against real data.
The same approach runs through everything else we publish:
Systematic Portfolio: ready-made portfolios for Macro & Megacaps and for Commodities. We publish the holdings and every change.
Multi-model Signals: a daily long-or-flat call on every name we cover. Each call shows how strongly our four models agree.
Research: company deep dives, macro commentary and essays on how we test.
Good places to start on the website:
Academy: more explainers like this one.
Start here: get our free weekly letter by email.
Macro & Megacaps Systematic Portfolio and Commodities: the two portfolios in detail.
Track record: live results since 3 August 2026, plus the ten-year backtest.
Upgrade: get the Systematic Portfolio, Multi-model Signals and research in full.
Common questions about base rates
What does base rate mean in stocks?
A base rate is the share of all comparable past cases in which an outcome happened. In SPY, the S&P 500 exchange-traded fund, with dividends counted, 54.05% of trading days and 81.83% of 1-year windows from 1993 to 2026 had a positive return. Each share is a base rate for its holding period.
How often does the stock market go up in a year?
In SPY, the S&P 500 exchange-traded fund, the return was positive in 26 of 32 calendar years from 1994 to 2025 (81.25%), with dividends counted. Across every 1-year window from 1993 to 2026, 81.83% were positive. The windows overlap, so they hold only about 33 separate years of history.
How often does the S&P 500 lose more than 20% in a year?
In SPY, the S&P 500 exchange-traded fund, 406 of 8,241 1-year windows from 1993 to 2026 lost more than 20% (4.93%), with dividends counted. The worst 1-year window fell 47.36%, from 5 March 2008 to 5 March 2009. Overlapping windows count the same fall many times.
What is the base rate fallacy?
The base rate fallacy is judging a case from its specific details while leaving out how often the outcome happens across all cases. SPY rose over 81.83% of 1-year windows since 1993, so a 1-year "up" call is right that often without any forecasting. A hit rate means little until it sits beside that figure.
DISCLAIMER: This article is intended for educational purposes only. Any information or analysis in this note does not constitute an offer to sell or a solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice, nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.
