Does market timing work? What the data says

"Time in the market beats timing the market" is one of the most repeated lines in investing. It is mostly right, but not for the reason usually given, and not without exceptions. This guide looks at the evidence: what missing a few months costs, why forecasting the market is so hard, an honest test of our own that mostly failed, and the difference between guessing and following a fixed, systematic process.

The cost of being out at the wrong time

The standard argument is that a handful of very good months produce a large part of long-term returns, and an investor who jumps in and out is likely to miss them. The data supports that:

$10,000 invested in the S&P 500 in 1993 and held through every month grew to about $302,000 (total return, to mid-2026). Missing just the 10 best months out of 401 cut that to about $103,000, a third of the result.

But the same chart shows the other side, which is rarely quoted: missing the 10 worst months would have turned the same $10,000 into about $1.05 million. Missing both the 10 best and the 10 worst left about $360,000, more than staying invested.

The reason is that the best and worst months cluster together. Many of the best months came right after crashes (in 2009, in 2020), inside the same volatile periods as the worst ones. So the real question is not "can you avoid the bad days?" but "can you avoid the bad periods without also missing the rebounds?"

Why forecasting the market fails

Most market timing is forecasting: deciding today whether the market will rise or fall next month or next year. The evidence against that is strong:

• Short-term returns are close to random. Economic data, valuations and sentiment explain little of the next month's return.
• Professional forecasters have a poor record of predicting recessions and bear markets before they start.
• Behaviour works against the timer. Investors sell after losses and buy after gains, the opposite of what timing needs. Checking a portfolio often makes this worse: Benartzi and Thaler (1995) called it "myopic loss aversion", where frequent evaluation makes losses feel bigger and leads investors to hold less risk than they planned.

Our own test: a timing rule that mostly failed

In White Paper No. 1, When Bonds Stop Hedging, we tested a rule that switched the bond part of a 60/40 portfolio into cash whenever stocks and bonds had been moving together, over 70 years of data. We pre-registered the test and then tried to prove the result was luck:

• The rule improved the Sharpe ratio only slightly: 0.54 against 0.51 for a plain 60/40 (backtested, hypothetical).
• Against 2,000 random switching rules with the same number of switches, it beat about 86–88 %. That is suggestive, but not statistically convincing.
• In 13 % of bootstrap resamples it did not beat 60/40 at all, and most of the gain came from 2022–2026.
• A version that added inflation as a second signal, pre-registered in advance, added nothing.

We also tested six real-time signals for predicting whether bonds would protect stocks over the next two years. The best one beat a naive guess by only 5–6 percentage points; rising inflation and the bond-price trend had no predictive value. Our conclusion was "probably small, not proven", and we published the result anyway.

Timing, regimes and systematic processes are different things

"Market timing" is used for three very different activities:

A systematic process does not try to predict. It reacts to what has already happened, with rules fixed in advance and tested on historical data. That is why it can step aside in long declines like 2000–02 or 2008, but will usually miss the first part of a V-shaped rebound like 2020.

For the S&P 500, the InvestWealth systematic strategy in the Research Terminal had a worst fall of about −30 % since 1993 against −55 % for buy & hold, with a Sharpe ratio of about 0.73 against 0.51 (backtested, after costs, hypothetical). The rules are not published, but every signal has been recorded and shown in the Terminal since 16 October 2026, with a three-month delay for free accounts.

So, does market timing work?

• Forecasting-based timing: for almost everyone, no. Missing a few good months costs more than the occasional good call earns.
• Reacting to the regime with a fixed process: it can work in the sense that matters most to many investors, which is limiting the depth of the worst falls (see max drawdown and the Sharpe ratio). It does not reliably beat buy & hold every year, and it needs discipline through the periods when it lags.
• Staying invested remains the right default for anyone who cannot stick to a fixed process through good and bad years.

Frequently asked questions