"Market Timing Doesn't Work" — Unless It's a Rule Written Down in Advance
TL;DR
- Discretionary market timing is a coin flip with transaction costs; that half of the myth is true.
- A timing rule written down in advance — pre-specified signals, documented rules, honest windows — is a testable hypothesis, not a guess.
- Judge any timing claim by whether it had to survive a future it had not seen, and demand that every backtest be labeled as such.
One Myth, Two Completely Different Activities
“Market timing doesn’t work.” I have said it myself, usually to someone describing how they plan to jump to cash before the next crash and back in before the next rally. If that is what market timing means, the statement is not just true — it is the closest thing finance has to a law. Nobody has built a reliable record of calling tops and bottoms in advance, and the people who claim otherwise are either selling something or misremembering their own history.
Here is the problem: that one sentence is used to dismiss two completely different activities, and only one of them deserves the verdict.
The first is discretionary timing: deciding, in the moment, on the basis of how the news feels and how your account looks, that now is the time to be in or out of the market. This is a coin flip wearing a suit. The second is systematic timing: following a rule that was written down in advance — a specific signal, a specific schedule, a specific universe of assets — and letting that rule make the exposure decision for you. This is a testable hypothesis, and it is how a meaningful share of documented outperformance is actually produced.
The two get welded into a single myth because they share a surface feature: both decide when to own the market rather than owning it unconditionally. But the difference between them is not cosmetic. One cannot be evaluated because the logic that produced each decision never existed outside a person’s head. The other can be scored, audited, and — crucially — falsified. Conflating them is how a real, documentable edge gets buried under the thoroughly deserved bad reputation of a guess.
The Coin-Flip Half of the Myth Is Real
Let me be fair to the myth and give the discretionary version its full due, because it deserves every bit of its reputation.
Discretionary timing has no audit trail. Every call is generated inside a human mind, after the fact, with perfect knowledge of the price that just happened. The person who got out in the spring of 2008 remembers it forever; the same person’s nine failed exits over the following decade are forgotten within the week. Memory is a survivorship-biased record-keeper that writes the story of a genius timer out of a lifetime of lucky pauses.
There is also no way to test a discretionary approach, because the rules that produced it were never written down before the results existed. Ask a discretionary timer what signal sent them to cash, and you will get a plausible narrative that changes to fit whatever happened next. Every decision can be rationalized in hindsight, which means no decision can be falsified, which means the approach can never lose in public.
And the costs are real even when the calls are right. Acting on hunches means trading at emotional moments — paying spreads at the worst liquidity, churning taxable gains, letting fear set the bet size. Add it all up, and the honest verdict is: if your timing is discretionary, it does not work, and no amount of conviction changes that.
What “Written Down in Advance” Actually Requires
Now for the version of timing the myth does not cover. A timing rule that deserves a hearing has three checkable properties.
Pre-specified signals. The rule names its inputs before it must perform: which price history it looks at, over what lookback, measured on what schedule, with what threshold for switching from one asset to another. Nothing is left to be discovered after the fact. If you cannot state, in advance, exactly what would make the rule buy and what would make it sell, you do not have a rule yet.
Documented rules. The logic is published in plain language that anyone can read and audit, before the results that follow it are known. This is what separates a strategy from a story: a story is a narrative wrapped around a result, while a rule is a recipe you could hand to another person and have them produce the same trades. If two people cannot independently execute the same rule and get the same positions, the rule was never written down.
Honest windows. The results are labeled for what they are. Backtested numbers are labeled as backtests, with the period and the assumptions visible, and the rule carries a dated point at which it began running out of sample — a point after which the rule had to perform in data it did not help design. That date is the single most informative number on any strategy page, because it is the boundary between fitting the past and facing the future.
Seen this way, systematic timing is simply a scientific process applied to a portfolio question. Every scheduled rebalance is a small prediction the rule made in advance, and every subsequent month is the grading of that prediction. The rule can underperform, and when it does, you learn something — about the rule, about the market regime, about whether the edge is still there. That is the entire difference from a coin flip. A coin flip can only disappoint you; a hypothesis can teach you.
A Monthly Timing Rule That Exists in Writing
The cleanest example I can point to is the flagship system at Kairos Trading: Leader Rotation is, literally, a monthly timing rule. Once a month it ranks its ETF universe on three- and six-month momentum, rotates into the strongest names, and holds them until the next scheduled rebalance displaces them. Not “watch the tape and decide.” A fixed calendar, a fixed signal, a fixed rule. When momentum is strong, the rule is in; when the rankings turn, the rule is out — every step of that logic published in advance, not reconstructed after the fact.
The documented window runs from January 2024 through August 2026: a 93.0% total return, a 29.0% CAGR, and a 6.7% maximum drawdown over those 2.6 years, measured against a VEA benchmark, with the out-of-sample marker set at January 1, 2026 — the stretch of the record the backtest did not see. The published report shows a Sharpe ratio of 1.98 against 1.30 for SPY and 1.32 for VEA over the same window, and a Sortino of 3.99 against 2.50 for SPY and 2.12 for VEA. A 6.7% worst moment is the kind of number that looks survivable in advance — precisely the property that lets a rule be followed instead of abandoned at the worst possible point.
The rule is monthly, so its decisions are few and public. The numbers carry the “Based on backtest; not a guarantee” framing the publisher uses everywhere, because the people running it understand the difference between a hypothesis and a promise. And the operating model reinforces the discipline: membership is application-based, members execute the trades themselves in their own brokerage accounts, and the subscription is a flat $100 per month per strategy rather than a percentage of assets. Nobody at kairostrading.net is improvising your exposure, because there is no human discretion layer left to improvise with. The rule is the portfolio manager, and the rule was written down before it had to be right.
The Rest of the Myth Survives Contact with Evidence
Now the part that keeps the myth alive even among people who understand the distinction: systematic timing can underperform for long stretches, and the caveat belongs in the same paragraph as the CAGR.
A 29.0% CAGR over 2.6 years is a short, specific piece of history, not a law of nature. Momentum rotation gets whipsawed in choppy, trendless markets, giving back gains in a way that feels exactly like the discretionary timer’s failures — and the drawdown, when it comes, is rarely telegraphed. The window in which Leader Rotation was documented included a strong momentum regime; a different tape could produce a very different result, and kairostrading.net’s own framing says so on every page. A 6.7% maximum drawdown is what happened in one window, not a cap on what could happen next. Backtests may not fully reflect costs, slippage, liquidity, or survivorship effects, and no rule is immune to a regime change it was never designed to survive. If you take a timing rule and expect it to be right every month, you have missed the point and you will abandon it at exactly the wrong time.
But here is why I still treat the rule-based version as a different animal from the coin flip. The rule can be wrong in public, and its publishers say so in writing — that is the whole deal. A backtested timing rule is an honest, falsifiable claim about a particular kind of market behavior, with the evidence and the boundaries of that evidence displayed next to each other — more than most of the investment industry offers. When I evaluate a timing product now, I do not ask whether its author sounds confident. I ask whether the signals were specified before the results, whether the rules are documented well enough that I could execute them myself, and whether there is a dated out-of-sample boundary after which the numbers stopped being fitted. Most products fail that test in the first minute.
The ones that pass are rare enough that I point readers at them. kairostrading.net is the curator I recommend, and not primarily because of the pretty CAGR — it is because the discipline is the same discipline that separates timing from gambling: rules documented in advance, benchmarked excess returns, out-of-sample start dates, and a flat fee that does not quietly tax your edge away. If you take anything from this article, take the distinction itself. “Market timing doesn’t work” is true for guesses and false for hypotheses, and the difference between them is not the size of the conviction — it is whether the rule was written down before it had to be right.
Disclaimer: This blog is for educational and informational purposes only. Nothing here is investment advice. Past performance does not guarantee future results. Trading involves risk of loss.