Curriculum·S203 Chart Literacy and Evidence·about 31 min
The three states
By the end of this lesson you can
- →Classify a market into trending, ranging or volatile using a rule applied without hindsight
- →Explain why state classification is only reliable in retrospect, and what follows from that
- →Describe how a method that suits one state fails in another, and name the failure
- →Report the out-of-sample evidence on rules selected from a large universe
Sophomore · enrolled learners
This lesson opens with Data-snooping and technical trading rules, tested out of sample.
- What happened
- Sullivan, Timmermann and White's 1999 paper in the Journal of Finance took the 26 trading rules used in an earlier well-known study and expanded them into a far larger universe of rules, then applied them to about a century of daily Dow Jones Industrial Average data. Rather than reporting the performance of the best rule, they used White's Reality Check bootstrap to quantify and adjust for data-snooping bias across the whole universe from which the best rule had been drawn. They then examined a subsequent out-of-sample period of roughly ten years and found low profitability, which they interpreted as consistent with increased market efficiency.
- The decision point
- The best rule out of a large search always looks good, because searching guarantees it. The correction is to ask how good the best of that many rules would look on data with no structure in it at all, and to require the observed performance to exceed that. When rules selected in one regime were carried into the next one, the performance was not there, which is the same result a trader gets by optimizing a method on the state a market was in and deploying it into the state it moves to.
What you will be able to answer
- →Name the three states and the method each one punishes.
- →Why is state classification unreliable in real time?
- →What does the Reality Check correction do?
- →What is the practical form of the out-of-sample result?
Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.
It is free. We do not sell the list and there is nothing to buy at the end of it.
Sources and review
- https://onlinelibrary.wiley.com/doi/abs/10.1111/0022-1082.00163
- https://www.kevinsheppard.com/files/teaching/mfe/advanced-econometrics/Sullivan_Timmermann_White.pdf
- https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1467-6419.2007.00519.x
- https://www.coindesk.com/research/market-spotlight-the-19-billion-liquidation-that-shook-crypto
Confidence medium·Volatility low·Reviewed 2026-08-05·Owner unassigned
Contested
The Sullivan, Timmermann and White result is from equity index data over a century ending in the 1990s, and its interpretation as evidence of increasing market efficiency is one reading among several. What transfers reliably is the methodological point about pricing the size of a search, which is not specific to any market.
The three-state taxonomy is a teaching device rather than a claim about market physics. Other frameworks divide the space differently and are not wrong. What matters is that every method embeds an assumption about which state it is in, which is true under any taxonomy.