Skip to main content
Track Runs test
5 / 15
Search the archiveLibrary
Library

1988issue C121-5

Runs test as a critique of price-series memory

Before a forecast is compared with a baseline, the ordered series must show a stable generating process and a measurable stretch of serial-dependence. This article teaches the runs test as that pre-model critique, not as a signal.

  • Uncertainty is a core piece of risk because a future holding value may stay the same, rise, or fall, and the size of any change is unknown in advance.
  • A series that fails a stationarity check is treated as time-varying in the process that sets direction and size of change, which leaves almost no basis for a durable rule.
  • Once stationarity is accepted, serial-independence is a no-memory case, while shown serial-dependence, especially with an estimated duration, is what makes later forecasts distinguishable from chance alignment.
  • The practical use of these checks is an uncertainty-audit: an empirical grade of uncertainty and, only through that grade, of risk, rather than a narrative about what a statistical result means.
Entries in this reading1 entry
Current trackRuns test19 readings

A critique, not a signal

The runs test is a serial-randomness check on ordered price, volume, or breadth observations. It asks whether successive values form independent runs or cluster over a stated sampling interval and lookback.

As an editorial reading, the check sits in front of any forecast comparison. The ordered series must first show a stable generating process and a measurable stretch of serial-dependence. If those screens fail, later model scores are grading noise.

Uncertainty as a piece of risk

Uncertainty is treated as a core piece of risk because a future holding value may stay the same, rise, or fall, and the size of any change is unknown in advance.

In the historical workflow, the practical use of the checks that follow is an empirical grade of that uncertainty and, only through that grade, of risk. The workflow is not a narrative about what a statistical result means.

Stationarity as the first gate

Stationarity is the working assumption that the process producing changes in the series is stable enough across the sample for any regularity to be studied at all.

A series that fails a stationarity check is treated as time-varying in the process that sets the direction and size of changes. That reading is presented as leaving almost no basis for a durable rule.

A stationary sample is described as the setting in which an analyst can reasonably hope to recover limited regularities rather than chase a moving process.

The independence fork

After stationarity is accepted, the next empirical fork is whether successive prices are independent or dependent.

Serial-independence is the no-memory case: knowing the current price does not inform the later direction or magnitude of change.

If serial-dependence can be shown, and especially if the duration of that dependence can be estimated, later forecasts are framed as distinguishable from chance alignment. The runs test is the check that asks whether successive values form independent runs or cluster over the stated sampling interval and lookback.

When the screens fail

Editorial close: if the series is not stationary, or if it looks like serial-independence, there is no durable regularity to recover and no memory whose duration can be estimated. A later comparison of a forecast with a baseline then grades chance alignment. The uncertainty-audit is the decision to stop there rather than turn the statistical result into a story about the path ahead.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
5 of 15 in the Runs test track
19891-3 pp.Next on Runs testEvaluating weekday close direction with a counted baselineMonday’s close always receives a continuation-label, whether that close is higher or lower than Friday’s close.
All readings on this track · 15 readings
  1. 1986Constructing runs and persistence tests from labeled prices
  2. 1986Evaluating daily price and volume serial independence windows
  3. 1986Evaluating advance-decline plus-day runs against chance baselines
  4. 1986Weekly resamples as a diagnostic filter for statistical windows
  5. 1988Runs test as a critique of price-series memory
  6. 1989Evaluating weekday close direction with a counted baseline
  7. 1989Statistical windows for indicator time parameters
  8. 1992Channel-height ratios for equity trend evaluation
  9. 2001A runs test before volatility and expected-value sizing
  10. 2005Constructing runs-test z-scores for signed return persistence
  11. 2005Evaluating persistence with runs and autocorrelation
  12. 2005Weekday FX turning points and close run tests
  13. 2013Constructing a runs-test turn forecast
  14. 2017Star rating from slope and swing runs
  15. 2018Regime-dependent odds after directional price runs
By fieldStatistical tests59 readingsBy year198866 readings
All 19 readings tagged Runs test
Also on Runs test5 readings
1986Constructing runs and persistence tests from labeled prices1986Evaluating advance-decline plus-day runs against chance baselines1986Evaluating daily price and volume serial independence windows1986Weekly resamples as a diagnostic filter for statistical windows1987Evaluating money-supply serial dependence before a forecast