Skip to main content
Track Dow Theory
15 / 29
Search the archiveLibrary
Library

2001issue C101-3

Why trend, range, and Dow rules need separate tests

Trend-following stays with prevailing direction, mean-reversion fades a defined trading range, and Dow Theory reads slower industrial versus railroad structure. A bull-and-bear reward that reduces to the range-normalized close-open does not measure location in the range or index confirmation, so each procedure still needs its own entry, exit, and abstention rules.

  • Trend-following is a single testable procedure that follows and exploits the prevailing direction of prices over the system holding period.
  • Mean-reversion looks for reversals only while the market is judged to be in a trading range, and is not a rename of trend-following.
  • Dow Theory is a slower OHLC reading of industrial and railroad averages used to judge whether a directional or ranging hypothesis is even in force.
  • The range-normalized close-open is bounded by 1 and -1 and is silent about where the change sits inside the bar, so it cannot replace a market-state gate.
Entries in this reading3 entries
Current trackDow Theory29 readingsMean reversion43 readingsTrend following89 readings

Not one signal under three names

A trend-following system is a procedure that follows and exploits the prevailing direction of prices. An antitrend, or range-reversal, method is a procedure that seeks reversals while a market is in a trading range. Long-horizon Dow industrial and railroad averages, with a cited start date of 1897, are treated as a distinct historical chart record for studying index structure.

Editorial view: those three procedures are not interchangeable names for one signal. Trend-following stays with prevailing direction. Mean-reversion fades only a defined trading range. Dow Theory reads slower industrial versus railroad structure. Each still needs its own entry, exit, and abstention rules, and a market-state gate before signals are combined.

Three procedures that do not replace one another

In the terms used here, trend-following is a single testable procedure that follows and exploits the prevailing direction of prices over the system holding period. The archive definition of an antitrend, or range-reversal, method is a procedure that seeks reversals while a market is in a trading range. That procedure is mean-reversion: a single testable procedure that looks for reversals only while the market is judged to be in a trading range.

Neither procedure stands in for the other. One is allowed only while prices have a prevailing direction to follow. The other is allowed only while the market is judged to be in a trading range.

Dow Theory is a slower OHLC structure reading of industrial and railroad averages used to judge whether a directional or ranging hypothesis is even in force. Editorial reading: that confirmation step is not a trend-following entry and not a mean-reversion fade. It is a prior question about whether the directional or ranging hypothesis is available to test.

A slower index record

The industrial and railroad averages, with a cited start date of 1897, are treated as a distinct historical chart record for studying index structure. Printed long-term index plates and separately obtained historical data that a researcher plots are presented as alternative ways to inspect that same record.

Editorial note: inspecting that record is not the same procedure as trend-following or mean-reversion. It does not supply their entry, exit, or abstention rules.

What the collapsed reward does not measure

A three-part bull-and-bear reward difference reduces, after summation, to close minus open divided by the bar high-low range, for both up-close and down-close bars. That reduced form is the range-normalized close-open: close minus open divided by that bar’s high-low range, an oscillator bounded by 1 and -1 that is silent about where the change sits inside the bar.

On a bar with high 50 and low 48, that reduced reading is the same for open 48 and close 48.1 as for open 49.9 and close 50.0. The first pair sits near the low of the bar. The second pair sits near the high. The collapsed reading does not distinguish them.

The multi-part reward form is defended only as a way to drop or reweight components and compare those variants in backtests across markets, time frames, securities, and conditions. That comparison is a formula-variant backtest: a comparison of which reward components are kept, dropped, or reweighted before any rule set is chosen for a market, time frame, or condition.

Editorial view: a formula-variant backtest still does not measure location in the range, and it does not measure industrial versus railroad confirmation. Those omissions are why the collapsed reward cannot serve as a shared test for trend-following, mean-reversion, and Dow Theory.

Editorial reading: keep the tests apart

Editorial conclusion: a reward that collapses to the range-normalized close-open is a classroom exhibit, not a shared rule set. Location in the range is not in that reading. Index confirmation is not in that reading.

Trend-following therefore stays with prevailing direction over the system holding period. Mean-reversion fades only while the market is judged to be in a trading range. Dow Theory remains a slower industrial versus railroad structure test of whether a directional or ranging hypothesis is even in force. Each method still needs its own entry, exit, and abstention rules.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
15 of 29 in the Dow Theory track
20011-3 pp.Next on Dow TheoryBear-market confirmation via prior correction troughsBear-market confirmation is a major decline that falls below the trough of the previous major bull-market correction, not a percentage label on its own.
All readings on this track · 29 readings
  1. 1982A bounded-risk entry separates a forecast from a trend
  2. 1984Critiquing reward bias, single-scale charts, and exact-turn forecasts
  3. 1990Constructing dual-average primary-trend confirmation
  4. 1991Two-average confirmation before a primary reversal call
  5. 1991Delayed confirmation is not Dow Theory divergence
  6. 1991Confirmation delay and breadth divergence in a two-average case
  7. 1992Utilities as a rate-regime lead for equities
  8. 1992Critiquing unconfirmed Dow rallies with volume
  9. 1993When Dow Theory signals fail after the decision-makers change
  10. 1994Market life expectancy as a risk filter
  11. 1994Score industrial and transport sync before calling an intermediate-trend signal
  12. 1997A 1995 industrial-average breakout mapped from component trends
  13. 1998Confirm Dow trends with Market breadth and Head and shoulders
  14. 1999Confirming an equity idea with rate, commodity, and index spreads
  15. 2001Why trend, range, and Dow rules need separate tests
  16. 2001Bear-market confirmation via prior correction troughs
  17. 2002Constructing a Dow line before breakout confirmation
  18. 2002Two-average confirmation as a swing-by-swing classroom drill
  19. 2002Withhold the hypothesis until the second average confirms: a 2001 case
  20. 2002A primary-bear case study in cycle speed, Dow theory, and pattern legs
  21. 2003Four index proxies as a bear-regime dashboard
  22. 2004Dual-average confirmation at shared prior highs
  23. 2004Confirmation as the second clock on a trend break
  24. 2004The confirmation-reaction planning window after a joint break
  25. 2005Dow confirmation as a two-average trend test
  26. 2008Related-average confirmation lag after a correction
  27. 2008Intermediate confirmation outranks secular phasing
  28. 2012Align swings to nested energy regimes
  29. 2016From nonconfirmation to a bearish primary trend change
By fieldTrend and mean reversion93 readingsBy year200176 readings
All 29 readings tagged Dow Theory
Also on Dow Theory5 readings
1982A bounded-risk entry separates a forecast from a trend1984Critiquing reward bias, single-scale charts, and exact-turn forecasts1997A 1995 industrial-average breakout mapped from component trends1993Switch trend following and mean reversion with an equity-curve filterMean reversion1994Evaluating weekly trend-following and mean-reversion timing rulesMean reversion