1992issue C011-5
Ranking futures trend tendency against a random walk
Futures can be ranked by trend tendency by comparing multi-horizon average high-low channel heights with a square-root-of-time random-walk baseline, rather than choosing one predetermined lookback.
- Rank a market's trend tendency by comparing multi-horizon average high-low channel heights with a square-root-of-time random-walk baseline, not with one predetermined lookback.
- Related contracts need not share one trend tendency, so energy markets in the same table still have to be ranked one by one.
- Keep the largest random walk index across a range of lookbacks so the sampling interval is chosen by the data, with a short-term reading on two-to-seven-day windows and a longer-term trend filter from eight days upward.
- Editorial reading: treat that ranking as a trend filter that decides which instruments even deserve a trend-following rule test.
How futures were ranked
Futures can be ranked by trend tendency by comparing multi-horizon average high-low channel heights with a square-root-of-time random-walk baseline, instead of one predetermined lookback. Trend tendency here means a market's measured inclination to expand range beyond a random-walk baseline across several channel lengths.
The channel-height ratio is the average high-to-low span over an n-day window divided by the average one-day span, compared with the square root of n. If prices followed a pure random walk, those ratios would sit on the square root of interval length. The measured ratios exceeded that line in every market in the table.
How the continuous series was built
The ranking used a spliced nearby series from January 1987 through June 1991. Each observation came from the highest-volume nearby contract, and rollover gaps were removed by shifting the new contract.
What the market table showed
Among 28 futures, live cattle and wheat measured as the weakest trenders, while corn ranked sixth. Crude oil and gasoline ranked near the top of that 28-market table, while heating oil ranked nearer the middle. Related energy contracts need not share one trend tendency.
Reading the random walk index
The random walk index is an observed high or low displacement divided by the distance a same-range random walk would be expected to travel over the same lookback. Readings above one are classified as trend rather than drift.
Day-to-day use keeps the largest index across a range of lookbacks so the sampling interval is chosen by the data rather than by a fixed moving-average length.
A short-term reading and a longer-term filter
Counts of which lookback produced the maximum index bent sharply between seven and eight days. That break supports a short-term random walk index on two-to-seven-day windows and a long-term random walk index from eight days upward, used as a trend filter.
On the long-term random walk index, the British pound crossed the 1.0 boundary with greater strength and for longer stretches than wheat, matching the pound's high rank and wheat's last-place rank.
All readings on this track · 10 readings
- 1984When random walks fail to explain market averages
- 1988Building probability-density adaptive weights and a chi-square trend filter
- 1990Withhold a random-walk label until grouping and lookback are checked
- 1991Constructing a variable-lookback random walk trend filter
- 1992Ranking futures trend tendency against a random walk
- 1995A normalized dispersion screen for trend-following
- 1997Why failed random-walk tests still need careful Monte Carlo work
- 2007Constructing a random-walk baseline from ordered returns
- 2011Constructing a cycle-free trend filter against an ATR-scaled random walk
- 2013How to evaluate a hybrid growth and money-flow scan