Evaluating time gaps in bond reversal patterns

The historical workflow timed reversals on daily Treasury bond futures by classifying reactionary extremes, keeping only the larger classified trends, and counting the days until the next opposite classified trend began.

Editorial reading: a head-and-shoulders, a double top or bottom, or a breakout confirmation is unfinished until that later opposite classified trend starts. The first average cross is scored as the end of the prior trend, not as pattern completion.

How reactionary extremes were labeled

Reactionary lows and highs were labeled with a 20-day moving average. The lowest price while the market was below the average was marked when price crossed above it. The highest price while the market was above the average was marked when price crossed below it. That marked print is the reactionary extreme.

Which moves became classified trends

A move between successive reactionary extremes was treated as an uptrend or a downtrend only when its distance exceeded 4 percent of the starting price level. Smaller swings between reactionary extremes were not treated as classified trends.

How a reversal interval was counted

A reversal was measured only when an uptrend was next followed by a downtrend, or a downtrend was next followed by an uptrend. Successive classified trends in the same direction were ignored.

Days between the end of one classified trend and the start of the opposite classified trend were counted. Reversals separated by more than one day were distinguished from those that flipped immediately. That day count is the reversal interval.

Ten years of daily Treasury bond futures

On 10 years of daily Treasury bond futures closes, 131 upmoves alternated with 131 downmoves as price crossed the average. Twenty-three upmoves met the uptrend test. The next classified trend was a downtrend 11 times and another uptrend 12 times. Eighteen downmoves met the downtrend test, and the next classified trend was an uptrend 11 times.

Immediate flips and delayed starts

Across 22 measured reversals, the opposite classified trend began immediately only 6 times. In 16 cases, or 73 percent of 22, at least a couple of days separated one trend’s end from the next opposite trend’s start.

A 1990 sequence with a long top

In one 1990 Treasury bond sequence, a 6.9 percent uptrend from 27 April to 14 June was followed by six average crosses over 34 days, then an 8.1 percent downtrend beginning 1 August.

Editorial reading: those 34 days are a time-consuming top between opposite classified trends, the setting in which a head-and-shoulders would be evaluated rather than a same-day flip.

A 1998 same-day top treated as an exception

A 29 July to 5 October 1998 advance of 10.3 percent was followed the same day by a decline of 6.7 percent through 6 November. That zero-day top was treated as an exception to the usual interval.

Editorial reading of the three patterns

Editorial reading: head-and-shoulders is a multi-swing reversal structure evaluated as the time-consuming top or bottom between opposite classified trends, not as a same-day flip.

Editorial reading: a double top or bottom is a two-extreme reversal condition. Its completion is checked by whether a later opposite classified trend appears after the first classified trend ends.

Editorial reading: breakout confirmation is treated as confirmation that price has left a prior trend. Here that confirmation is a moving-average cross that later precedes a classified move in the opposite direction.

Editorial reading: the first average cross can therefore mark the end of the prior classified trend and still leave the reversal pattern unfinished until the opposite classified trend begins.