Evaluate a mechanical futures system as one procedure

A single testable procedure

The evaluated system is framed as a single mechanical trading system. It joins a prior rule set to a volume-weighted average price construction and uses two derived indicators, slope and acceleration of a two-step linear-regression smoother, to control entries and exits.

Bars and the volume-weighted series

Decision bars are the 30-minute chart, while the volume-weighted series is taken from 15-minute bars. The smoother replaces the ordinary average inside that volume-weighted construction.

Slope, acceleration, and gated exits

Acceleration is treated as the rate of change of slope, approximated by the difference of consecutive slope values. A sign change in acceleration is used as a reversal cue for slope and as a required positive condition on a named transition entry.

A named run exit is defined exclusively from slope and acceleration and is taken when slope reaches a peak and acceleration changes sign. That pairing is presented as able to block earlier channel-style exits.

When slope amplitude is small, entries and exits are prohibited so that channel-like stretches are not treated as actionable. This is offered as a way to reduce whipsaws relative to fixed-percentage or fixed-dollar profit targets.

Partitions from curve pairs

Curve pairs are treated as partitions. Each above-or-below relationship doubles the state space, so two pairs yield four partitions and three pairs yield eight, and the system uses several such relationships for multiple entries.

Fast remaining above slow is treated as a stronger rising-state cue than color or position versus the raw series. Unlike a simple average, sitting below the underlying series does not by itself imply that series is rising.

Independence and the midnight constraint

The fast and slow volume-weighted smoothers share much of the ordinary fast and slow behavior but show a midnight reinitialization kink. Entries that use either volume-weighted smoother are not allowed between midnight and 1 am.

Independence between ordinary and volume-weighted smoothers of the same length is the evaluation reason for combining them. Matched lookbacks still produce dissimilar paths that add entry and exit information.

How robustness is argued

Robustness testing is argued from complementary pairing of two futures books and from the observation that color changes on the smoothers precede crossovers, while acceleration sign changes occur even earlier than those color changes.