A signed hourly swing catalog as a next-session chi-square check

Evaluation as a catalog of signs

Editorial framing. This piece treats evaluation as a closed catalog of signed end-of-session hourly moves. Each bar is reduced to plus or minus, every short lookback key is enumerated, and the next-session direction count is what gets inspected.

A signed hourly swing is an hourly price change reduced to plus or minus, with volume and the size of the move omitted. Keeping only the sign of each hourly price swing turns the last hours of a session into a finite catalog of short sequences.

A directional pattern is an ordered short sequence of those signed hourly swings, treated as one discrete catalog key.

The four-hour catalog and the three-hour catalog

The lookback window is the fixed number of trailing hourly intervals encoded into one pattern key. The sampling interval is the hourly bar used to assign each plus or minus in the sequence.

Using the last four hourly signs produces 16 possible keys, which is the four-hour catalog, the complete set of signed sequences that can be formed from the last four hourly intervals. Using the last three hourly signs produces eight possible keys, which is the three-hour catalog.

Counts on the following session

Each catalog key was evaluated by counting next-session advances and declines in a 19-year archive of hourly changes. A next-session direction count is that tally of advances versus declines on the session after a given catalog key.

A chi-square test is a statistical check of whether an observed next-session advance and decline split departs from a stated chance baseline. Editorial reading uses an even-split baseline, a simple null that treats the next session as equally likely to advance or decline, so a pattern table is read as a test of departure from chance.

An illustrated four-hour key

One illustrated four-hour key of advance, advance, decline, then advance was observed 142 times and was followed by 90 next-session advances and 52 next-session declines.

A chi-square test was applied to that four-hour cell’s next-session counts and was reported as occurring by chance fewer than once in a thousand repetitions.

A four-decline key and a four-advance key

The same evaluation contrasted a four-decline sequence with a four-advance sequence and reported that the two cells did not share the same next-session balance.

How both catalogs were presented

Two figures presented the four-hour catalog and the three-hour catalog together with the associated next-session counts.

Editorial reading. Once both catalogs sit beside their next-session counts, the chi-square question is whether a cell leaves the even-split baseline, not whether a directional pattern deserves a name as a forecast rule.