Rational group structure with a trend screen, RSI, and bands

Two-pass industry grouping

Industry groups can be assembled in two passes. First, firms are placed by line of business. Then members whose price path is a gross outlier versus the rest of the bucket are dropped. That two-pass method is a rational group structure.

The grouping is treated as ongoing work. New firms are added and sectors are rewritten as mergers, acquisitions, and changes of business line occur.

Telephone companies were placed in a communications sector rather than utilities. Utilities were split into natural-gas, water, nuclear-electric, and traditional-electric buckets because those subsets often trade to different agendas.

Momentum ranks and the new-high board

A composite momentum score can be centered on a four-week horizon by blending several related momentum measures. Group tables can list short momentum, intermediate momentum, and money-flow rank together. The same tables can flag 52-week, 26-week, and 13-week new highs or new lows as an early-warning layer.

On the illustrated board, real estate investment showed short momentum, intermediate momentum, and a money-flow rank across the 52-week, 26-week, and 13-week new-high lists. Consumer services and networking appeared on the 13-week new-low list.

How the three screens sit in order

A trend filter keeps only groups whose ordered price, volume, or breadth series still point in one direction over a defined lookback.

The relative strength index is a bounded oscillator built from successive price changes. In this stack it scores how persistently a group is advancing or fading after the trend screen has already kept the group.

Bollinger Bands are a volatility envelope around a moving average of the group series. They are used last, to judge whether the latest print is stretched versus recent range.

Graded membership

A robustness check for a graded, rather than crisp, decision rule is whether the same underlying function still classifies readings after the environment shifts. The check fails if the rule depends on one fixed trigger. That graded assignment is fuzzy membership.

An explicit baseline

Editorial use of the archive workflow is to write the order of screens as an explicit quantitative baseline. The same baseline can then be compared with a later out-of-sample check. The archive itself is not being asked to certify that later check.