Four-dimension relative strength as rank rotation

Score first, then rank

The four-dimension procedure scores a security by pairing it with many reference series rather than with a single index, then collapses those pair scores into one rank.

Rank-rotation, as used here, scores each candidate against a shared basket of market references, ranks the resulting scores, and uses that order to decide which holdings to keep, rotate, or skip.

TradersWeek editorial view: treat portfolio construction as that ranking procedure, not as a single-index residual.

Build each pair the same way

Each pair is formed as the candidate close divided by a reference close, then passed through the same multi-horizon moving-average state machine.

The state machine uses a fast exponential average of the pair series, then medium, slow, and very-slow simple averages of that fast line. It maps the four-line order into discrete tiers of 10, 9, 5, or 0.

Collapse the basket into one composite

The four-dimension score is the stacked moving-average state of a price-ratio series, coded as a discrete tier and then averaged across many reference pairs.

A composite oscillator is the mean of the pair tiers across the reference basket, scaled, then smoothed. A shorter average of that oscillator is plotted as a signal line.

The signal line separates a raw rank from its recent trend. Once the basket is reduced to a single numeric score that can be ranked across candidates, that score is a relative-strength-index: a multi-horizon comparison of a security's price path to several reference series.

Use a cross-context reference basket

Reference baskets in the supplied implementations span large-cap, mid-cap, small-cap, growth, value, emerging-market, real-estate, gold, utility, and bond series. The rank is a cross-context score, not a single-index residual.

Sort many targets on one sheet

A batch ranking sheet evaluates many target symbols against the same benchmark set. It can sort first by one composite field and then by a tie-break field, or the reverse, and still produce the same order.

Keep every series on the same date

Target and benchmark prices must be retrieved as of the same date. Otherwise the ranks mix unmatched observations.

Shared-date evaluation means retrieving every target and every benchmark as of the same calendar date so ranks compare like observations.

Benchmark history need be pulled only once per evaluation day unless the mix of references changes. Per-symbol scoring after that is treated as inexpensive.