Rebuild every industry as a share of one rank scoreboard

One headline trend can hide several cycles

A diversified equity market can host several business cycles at once, so one broad index can hide an industry that is not moving with the headline trend.

Dividing a published sector index by a base index inherits outlier price noise and capitalization or price weighting, so large names can mask small-cap leadership inside the same group. That conventional quotient is a relative-strength ratio.

When names inside one industry do not move tightly together, a capitalization-weighted ratio plot becomes noisy while a ranked rebuild stays cleaner. A sector is any basket assumed to share positive price co-movement, whether an industrial family or a non-industry split such as value versus growth, large versus small capitalization, or a price-level cohort.

Rank the universe, then share it out

Ranked relative strength rebuilds each industry by ranking the full universe on one-day closing change, averaging those ranks inside the group, normalizing the averages so they sum to one, and accumulating each day's gap from an equal-share baseline. The cumulative series is the reconstituted sector index that later rules compare, rotate, and overlay.

The construction uses daily closes on the 1,500 names of the S&P Super Composite, bucketed into ten GICS industries, because liquidity in that universe supports complete, continuous data.

Logarithmic and percentage one-day changes produce the same rank order. With 1,500 names the median sits at the 750th position, and a sector scores above 0.50 when more than half its names beat that median.

If all ten industries had identical returns, each would hold a 0.1 share. Persistent gaps from that equal-share baseline are treated as strength or weakness and graph as a steadily rising or falling cumulative line.

A hyperbolic-tangent map with scaling factor 3 assigns each rank a score from near 0 to near 1, with 0.5 at the median, before the same average, normalize, and accumulate steps.

Read order, slope, and the ten-way share

Rank-rotation selects, avoids, or stands aside from industries by the order and slope of their reconstituted rank series rather than by a single broad-market average.

Industry-rotation reads the ten-way share of ranked strength as the market regime that tells a single name whether its group is leading, lagging, or mixed.

When the ratio and the ranks disagree

On the information-technology window, the standard ratio and the ranked series diverged. The ratio did not take out its earlier peak while the ranked series made a higher high, and from mid-April 2003 into early July that industry led the ranked panel.

From that divergence through the next labeled high, nearly 20% of information-technology names sat in the top 10% of the cross-section, ahead of every industry except health care. That is the breadth a rank-rotation rule would have read instead of the ratio's implied downturn.

Apply the same ranking to a price-level cohort

The same ranking applied to two price-level cohorts split by a 200-day average and a $10 cutoff showed a sub-$10 leadership turn in early April.

An MACD overlay on that reconstituted series confirmed the turn by month-end, and the episode lasted into early July. MACD here marks turning points and zero-line crosses on the reconstituted ranked series. It is not a substitute for the ranking construction.