Comparative relative strength

Definition

Comparative relative strength (RS) measures how a security performs RELATIVE to a benchmark or peer — typically as a ratio line (stock price ÷ index price) or a trailing relative return. It is the selection principle behind momentum investing at the individual- name level: Levy (1967) provided the early academic evidence that high-RS stocks continued outperforming, a generation before Jegadeesh-Titman formalized cross-sectional momentum. Distinct from indicator-rsi (an internal oscillator that unfortunately shares the name), comparative RS answers the portfolio question directly: of everything I could hold, what is actually leading?

How it works / structure

• The ratio line: plot security/benchmark; rising = outperformance regardless of absolute direction (a stock falling less than the index has rising RS — the classic bear-market leadership tell for the next cycle, practitioner canon labeled as such).
• Quantified forms (engine-relevant): trailing relative return over 3-12 months (the strategy-momentum formation window, standardly skipping the last month for short-term reversal); RS RANK across a universe (percentile of trailing return — Levy’s construction, and the IBD-style 1-99 rating popularized from it, labeled practitioner); ratio-line trend filters (RS above its own moving average).
• Why it persists: the momentum literature’s explanations apply directly — underreaction and slow information diffusion (strategy-momentum carries the full evidence stack including the crash risk); RS is momentum’s per-name measurement instrument.
• The rotation application: sector/asset-class RS ranks drive strategy-sector-rotation implementations; relative-strength switching between broad assets is among the oldest systematic rules with out-of-sample history (documented in the tactical-allocation literature, dispersed results).

When it applies

Name selection within momentum and growth frameworks (buy leadership, not laggards — the RS discipline); sector/asset rotation ranking; relative weakness screening for short candidates or avoidance; bear-market next-cycle-leadership scans (what holds RS through the decline). Requires a universe and a benchmark choice — both are parameters that change results (quant-backtest-hygiene labeling rule).

Risk profile & failure modes

• Momentum’s inherited risks: RS selection walks into momentum crashes (2009-style junk rallies invert leadership violently — the documented −70%+ momentum-factor months) and high turnover costs; every strategy-momentum caveat applies unchanged.
• Benchmark sensitivity: RS against the S&P vs equal-weight vs sector gives different ranks — benchmark shopping is a researcher degree of freedom that must be pre-specified.
• Ratio-line illusions: a rising ratio of two falling prices still loses money — RS is a RELATIVE statement; absolute risk controls remain mandatory.
• Turn-point lag: RS is trend-following by construction — leadership rotates BEFORE ranks update; rotation regimes (2000, 2022 growth-to-value) fire the whipsaw cost up front.

Evidence & limits

Levy (1967) is the early peer-reviewed evidence; the modern momentum literature (cited in strategy-momentum) is the durable foundation — cross-sectional relative return persistence is among the most replicated anomalies, with the crash caveat equally documented. Rating-style transformations (1-99 ranks) are practitioner conventions on top of the same signal.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Top-decile 6-month RS names in the S&P 1500 beat bottom-decile by 4%+ annualized over 3 years, skip-month construction (Levy/momentum replication)” — falsified by the decile spread.
• “Stocks with rising RS through a 15%+ index correction outperform in the 12 months after the low (bear-leadership thesis)” — falsified by the cohort’s forward relative return.

Cross-references

• The factor it instruments: strategy-momentum; the rotation user: strategy-sector-rotation
• The name-collision disambiguation: indicator-rsi (internal oscillator, different object)
• The hygiene constraints: quant-backtest-hygiene