15.1 Introduction to Relative Strength
Key Takeaways
- Comparative relative strength measures one asset's performance versus a benchmark or peer; it is not Welles Wilder's Relative Strength Index (RSI).
- A security can fall in price and still show rising relative strength if it declines less than the benchmark.
- Three assessment methods are the ratio chart, universe ranking or percentile, and the RS line versus its own moving average.
- A universe can be ranked cross-sectionally by trailing performance or self-relatively by price or RS versus its own moving average.
- Benchmark choice, lookback, survivorship bias, and risk differences can all distort relative-strength conclusions.
Comparative relative strength (RS) is how one security, sector, or asset class performed versus a benchmark or a peer. It is a ratio or a rank. It is not the Relative Strength Index (RSI), Welles Wilder's 0–100 oscillator of a single security's own average gains versus average losses.
The 2026 CMT Association Program Guide places this unit in Section Nine: Comparative Market Analysis. CMT Level I items love the name collision. If a stem says "relative strength" and the choices mix ratio charts with 70/30 oscillator readings, the oscillator is the trap.
What comparative relative strength is
Relative strength analysis asks a competition question: who is winning the fight for capital? You divide one price series by another, or you rank many series by trailing performance, so you can see outperformance and underperformance even when every price is rising or every price is falling.
The simplest construction is a ratio:
RS = Price of asset A ÷ Price of asset B (benchmark or peer)
- A rising ratio means A is outperforming B (A is up more, or down less).
- A falling ratio means A is underperforming B.
- A flat ratio means A and B are moving at about the same rate.
Asset B is a choice, not a law of nature. Common denominators on CMT-style questions include a broad equity index, a sector exchange-traded fund (ETF), a peer stock, a bond proxy, or another asset class. Change the denominator and you change the question.
RS is not RSI
| Feature | Comparative relative strength (RS) | Relative Strength Index (RSI) |
|---|---|---|
| What it compares | One asset versus a benchmark or peer | One asset versus its own recent gains and losses |
| Typical output | Unbounded ratio, rank, or percentile | Bounded oscillator, usually 0–100 |
| Classic construction | Price A / Price B, or trailing-return rank in a universe | Wilder average gain ÷ average loss, often 14 periods |
| Reads as | Leadership versus laggardship | Internal momentum; overbought/oversold of that name |
| Can rise while price falls? | Yes, if the name falls less than the benchmark | RSI can bounce from oversold while the name is still in an absolute downtrend |
J. Welles Wilder introduced RSI in 1978 in New Concepts in Technical Trading Systems. The word "relative" inside RSI means "gains relative to losses on this chart," not "this stock relative to the S&P 500." A stock can print RSI 75 (internally hot) while its ratio versus the index is rolling over (losing leadership). The reverse also happens: RSI can look washed-out on a name that is still a relative leader because it is dropping less than the tape.
Principles behind RS analysis
- Capital is finite. Flows toward names and groups that are already winning tend to continue until the relative trend breaks.
- Outperformance is not the same as a rising price. Losing 4% while the benchmark loses 10% is still relative strength.
- Underperformance is not the same as a falling price. Gaining 3% while the benchmark gains 12% is still relative weakness.
- Relative trends can persist. Intermediate-horizon ranking (on the order of Levy's classic 26-week window rather than a handful of sessions) is the empirical backbone of RS as a selection tool.
- The RS line is chartable. Treat the ratio as if it were a price: trend, support and resistance, breakouts, and moving averages all apply to the ratio itself.
Absolute trend versus relative trend
Absolute trend is the direction of the security's own price: higher highs and higher lows, or the opposite. Relative trend is the direction of the RS ratio or rank versus the chosen comparison series. CMT Level I expects you to hold both ideas at once. Four combinations show up on exams and on screens:
| Absolute trend (price) | Relative trend (RS) | How to read it |
|---|---|---|
| Up | Up | True leader — rising and beating the comparison series |
| Up | Down | Laggard rally — the name is up, but the tape (or peer) is up more |
| Down | Up | Defensive / holding-up leadership — price is under pressure, but the name is declining less |
| Down | Down | True laggard — falling and losing the comparison |
Worked numbers
Case A — relative strength in a decline. Stock Z closes at $40, then $38 (−5.0%). The S&P 500 proxy moves from 5,000 to 4,600 (−8.0%). RS starts at 40 / 5,000 = 0.00800 and ends at 38 / 4,600 ≈ 0.00826. Price fell; the ratio rose. Z is in an absolute downtrend and a relative uptrend versus that benchmark.
Case B — relative weakness in a rally. Stock Y closes at $80, then $84 (+5.0%). The same benchmark moves from 5,000 to 5,400 (+8.0%). RS starts at 80 / 5,000 = 0.01600 and ends at 84 / 5,400 ≈ 0.01556. Price rose; the ratio fell. Y is in an absolute uptrend and a relative downtrend.
Exam trap: "The ratio rose, therefore the stock must be making new price highs." False. The ratio only says who won the comparison.
Three methods of assessing relative strength
1. Ratio chart
Plot A divided by B as its own series. Apply the same technical toolkit you would apply to a stock: trendlines, breakouts, moving averages, even point-and-figure on the ratio. A ratio making a new high while price is still below its old high is a leadership tell you will see again in the next section.
2. Ranking / percentile
Compute a performance statistic for every name in a defined universe (for example, 13-week percent change, or a weighted blend of several lookbacks). Sort best to worst and convert the order to a percentile or a 1–99 rating. A rating of 90 means that name beat 90% of the universe over the window. This is how many momentum screens and commercial "RS ratings" are built. Ranking answers "where does this name sit in the crowd?" in a way a single ratio versus one index does not.
3. RS line versus its own moving average
Take the ratio (or a related RS series) and overlay a moving average of that RS line. When the RS line is above its average, the relative trend is up; a cross below flags a relative-trend turn. This method turns a squiggly ratio into a regime: in a relative uptrend or not. It is the method that most directly pairs with "buy leaders whose relative line is still in force."
| Method | Input | What "strong" looks like | Best use |
|---|---|---|---|
| Ratio chart | Price A / Price B | Rising ratio, ratio breakouts | Two-asset comparison, intermarket pairs, stock vs index |
| Ranking / percentile | Trailing return (or blend) for every name | High percentile (for example 80–99) | Screening a universe; momentum sleeves |
| RS line vs its MA | Ratio series + moving average of the ratio | RS above its own average | Timing relative-trend turns; filters on a watchlist |
Considerations when you use RS
Benchmark choice. Stock versus a broad index answers market leadership. Stock versus its sector ETF answers group leadership. KO versus PEP answers a pairs question. A semiconductor that beats the S&P 500 but lags its chip-sector proxy is a market participant, not a group leader. Write the comparison question down before you trust the rank.
Lookback. A 4-week rank is noisy and high-turnover. An intermediate window (13–26 weeks is the neighborhood of classic RS research) balances persistence against decay. A 12-month window is slower; commercial 1–99 ratings often weight recent quarters more heavily than the distant months inside that year. Change the lookback and the "leader" list changes. There is no single official CMT lookback — the exam tests that you know the choice matters.
Survivorship. If your historical universe quietly drops names that merged, bankrupted, or delisted, the survivors' past RS ranks look cleaner than live trading would have been. Momentum backtests that ignore delistings overstate results. Build ranks with the names that existed on the ranking date, not only the names that still trade today. The same Comparative Market Analysis outline that teaches RS also asks you to define survivorship bias on indexes — carry that idea into RS ranking.
Risk differences. Raw percent-change ranks overweight high-beta, high-volatility names in a strong tape. A +40% mover with violent swings is not the same animal as a +18% mover with half the volatility. Some practitioners rank residual performance (return after a beta adjustment) or return per unit of volatility. Level I at least requires you to know that unadjusted RS is not a risk-adjusted alpha statistic. If two names have the same RS rank, they need not have the same drawdown profile.
Two approaches to ranking a universe
The Program Guide asks you to describe two approaches to ranking a universe of securities. They are not the same as the three assessment methods above; they are two ways to order the crowd.
Approach 1 — Cross-sectional performance ranking. Give every name the same lookback. Rank by trailing percentage price change, or by excess return versus a shared benchmark, or by a weighted multi-period blend (recent months counting more than distant months). Convert to percentiles. Buy or overweight the top slice; avoid or underweight the bottom. This is the classic "who beat whom in this universe?" sort used in Levy-style tests and in many RS ratings.
Approach 2 — Self-relative (time-series) ranking. For each name, compare current price to that name's own moving average (a widely cited technician construction is current weekly close divided by a 26-week average), or compare each name's RS line to that RS line's own moving average, then rank those ratios. You are ordering the universe by how strong each name's own relative-trend regime is, not solely by who printed the largest raw percent change versus peers. Two names with similar 13-week returns can rank differently if one is extended far above its average and the other has only just crossed back above it.
In practice, desks often combine them: cross-sectional percentile to pick the pond, then RS-versus-its-MA to decide whether the relative trend is still in force.
How portfolio managers and individuals use RS
Portfolio managers use RS to rotate sectors, overweight leaders and underweight laggards versus a policy benchmark, and run systematic screens that still respect liquidity, volatility, and mandate rules. RS is a way to spend risk budget on names that are already attracting capital, and to stop arguing with persistent laggards. A PM may rank 500 names weekly, hold a top-quintile sleeve, and replace names that fall through a rank or RS-versus-MA threshold.
Individual investors and traders usually run a smaller process: a handful of ratio charts (favorite stocks versus SPY, or a sector ETF versus the index), a published RS rating as a filter, and a rule such as "do not add to names with falling RS." They rarely need a 3,000-name ranking engine. They do need the same conceptual split: absolute trend is not relative trend, and RSI is not RS.
Neither group should treat RS as a stand-alone order ticket. It is a selection and environment tool. Combine it with the security's own trend, liquidity, and a risk-exit. Leaders fail. Crowded RS lists can reverse violently after a shock — that is a limitation, not a reason to confuse RS with RSI.
Key Takeaways
- Comparative RS is performance versus a benchmark or peer; RSI is a bounded internal-momentum oscillator of one security.
- Absolute trend is the price; relative trend is the ratio or rank.
- Assess RS with a ratio chart, percentile ranking, and the RS line versus its own moving average.
- Rank a universe cross-sectionally (trailing performance) or self-relatively (price or RS versus its own average).
- Watch benchmark, lookback, survivorship, and risk before you trust a leader list.
Which statement best distinguishes comparative relative strength from the Relative Strength Index (RSI)?
A stock falls 6% while its benchmark falls 12%, and the stock-to-benchmark ratio rises. This most directly shows:
When ranking a universe on trailing relative strength, which issue most directly makes historical tests look cleaner than live results because failed or delisted names vanish from the database?