Skip to content
Menu
Finabase Learn

Strategies · Trend and momentum

Relative strength rotation: holding the strongest fund, tested

A relative strength rotation holds whichever fund has risen most over recent months. The exact rules, a 22-year test on five ETFs with costs, and the years it lost badly.

On this page
  1. The idea
  2. The rules
  3. What the backtest shows
  4. Where it fails: the losing periods
  5. Compared with its neighbours
  6. How Finabase does it
  7. Frequently asked questions
  8. Summary
  9. Check your understanding

A relative strength rotation keeps a short list of funds, ranks them once a month by how much each has risen lately, and holds the one at the top. When another fund takes the lead, it switches. The idea rests on relative momentum: across many studies since the 1960s, assets that did better than their peers over the past several months have tended, on average, to keep doing better for a while.

This lesson states a rotation rule exactly and tests it on five large exchange-traded funds from 2004 to 2026. The rule finished ahead of SPY, the S&P 500 fund, by 0.7 points a year. That headline hides nearly everything that matters: a 61% fall in 2008, eleven years of trailing since 2015, trading costs of almost a point a year, and one fund on the list that beat the whole strategy by simply being held.

The idea

Relative strength compares assets with each other. It does not ask whether a fund is going up. It asks which fund is going up most, or falling least. The Relative strength and market breadth: ratio lines, A/D and % above MA lesson shows how to read it on a chart as a ratio of two prices.

A rotation turns the comparison into a holding. The bet is that leadership lasts: the fund that led over the past six months is more likely than the others to lead next month. Jegadeesh and Titman found this among US stocks in 1993, using lookbacks of three to twelve months, and later studies found it among countries, sectors and asset classes.

Two features follow from the design, and both show in the test.

  • It is always invested. The strongest of five falling funds is still falling. Relative strength alone never moves to cash. Time-series momentum is the rule that does.
  • It chases. By construction it buys what has already gone up. When leadership changes hands quickly, it keeps arriving late.

The rules

The list, fixed in advance: SPY (S&P 500), QQQ (Nasdaq-100), IWM (US small companies), EFA (developed markets outside North America) and EEM (emerging markets).

Checked once a month, on the last trading day:

  1. For each fund, divide today's close by its close six months earlier and subtract one. That is its six-month return.
  2. Hold the fund with the highest six-month return for the next month, with the whole amount.
  3. If that is the fund already held, do nothing. If not, sell and buy at today's close.

The list matters as much as the rule. These five were chosen because they are large, old enough to test and cover different parts of the stock market. They are all stock funds, so the test says nothing about rotating between stocks, bonds and gold.

Which of the five funds the rule held, 2004–2026 Cash SPY QQQ IWM EFA EEM Fund held 2005 2010 2015 2020 2025
Which of the five funds the rule held, month by month, 2004 to 2026. Each jump is a switch, and each switch is a sale and a purchase.

A worked example

At the end of April 2019 the rule held EEM. One month later the six-month returns were: SPY −0.1%, QQQ 2.7%, IWM −4.4%, EFA 1.0% and EEM −0.9%. QQQ led, so the rule sold EEM and bought QQQ at the May close.

QQQ then stayed on top at every month-end for a year and a half. The rule did not trade again until the end of November 2020, when IWM's six-month return of 30.3% passed QQQ's 28.4%. That stretch is why 2020 was the rule's best year: it gained 52.3% while SPY gained 16.2%.

What the backtest shows

The test, stated in full:

  • Data. Month-end closes of the five funds. EEM, the youngest, begins in May 2003. Twelve months are set aside so that every lookback in the comparison table starts on the same date, and the test runs from May 2004 to October 2026: 270 months.
  • Trading. Ranking and trading both use the month-end close.
  • Costs. 0.10% of the amount on each sale and each purchase, so a switch costs about 0.20%.
  • Left out. Dividends, taxes and each fund's expense ratio beyond what is already in its price. The funds' dividend yields differ (EFA and EEM pay more than QQQ), so a ranking on total return could differ slightly from this one.
  • The comparison is SPY bought once and held.
SPY, QQQ, IWM, EFA, EEM monthly, 2004–2026: holding the strongest over six months against SPY, 0.10% per trade, no dividends $1 $2 $5 Growth of $1 SPY, held throughout Strongest of five 0% −20% −40% −60% Fall from the peak 2005 2010 2015 2020 2025
Five funds monthly, 2004 to 2026: holding the strongest over six months against SPY held throughout, 0.10% per trade, dividends not included. Top: growth of $1 (log scale). Bottom: the fall from the highest value so far.
Measure Strongest of five SPY, held
Total return 688.4% 569.6%
Return per year (CAGR) 9.6% 8.9%
Deepest fall from a peak −61.0% −52.2%
Longest time below a peak 72 months 64 months
Time invested 100% 100%
Switches of fund 83 0

The 83 switches turned the whole portfolio over 3.7 times a year. At 0.10% a trade that cost about 0.9 points of return a year: with no costs the rule returns 10.5% a year. A rotation trades far more than the monthly trend rules in this section, so the cost assumption matters far more.

The rule held QQQ in 38% of months, EEM in 27%, IWM in 19%, EFA in 9% and SPY in 7%.

What each fund did when simply held over the same months:

Held throughout Return per year Deepest fall
QQQ 14.0% −50.0%
SPY 8.9% −52.2%
IWM 7.4% −53.5%
EEM 5.7% −61.9%
EFA 3.7% −59.7%
All five, equal parts 8.2% −55.2%

The rotation beat four of the five funds and the equal mix. It trailed QQQ by 4.4 points a year. Of course nobody knew in 2004 that QQQ would be the winner, and that is the honest way to read the table: the rule did better than holding everything equally, and much worse than hindsight.

The two halves of the period

Period Strongest of five SPY, held
May 2004 – January 2015 10.4% a year 5.5% a year
January 2015 – October 2026 9.0% a year 12.0% a year

The whole advantage was earned before 2015, and most of it before 2008. From 2004 to 2007 emerging markets led month after month, and the rule gained 31.2% in 2005 and 31.6% in 2007 while SPY gained 3.0% and 3.2%. Since 2015 it has trailed SPY in eight of twelve calendar years, counting 2026 to October.

Where it fails: the losing periods

A broad decline: 2008. The rule lost 46.0% in 2008 against 38.3% for SPY, and 61.0% from October 2007 to February 2009. Between April 2008 and February 2009 it switched nine times, and on most of those dates every fund on the list had a negative six-month return. In November 2008 it moved to SPY because SPY was down 35.8% and the others were down more. Picking the least bad of five falling funds is not protection.

Leadership that keeps changing: 2022. The rule lost 31.2% while SPY lost 19.5%. It held QQQ into January 2022, moved to SPY, and then made six more switches in eight months (EFA, EEM, IWM, SPY, IWM, EFA), each into a fund that had just fallen a little less than the rest.

A steady leader it kept leaving: 2024. SPY gained 23.3% and the rule gained 6.3%. Five switches between July and November, among funds whose six-month returns were within a few points of each other, each cost two trades (about 0.20%), and the rule ended the year 17 points behind.

Other ways it fails or misleads:

  • The list is a choice. Add a bond fund or drop EEM and the history changes. A test of one list on one period is one draw.
  • Taxes. With 83 switches, nearly every gain is realized within a year and taxed as ordinary income in a US taxable account (IRS Topic 409). SPY held throughout defers all of it.
  • Close calls decide trades. In January 2010 SPY led QQQ by 8.7% to 8.5%. A rule that switches on a 0.2 point gap is acting on noise.

Neighbouring settings

Lookback Hold the top one Hold the top two
3 months 10.0% a year 9.1% a year
6 months 9.6% a year 8.7% a year
9 months 8.5% a year 9.2% a year
12 months 10.5% a year 10.2% a year

Every version fell between 56% and 63% from peak to trough. The returns run from 8.5% to 10.5% with no pattern: nine months is the worst for one fund and the second best for two. When neighbouring settings scatter like this, the differences are luck, and the spread itself, two points a year, is a fair measure of how uncertain the 9.6% is.

Compared with its neighbours

Rule Can hold cash Trades Deepest fall in its test
Relative strength rotation (this page) No About 4 a year −61.0%
Time-series momentum on SPY Yes One in two years −20.8%
SPY held throughout No None −52.2%

Relative and absolute momentum answer different questions. One picks what to hold; the other decides whether to hold anything. The tests on this page and in Time-series momentum: the 12-month return rule, tested cover different periods, so the falls are not a like-for-like comparison, but the reason for the gap is structural: only one of the two rules can leave the market.

How Finabase does it

Finabase does not rank funds for you or tell you which one to hold. Two tools let you explore the pieces.

  • Strategy. The Strategy tool is built for allocation rules of this kind: if a condition holds, put a weight in one symbol, otherwise in another. It replays the rule over two years of real prices. Two years of a rotation is about eight switches, enough to see how the rule behaves and far too few to judge it. It charges no trading costs, which matters for a rule that trades this often.
  • Chart. The Rate of Change (ROC): percentage momentum and how to read it study shows a fund's return over a chosen number of bars. Set to 126 daily bars it is the six-month return the rule ranks by; the link above opens it on QQQ.
  • Backtest. The Backtest tool works on one symbol at a time, buys a fixed $10,000 on each signal and charges no costs. It cannot test a rotation.

Frequently asked questions

Does a relative strength rotation beat the market?

In this test it beat SPY by 0.7 points a year over 22 years, before taxes, with a deeper worst fall. It was ahead by a wide margin until 2015 and behind afterwards. A different list of funds or a different period would give a different answer.

Why did the rotation fall more than SPY in 2008?

Because it must always hold one of the five funds, and in 2008 all five fell. It also entered the decline holding emerging markets, which fell furthest, and then switched nine times in eleven months.

How often should a rotation rebalance?

Monthly is the common choice and the one tested here. Checking more often means more switches and more costs; this version already lost about 0.9 points a year to a 0.10% cost per trade.

What is the difference between relative strength and RSI?

They are unrelated despite the name. Relative strength compares one asset with another. The Relative Strength Index compares an asset's recent gains with its own recent losses.

Summary

A relative strength rotation holds the fund with the best recent return and switches when another takes the lead. On five stock funds from 2004 to 2026 it returned 9.6% a year against 8.9% for SPY, after costs that took almost a point a year, and its deepest fall was 61.0%. It earned its whole lead while emerging markets ran from 2004 to 2007, lost more than the market in 2008 and 2022 because it can never hold cash, and has trailed SPY since 2015. The settings table scatters without pattern, which is a sign that the small lead is not a dependable one.

Check your understanding

4 questions on this lesson. Each answer comes with the reason, and nothing is scored.

  1. At a month-end the six-month returns of the five funds are −36%, −42%, −37%, −46% and −55%. What does the rotation hold next month?

    1. Cash, because every fund is falling
    2. The fund that is down 36%
    3. Equal parts of all five
    Show the answer

    B. The fund that is down 36% Relative strength only ranks. The least bad fund is still the top of the list, so the rule holds it. These were the returns in November 2008, when the rule moved to SPY.

  2. The rule returned 9.6% a year with a 0.10% cost per trade and 10.5% with no cost. Why is the gap so large compared with a monthly moving average rule?

    1. The funds are more expensive
    2. It switched 83 times, turning the whole portfolio over about 3.7 times a year
    3. Costs rise with the price of the fund
    Show the answer

    B. It switched 83 times, turning the whole portfolio over about 3.7 times a year Each switch is a sale and a purchase of the whole amount. A rule that trades four times a year pays roughly eight times the cost of one that trades once every two years.

  3. QQQ returned 14.0% a year held throughout, more than the rotation’s 9.6%. What is the fair conclusion?

    1. The rotation should have held only QQQ
    2. Hindsight picks the winner; the rule beat four of the five funds and the equal mix without knowing it
    3. The test is wrong
    Show the answer

    B. Hindsight picks the winner; the rule beat four of the five funds and the equal mix without knowing it Nobody could know in 2004 which fund would lead for 22 years. The table shows what the rule added over holding everything equally, and how far short of perfect foresight it fell.

  4. Eight combinations of lookback and number of funds returned between 8.5% and 10.5% a year, in no clear order. What does that suggest?

    1. The 12-month lookback is best
    2. The differences are mostly luck, and the spread shows how uncertain any one result is
    3. Longer lookbacks are always better
    Show the answer

    B. The differences are mostly luck, and the spread shows how uncertain any one result is If one setting were truly better, its neighbours would be nearly as good and the table would show a slope. A scatter means the ranking of settings would probably change on another period.

Sources and further reading

  1. Narasimhan Jegadeesh and Sheridan Titman, “Returns to Buying Winners and Selling Losers,” Journal of Finance 48(1), 1993 — relative momentum among US stocks
  2. Robert A. Levy, “Relative Strength as a Criterion for Investment Selection,” Journal of Finance 22(4), 1967
  3. Mebane T. Faber, “Relative Strength Strategies for Investing” (2010) — rotation among sectors and asset classes
  4. IRS — Topic no. 409, Capital gains and losses
  5. Finabase chart data — monthly bars for SPY, QQQ, IWM, EFA and EEM, May 2003 to October 2026; rules and results computed by Finabase