At a glance
- RSL strategies belong to the momentum approaches: the stocks of a universe are ordered by Levy's Relative Strength (RSL), with the selection concentrating on the top-ranked ones.
- The measure goes back to Robert A. Levy (1967). The implementations common today, with rankings, a weekly rhythm and exit thresholds, developed later from his idea.
- This page describes the strategy and its publicly documented variants. It is not investment advice and gives no instruction to act.
- How to use the ranking in Foliograph is described under RSL ranking in Foliograph.
The idea
Momentum describes the empirical tendency of stocks with a strong past price performance to remain relatively strong over certain periods afterwards. An early study of this principle comes from Robert A. Levy, who in 1967 in the Journal of Finance ranked stocks by their relative strength against their own price history.
Levy examined 200 stocks of the New York Stock Exchange from October 1960 to October 1965, ordered them every week by his measure C/A26 and compared, among other things, the strongest with the weakest groups. Over the following 26 weeks stocks from the previously strong rank groups did better on average than stocks from the weak ones.
Narasimhan Jegadeesh and Sheridan Titman documented in 1993, with a different method (ranking by past returns), a continuation of relative stock returns over 3 to 12 months as well. Since then an extensive academic literature on momentum has emerged. It supports the effect in general, not Levy's specific measure.
The measure
RSL = last weekly close / average of the last 27 weekly closes
Levy called the measure C/A26: the current price (C) divided by the average (A) of the current weekly price and the 26 weeks before. The average therefore covers 27 prices. Some programs use only 26 prices, which makes the values come out slightly different.
- An RSL of 1.20 means: the price is 20% above its average.
- An RSL of 0.90 means: the price is 10% below it.
- The RSL compares every stock with its own past. Only the ranking turns it into a comparison between stocks.
Some programs use about 130 trading days instead of 27 weekly closes. This covers roughly the same horizon, but the values are not identical.
A common implementation
A common implementation of the RSL idea looks like this. It does not come from a professional publication and is only described here:
- Universe: the basis is an index or a sub-index, for example the S&P 500.
- Ranking: all stocks of the universe are sorted by RSL, the highest RSL is rank 1.
- Selection: for example the five or ten top-ranked stocks are held, each with the same amount. Many keep an eye on the spread across industries and fall back on lower-ranked stocks of other industries when one industry piles up.
- Rhythm: the ranking is checked once a week, for example on Monday with Friday's closing prices.
- Replacement: a stock leaves the portfolio when it is no longer in the top quarter of the ranking (in the S&P 500 from about rank 126). Its place goes to the highest-ranked stock not yet held.
This rule set does not come from Levy's study in the Journal of Finance. Levy examined the measure and several rank groups. The specific rules for portfolio, replacement and spread were derived from it later.
Checking the ranking
A ranking is a calculation over past prices. People who follow such an implementation therefore often do not take the top ranks unchecked. Two cases come up again and again.
Clusters in one industry. When several stocks of the same industry or sub-industry did well over the last months, they sit at the top of the list together. Five semiconductor stocks can therefore form a considerable industry risk, even though they are formally five separate stocks. They often react to similar industry and economic news. If, for example, three of the first five stocks are semiconductor stocks, some take two stocks of other industries from ranks 6 to 10 instead, so that the selection spreads over several industries. A spread across different segments such as software, hardware and IT services can reduce the concentration risk but does not remove a common sector risk.
Announced takeovers. When a takeover offer becomes known, the price often jumps towards the offered price. Afterwards it often moves in a fairly narrow range as long as the market expects the takeover to close. The RSL rises sharply without a trend having formed and the stock suddenly appears high up in the ranking. The further price development then depends largely on the offer price and the likelihood of the takeover. After a takeover has closed, the stock is usually removed from the index. Signs are a single large jump in the chart, an unusually steady price afterwards and news about the company.
Andreas Clenow also uses a filter for unusually large price moves in Stocks on the Move: stocks with a single move of more than 15% within the last 90 trading days are excluded for the time being. Clenow does not rank by the RSL but by his own trend measure over 90 days. The filter can be applied to an RSL ranking just as well.
Publicly described variants
The following extensions are described in articles and books. Each changes how often positions change and how the portfolio behaves in weak market phases.
| Variant | What it changes | Source |
|---|---|---|
| RSL threshold 1 | Additional exit when the RSL of the stock falls below 1, the price falls under its average. | Bauer, Derivate Magazin 2/2012 |
| Comparison with the market | Exit when the RSL of the stock falls below the average of all stocks of the universe. | common variant, no professional publication |
| Market filter | New positions only while the RSL of the index itself is above 1, the index is above its average. | Bauer, Derivate Magazin 1/2012 and 2/2012 |
| Rank indicator | The percentage rank position counts instead of a fixed rank, for example "top 20%". This makes rankings of indices of different size comparable. | Paesler (2011), Bauer, Derivate Magazin 1/2012 |
| Trailing stop | Exit when the price falls a fixed percentage below its high since entry (15% in the article). | Bauer, Derivate Magazin 2/2012 |
| Longer rhythm | Check every six months instead of weekly. Fewer changes mean lower costs. | Bauer, Derivate Magazin 1/2012 |
| Gap filter | Stocks with a single move of more than 15% within the last 90 trading days are left out, for example after a takeover offer. | Clenow (2015) |
| Skip the last month | Performance is measured over 12 months without the last month, because very short-term moves often reverse. | Jegadeesh (1990), Carhart (1997) |
| Nearness to the 52-week high | Rank by the distance to the 52-week high instead of past return. | George and Hwang (2004) |
| Absolute momentum | Select only when the strongest assets also show a positive return above risk-free investments. | Antonacci (2014) |
| Account for volatility | Scale the share of the strategy by its recent volatility to dampen sharp setbacks. | Barroso and Santa-Clara (2015), Daniel and Moskowitz (2016) |
A three-part article series by Gregor Bauer in Derivate Magazin combined for example RSL, market filter, rank indicator and stop and examined these rules in a look back over about 20 years on DAX stocks. The tests come from Oliver Paesler, and the RSL was computed with 130 daily prices instead of weekly prices. Such historical simulations do not carry the same weight as peer-reviewed scientific studies. They depend heavily on the data, the index composition, the parameters and the costs taken into account.
What Foliograph covers
Foliograph shows the RSL ranking and the related measures. It applies no rules and suggests no values. You can set your own criteria for highlights as well as your own thresholds for held stocks.
| Measure or variant | Setting in Foliograph |
|---|---|
| Top-ranked stocks | Highlight rows: "Within the first" |
| Rank indicator | "Top %" respectively "Outside top %" |
| Top quarter | Thresholds for held stocks: "Outside top %" or "Outside the first" |
| RSL threshold 1 | "RSL at least" respectively "RSL below" |
| Comparison with the index | "RSL above the index RSL" respectively "RSL below the index RSL" |
| Comparison with the market | "RSL above the average RSL" respectively "RSL below the average RSL" |
| Market filter | "Index RSL above 1" |
| Weekly rhythm with Friday's prices | "Completed weeks only" |
| Trend of the stock | SMA pair of the weekly closes, default 10 and 21 weeks |
| Spread across industries | Sector and Industry columns, both with a filter in the header row |
| Gap filter | Jump 90d column (last 90 trading days) and "No price jump above (%)", plus the chart panel (click on the row) |
The period can be changed. The default is 27 weekly closes, matching the period length of Levy's C/A26. Details are under RSL ranking in Foliograph.
Limits
- Original study: Levy's sample was assembled in hindsight. It only contained stocks that were still listed on the NYSE in 1965. Stocks that disappeared in between are missing from it. This can make the results look too favourable.
- Price data: Levy used price series that were also adjusted for dividends and other distributions. Foliograph uses split-adjusted prices from IBKR, which do not include dividends. The RSL values can therefore differ slightly despite the same formula.
- Costs and replication: Michael Jensen and George A. Benington checked Levy's rules in 1970 with further samples, mostly outside his study period. After transaction costs they found no advantage over buy and hold. They attributed Levy's result to a selection bias, because Levy had picked his rule from 68 rules he tested. A weekly rhythm creates many changes, so commissions, spreads and taxes weigh accordingly.
- Momentum crashes: momentum strategies can go through rare but severe loss phases. Daniel and Moskowitz (2016) show that these occur especially after strong market declines and during the fast recoveries that follow.
- Survivorship bias: computing past rankings with today's members of an index leaves out stocks that have left the index. Historical results can be distorted by this and in particular look too favourable.
- The past: all variants evaluate past prices. Historical relationships do not guarantee that the momentum effect or a specific strategy will repeat in the same way.
Sources
Levy and replication
- Robert A. Levy: Relative Strength as a Criterion for Investment Selection. The Journal of Finance, Vol. 22, No. 4 (1967), pp. 595-610. DOI 10.1111/j.1540-6261.1967.tb00295.x
- Michael C. Jensen, George A. Benington: Random Walks and Technical Theories: Some Additional Evidence. The Journal of Finance, Vol. 25, No. 2 (1970), pp. 469-482. DOI 10.1111/j.1540-6261.1970.tb00671.x
Academic momentum research
- Narasimhan Jegadeesh: Evidence of Predictable Behavior of Security Returns. The Journal of Finance, Vol. 45, No. 3 (1990), pp. 881-898. DOI 10.1111/j.1540-6261.1990.tb05110.x
- Narasimhan Jegadeesh, Sheridan Titman: Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. The Journal of Finance, Vol. 48, No. 1 (1993), pp. 65-91. DOI 10.1111/j.1540-6261.1993.tb04702.x
- Thomas J. George, Chuan-Yang Hwang: The 52-Week High and Momentum Investing. The Journal of Finance, Vol. 59, No. 5 (2004), pp. 2145-2176. DOI 10.1111/j.1540-6261.2004.00695.x
- Mark M. Carhart: On Persistence in Mutual Fund Performance. The Journal of Finance, Vol. 52, No. 1 (1997), pp. 57-82. DOI 10.1111/j.1540-6261.1997.tb03808.x
- Pedro Barroso, Pedro Santa-Clara: Momentum Has Its Moments. Journal of Financial Economics, Vol. 116, No. 1 (2015), pp. 111-120. DOI 10.1016/j.jfineco.2014.11.010
- Kent Daniel, Tobias J. Moskowitz: Momentum Crashes. Journal of Financial Economics, Vol. 122, No. 2 (2016), pp. 221-247. DOI 10.1016/j.jfineco.2015.12.002
Books
- Gary Antonacci: Dual Momentum Investing. McGraw-Hill, 2014.
- Andreas F. Clenow: Stocks on the Move: Beating the Market with Hedge Fund Momentum Strategies. 2015.
German-language RSL variants
- Gregor Bauer: Relative Stärke, Strategien zur systematischen Aktienauswahl, parts 1 to 3. Derivate Magazin 1/2012, 2/2012 and 3/2012 (tests by Oliver Paesler).
- Oliver Paesler: Das Konzept der Rang-Indikatoren. Paper for the VTAD Award 2011.
Related
This page is also in the app under Help, in the language of the interface. Question not answered? Write to info@foliograph.ch or use the contact form.