RSI - Relative Strength Index – signal statistics for the Nordic markets, 1996 to 2018

Published in English on 20 February 2019. Norwegian original here >>

Is overbought RSI a sell signal? Or is RSI first and foremost a momentum indicator, meaning buy when RSI is high? We have studied price development following RSI at extreme levels on the Stock Exchanges in Oslo (Norway), Stockholm (Sweden), Copenhagen (Denmark) and Helsinki (Finland).

RSI - Relative Strength Index

RSI is an oft-used indicator in technical analysis. RSI means Relative Strength Index, and it measures how well a stock has performed compared to itself. The figure is calculated by looking at the strength of days with rising prices compared to strength of days with falling prices over a certain period of time and it gets a value between 0 and 100.

RSI is calculated thus:

      priceUp  = sum(changes(upDays))
      priceDown=-sum(changes(downDays))
      rsi= 100 * priceUp/(priceUp+priceDown)

 

RSI is often calculated for the past 21 days (RSI21). A stock which has risen a lot compared to reactions back during the time period has high RSI. Similarly, a stock which has fallen a lot compared to reactions upwards during the time period has low RSI. RSI is used both as an indicator of whether a stock is overbought or oversold, and whether it has positive or negative momentum.

How can a stock’s RSI value predict future price development? Will a stock that has risen a lot, and reached a high RSI value, most likely keep rising or will it react back? Will a stock with low RSI, which indicates it is oversold and has negative momentum, see a reversal upwards or continue falling?

RSI as overbought/oversold indicator

A stock is overbought when it has risen a lot in a short time, without any significant reactions back during the time period. The idea is that it has risen too much and will have a reaction back soon, indicating that it is time to sell the stock. Similarly, a stock is oversold when it has fallen a lot in a short time. This may indicate that it will have a reaction up soon and should be bought before then. See example below.

Figure 1: RSI21 in Carlsberg B. Green areas show oversold RSI (below 30) and red areas overbought RSI (above 70).

RSI as momentum indicator

RSI is also used as a momentum indicator. The idea is that major price movements indicate that the investors as a group are in motion. It is thought necessary to join the early movements, as even more investors will join later and drive the price further in the same direction. See example below.

Figure 2: RSI21 in D/S Norden. Red areas show high momentum and positive (RSI above 70) and green areas show high momentum and negative (RSI below 30).

Research into RSI

In 2015 Investtech conducted a study looking to determine to which degree RSI can be used to predict future price development. The project was part of a bigger signal analysis project, and supported by the Research Council of Norway. In 2019 we have conducted new analyses with a bigger data set and are reporting the new research results.

We previously found that RSI was a good momentum indicator and did not perform well as an overbought/oversold indicator. How have the RSI signals performed in recent years and what does the statistics say about consistency over time? Can the RSI indicator still be used to find good buy signals?

Data Set

We have studied stocks on the Nordic markets Norway, Sweden, Denmark and Finland. For Norway we have data from 1996, and for Sweden, Denmark and Finland from 2003. For all countries we looked at data until 31. December 2018. We have identified signals until 27 September 2018, in order to study price development for 66 days (3 months) after the last signal.

All stocks that have been listed in the period are included. Stocks that have been delisted due to for instance mergers, takeovers and bankruptcy are included. All prices are adjusted for splits, dividend payments, reverse splits, and other corporate capital changes. Minor dividends in foreign currencies are generally not registered, with the exception of dollar dividends on the Oslo Stock Exchange.

Minimum liquidity requirement was set to 500,000 krone as a daily average for the past 22 days. For Finland we set the limit to 50,000 Euro.

RSI21 to critical levels

We have studied what happens when RSI falls below 30 and breaks above 70. According to technical analysis literature, these are critical levels and are used as buy and sell signals.

When RSI fell below 30, this triggered a sell signal based on the momentum strategy. When RSI broke above 70, a buy signal was defined based on the momentum strategy. Please note that this is the opposite of how many traders use RSI and the opposite of the signal definition in the 2015 report.

Number Norway Sweden Denmark Finland Total
Buy signal RSI21 breaks above 70 9,258 16,472 4,900 5,103 35,733
Sell signal RSI21 breaks below 30 6,691 10,846 2,972 3,081 23,590
Total number 15,949 27,318 7,872 8,184 59,323
Share 27 % 46 % 13 % 14 % 100 %

The table shows number of buy and sell signals based on a momentum strategy for RSI21.  Investtech’s computers identified a total of 35,733 cases where RSI broke above 70. This is a buy signal based on momentum strategy and are called “buy signal” in the charts below. Our computers also identified 23,590 cases where RSI21 broke below 30, which is a sell signal in the momentum strategy and is called “sell signal” in the charts. The data country distribution is 27 % from Norway, 46 % from Sweden, 13 % from Denmark and 14 % from Finland.

The signals were identified by studying all cases where RSI crossed the critical levels. Only signals where it was more than 14 days since the previous signal in the same stock were counted, in order to avoid near-doubles. Yet a big number of signals will be quite similar, because quite a few stocks follow each other closely in terms of prices. Especially when the exchange has risen or fallen a lot in a short time, many stocks will trigger signals at the same time. The signals (samples) will therefore not be independent, and statistical measures for uncertainty (standard deviation) cannot be used indiscriminately.

Buy and sell signals are compared to each other and the benchmark index. We used the standard dividend adjusted indices, respectively Hovedindeksen (OSEBX), OMX Stockholm Benchmark GI (OMXSBGI) and OMX Copenhagen 25 GI (OMXC25GI). For Finland we used an average of the Scandinavian indices and OMX Helsinki (HEX), as Nokia weighs heavily in the HEX index and the company’s fluctuations made the HEX a less relevant benchmark.

RSI21 Norway.

 

RSI21 Sweden.

 

RSI21 Denmark.

 

RSI21 Finland.

 

Figure 3: Price development the first 66 days following RSI momentum signals in the Nordic markets.

The chart shows average price development following buy and sell signals from RSI21 used as a momentum indicator. The signals are triggered on day 0. Only days when the exchange is open are included, so 66 days equal approximately three months. Buy signals are the blue line and sell signals are the red one. The shaded areas are the standard deviation of the calculations. Benchmark index is the black line.

Average return after 66 days Norway Sweden Denmark Finland Weighted average
Buy signal RSI21 breaks above 70 6.4 % 5.1 % 5.8 % 4.5 % 5.5 %
Sell signal RSI21 breaks below 30 -0.9 % 1.1 % 0.5 % 1.0 % 0.4 %
Benchmark 3.1 % 3.6 % 3.3 % 2.9 % 3.4 %

The results are quite uniform across the four markets. Stocks with buy signal, i.e. those that have risen a lot and have high RSI, continue to rise. During the 3 months following RSI21 breaking above 70, these stocks have on average risen by 5.5 %. Compare benchmark indices that have risen 3.4 % in an average 66 day period.

Stocks with sell signals, that had fallen a lot in a short time, making RSI21 break below 30, have moved sideways in the next 3 months, with an average rise of 0.4 per cent.

Annualized excess return Norway Sweden Denmark Finland Weighted average
Buy signal RSI21 breaks above 70 14.4 %p 6.5 %p 10.8 %p 6.9 %p 9.1 %p
Sell signal RSI21 breaks below 30 -15.8 %p -10.2 %p -11.3 %p -7.5 %p -11.6 %p

%p= percentage point

The table shows annualized excess return in percentage points, called %p. The figure is estimated by repeating the 66 day return four times and deducting the 66 day return of benchmark index four times.

Figure 4: Price development at RSI21 signals in Norway. Same as figure 3, but including price development in the days before the signal was triggered.

Figure 3 clearly shows that stocks that have risen on average continue to rise, and that stocks than have fallen continue their weak performance.

The charts indicate that RSI should be used as a momentum indicator and not as an oversold/overbought indicator. This is the direct opposite of traditional technical analysis literature’s presentation of RSI and contrary to how many investors use RSI.

However, we will see that the results are more nuanced when we also include stock liquidity in the analysis.

Sensitivity analysis

A sensitivity analysis gives an indication of how robust the RSI indicator is as an investment instrument. We want to vary parameters in the formulas to investigate result fluctuations.

We have conducted a sensitivity analysis for the Norwegian and Swedish markets for RSI duration (for instance 14, 21 and 30 days), trigger value (signal at RSI above 60, 70, 80) and manner of signal identification (signal at break to extremes or when RSI reverses from extreme values).

The combined results show that the RSI indicator is robust in terms of RSI duration, trigger value and signal identification.

Results the past five years and over time

Over the past ten-fifteen years, several articles have been published in international journals that conclude that momentum can be a good indicator to predict future excess return from stocks. So-called «quant funds» have also appeared, where investments are made quantitatively based on a number of factors but where different forms of short or long term momentum are also included.

The question is whether the increasing focus on momentum and the historically good results have made more investors position for these effects making the results weaker over time.

Stock markets have developed quite differently in the early 2000's compared to the past few years. We study relative numbers to get a fair comparison. We look at average return for these signals as compared to average return for the benchmark indices.

Relative return after 66 days Full period (1996-2018) Past 3 years (2016-2018) Past 5 years (2014-2018)
Buy signal RSI21 breaks above 70 2.1 %p 1.7 %p 1.9 %p
Sell signal RSI21 breaks below 30 -2.9 %p -0.4 %p -1.1 %p

%p: percentage points difference compared to average benchmark.

Annualized absolute return: Full period (1996-2018) Past 3 years (2016-2018) Past 5 years (2014-2018)
Buy signal RSI21 breaks above 70 22.4 % 13.8 % 15.7 %
Sell signal RSI21 breaks below 30 1.7 % 4.9 % 3.7 %

The top table shows that average excess return for stocks with buy signals was 1.7 percentage points the past five years. These are strong numbers, if a little weaker than the 2.1 percentage points excess return for the whole time period. Annualized the excess return is 7.7 percentage points the past five years, versus 9.2 percentage points for the whole period.

Absolute numbers are also strong. The statistics show an average annualized return of 22.4 per cent for the whole period and 13.8 and 15.7 per cent for the past three and five years. Note that the interest level at the start of this period was significantly higher than today, meaning it is not entirely correct to compare these numbers.

The charts below show the development of signal stocks the past five years in each of the Nordic markets.

RSI21 Norway.

 

RSI21 Sweden.

 

RSI21 Denmark.

 

RSI21 Finland.

 

Figure 5: Price development at buy and sell signals from RSI momentum the past five years in the Nordic countries.

Norway.

 

Sweden.

 

Denmark.

 

Finland.

 

Figure 6: Relative return 66 days after buy signal from RSI in five-year intervals. Note that the first period is either 1996-1996 or 2003-2004 and that the last period if the four years 2015-2018.

The above figure shows that relative return for stocks with momentum signals from RSI vary over time and between markets. This is to be expected and is mostly due to ordinary variations in the markets from year to year. It is hard to see a clear trend in the charts above. Combined results from the Nordic countries, table above, also show small changes.

Note that the numbers are relative. All positive figures in the histograms above show that stocks with buy signals have outperformed index.

Liquidity

Big companies, which tend to have high turnover at the stock exchange, generally get more attention from analysts, investors and media than smaller companies. This means that information about the bigger companies reach the traders faster than it does for the smaller companies. This difference may cause the signal power of RSI, and the chances of excess return by following these signals, to vary between smaller and bigger companies.

We split the companies into two groups of more or less the same size; those with average daily turnover above and below five million krone. For Finland we used 0.5 million euro as the limit. We studied stocks with buy and sell signals from RSI momentum. The results are shown below, with smaller companies on the left and bigger companies on the right.

a) Norway smaller companies.

 

b) Norway bigger companies.

 

a) Sweden smaller companies.

 

b) Sweden bigger companies.

 

a) Denmark smaller companies.

 

b) Denmark bigger companies.

 

a) Finland smaller companies.

 

b) Finland bigger companies.

 

Figure 7: Return for stocks with buy and sell signals from RSI. Stocks with average daily turnover below five million krone on the left and above five million krone on the right.

Stocks with buy signal where RSI21 breaks above 70
Relative return after 66 days Norway Sweden Denmark Finland Weighted average
Smaller companies 5.2 %p 2.3 %p 3.7 %p 2.0 %p 3.2 %p
Bigger companies 1.5 %p 0.6 %p 1.3 %p 1.2 %p 1.0 %p

Stocks with sell signal where RSI21 breaks below 30
Relative return after 66 days Norway Sweden Denmark Finland Weighted average
Smaller companies -3.4 %p -3.0 %p -4.8 %p -2.3 %p -3.2 %p
Bigger companies -4.8 %p -1.8 %p -0.8 %p -1.3 %p -2.5 %p

%p: percentage points

In smaller companies, stocks with buy signal from RSI momentum gave an excess return of 3.2 percentage points vs benchmark index in 66 days. The bigger companies gave an excess return of 1.0 percentage points.

Buy signals from RSI momentum thus appear to have stronger signal power for smaller companies than bigger ones. It looks as though buy signal from RSI momentum is more reliable if the stock has a daily turnover of less than five million krone per day than if they have turnover above five million. However, buy signal from RSI momentum indicates future excess return for bigger companies as well. Average annualized excess return has been 13.8 percentage points at buy signal from RSI momentum for smaller companies and 4.4 percentage points for bigger companies.

The sell signals show more consistent predictive power for low and high liquidity stocks. Smaller companies with sell signal from RSI momentum gave a negative excess return of 3.2 percentage points the next three months. The bigger companies gave a negative excess return of 2.5 percentage points. Average annualized excess return was -13.0 percentage points from sell signal from RSI momentum for smaller companies and -9.9 percentage points for the bigger companies. Thus, sell signals seem to have good signal power for both bigger and smaller companies.

We have different figures for the four Nordic countries, but the relative conditions are quite similar. Consistency across countries strengthens the results: RSI momentum has shown strong predictive power for smaller companies, both for buy and sell signals. RSI momentum has shown moderate predictive power for buy signals from bigger companies and strong predictive power for sell signals from bigger companies.

Short term or long term indicator

We have studied what happens for the three months after RSI reached extreme values of above 70 or below 30. The results indicate that RSI can be a good indicator to time trades for medium long term investors.

However, many investors and traders use RSI first and foremost as a short term indicator, to time their trades in the coming days and weeks.

What happens if we study return for the first 22 days after signal instead?

RSI21 Norway.

 

RSI21 Sweden.

 

RSI21 Denmark.

 

RSI21 Finland.

 

Figure 8: Price development first 22 days after RSI momentum signals in the Nordic countries.

Relative return after 22 days Norway Sweden Denmark Finland Weighted average
Buy signals RSI breaks above 70 1.3 %p 0.8 %p 1.2 %p 0.8 %p 1.0 %p
Sell signals RSI breaks below 30 -1.1 %p -0.4 %p -0.7 %p -0.4 %p -0.6 %p

The charts and table give an annualized excess return of 13.2 percentage points for buy signals, when repeated 12 times in a year. This can be compared with 9.1 percentage points excess return using the 66 day figures repeated four times per year. The sell signals give 8.0 percentage points negative excess return using the 22 day figures vs 11.6 percentage points with the 66 day figures.

Combined the buy signals from RSI have given even better annualized excess return when used as a short term trading strategy. However, sell signals have given slightly lower negative excess return when used short term than medium long term.

Summary

We have studied 35,733 cases where RSI21 breaks above 70 and 23,590 cases where RSI21 breaks below 30. We looked at what happened after these signals were triggered. We have used 23 years of stock exchange data from Norway and 16 years for the other countries. The data distribution is 27 % from Norway, 46 % from Sweden, 13 % from Denmark and 14 % from Finland.

The results are consistent across the four markets and show that stocks with high RSI, above the critical 70 level, have continued to rise. Similarly, stocks with low RSI, below the critical 30 level, continue to underperform compared to benchmark. Excess return three months after buy signal was 2.1 percentage point vs benchmark. Negative excess return three months after sell signals was 2.9 percentage points on average. Annualized this is 9.2 percentage points excess return and 11.7 percentage points negative excess return respectively.

The results are quite consistent over time. The past three and five years’ excess return from buy signals from RSI momentum was 1.7 and 1.8 percentage points respectively, compared to 2.1 percentage points for the whole period.

We split the companies in smaller and bigger companies. The results were stronger for smaller companies than the bigger ones, especially for buy signals from RSI momentum. Smaller companies gave 3.2 percentage points average excess return in three months, vs 1.0 percentage points for the bigger companies.

The results above are for the medium long term, 66 days, after RSI21 signals. Many investors use RSI more as a short term indicator. We also looked at return after 22 days. Annualized return for buy signals were 13.2 per cent for 22 day figures vs 9.1 percentage points for 66 day figures. This indicates that RSI21 can be a good indicator for medium long term investments, and even better for shorter term trading.

Combined the results show that RSI is well suited as a momentum indicator and that investments based on buy signals from RSI momentum give statistically better return than average benchmark.

 

We conducted the same study on the Indian market and the results show that buy signals from RSI momentum have given an annualized excess return of 9 percentage points in India. Read the report here.

 

Keywords: Buy signal,Copenhagen,h_RsiHigh,h_RsiLow,help topic main report,Helsingfors,Momentum,Oslo,Overbought,Oversold,RSI,Sell signal,statistics,Stockholm.

Written by

Geir Linløkken
Head of Research and Analysis
at Investtech

"Investtech analyses the psychology of the market and gives concrete trading suggestions every day."

Espen Grønstad
Partner & Senior Advisor - Investtech
 


Investtech guarantees neither the entirety nor accuracy of the analyses. Any consequent exposure related to the advice / signals which emerge in the analyses is completely and entirely at the investors own expense and risk. Investtech is not responsible for any loss, either directly or indirectly, which arises as a result of the use of Investtechs analyses. Details of any arising conflicts of interest will always appear in the investment recommendations. Further information about Investtechs analyses can be found here disclaimer. The content provided by Investtech.com is NOT SEC or FSA regulated and is therefore not intended for US or UK consumers.


Investtech guarantees neither the entirety nor accuracy of the analyses. Any consequent exposure related to the advice / signals which emerge in the analyses is completely and entirely at the investors own expense and risk. Investtech is not responsible for any loss, either directly or indirectly, which arises as a result of the use of Investtechs analyses. Details of any arising conflicts of interest will always appear in the investment recommendations. Further information about Investtechs analyses can be found here disclaimer. The content provided by Investtech.com is NOT SEC or FSA regulated and is therefore not intended for US or UK consumers.

Titlex

OK