Everything I've learned about the RSI

24 พ.ย. 2022

BTCUSDT

In this post, I'll make an attempt to share everything I've learned over the Relative Strength Index (RSI) Over the past 24 months.

Nothing described in this post is financial advice, it's just me, sharing thoughts and ideas with you.

nb: this post is more suited for traders and investors that are already educated about the RSI Indicators.


A brief introduction about the indicator itself :

The relative strength index (RSI) is a momentum indicator used in technical analysis. RSI measures the speed and magnitude of a security's recent price changes to evaluate whether it's better to buy, sell, or wait.
The RSI is displayed as an oscillator (a line graph) on a scale of zero to 100.
The RSI is probably the most used oscillator in finance nowadays, by both retail traders and institutions, hence meaning that when used well, it can be used as a great edge to profitability.

RSI popular uses :

- An asset is usually considered overbought when the RSI is above 70 and oversold when it is below 30.
- The RSI can give us insights on a potential trend's loss of momentum or validity when the price pivots levels are diverging with the RSI indicator (hidden and regular divergences)
- The most popular RSI length is 14 periods.



My findings


1. Overbought and oversold: myth or reality?

RSI's 30 and 70 levels never proved themselves to be a strong enough edge for me to be used as a standalone signal for trade entries.

As an example, just look at the irregularity of the results you would get when using just these zones :


My take on it is that as a price oscillator when it crosses into extremes, it simply means price momentum is at extreme levels. To me it's basically like a mountain cyclist in the middle of a race: he might very well go faster and higher, however, the quicker and higher he goes the more unlikely he is to keep up with that speed. Eventually, he might either decrease its speed or even go backward.


What does this tell us ?

The RSI 30 and 70 levels seem to be better used when used as timing indicators. For example, the 70 and 30 levels could be used as a filter for a trader to eliminate market noise when using a trend reversal strategy (mean-reversion). For trend traders, the levels could be used to timing signals where they'll start looking for price to do a pullback (consolidation) to get in the trend.

My experience using the 30 and 70 levels as exit signals however has been better (when it comes to using it as the only signal for a trade exit).
Say you are long on BTCUSD, in profit, and you get an RSI closure above 70. Well, in that case, you could exit 50% of your position and wait for the oscillator to cross down the 70 levels to exit the rest (as the overbought and oversold zones are rarely a defining factor for trend reversals and corrections).


2. Divergences in the overbought and oversold zones :

The lower the time frame you are trading on is, the higher the noise when it comes to divergences, especially with volatile assets such as BTCUSD. So you might want to filter out most of the ones you see to only take the best ones.
On the 15M and 5M timeframes, on BTCUSD, I find that on average about 1/3 of the divergences I see play out. However, we are not expected to take every divergence we see.

Here's what has helped me get better results with divergences :



- When approaching supply and demand zones, especially the higher timeframe ones, we might want to be more aggressive with the divergences we enter into. As the hit rate is not always amazing, the R:R is usually much better, and if the trade works out, it might give you great results which accounts for the low win rate.
- If you want to increase your win rate, I also find that going for higher timeframes is usually better when it comes to divergences.
- Take only divergences where RSI divergence's first pivot point is over 70 or under 30. Ideally, you don't want the noise to go below 60, or above 40, so that your trade has the necessary momentum to play out.
- For extra confirmation, wait for a break of the noise level to enter the trade.
- Regular and hidden divergences play hand in hand creating a form of momentum equilibrium. Hidden divergences always create regular divergences and vice versa. Hence a hidden divergence can be considered an early pullback warning to get in a bigger-picture trend.
- Regular divergences tend to play out better than hidden divergences. This is especially true when the volume is decreasing, or after a longer period of consolidation when volatility has been contracting and might be about to expand soon.
- Regular divergences in strong trends can be both a disaster and a treat. "The trend is your friend". This saying is especially true here. However, 2-3 drives of regular divergences are a great indication of a potential reversal, with enough confirmation factors to produce (often time) a great entry.
- The angle of the trend line between divergences pivot points, both on the price chart and the RSI, can be a good indication of the severity of the divergence occurring.
- The ideal lookback period for detecting divergences for me has proved to be between 5 and 28 bars. (Below 5 bars is not enough to confirm a true pivot point for me and above 28 bars has probably already played out in past price movements).
- Like all edges, using a divergence strategy always produces better results when used in confluence with other signals. I find the best confluences happen when divergences occur: alongside a stochastic cross, near medium-slow moving averages, near horizontal supply and demand zones, alongside volatility expansion, when the volume is decreasing (meaning market makes are in disagreement with the move occurring), near Bollinger bands 2.5 to 3 standard deviations (period 20).
- Convergence between your timeframes and higher timeframes is key to understanding how to better choose your trades. Try to play the big divergences but enter smaller timeframes divergences.
- When you lose a divergence trade, don't get disappointed. Jump back in because often time, and price will need to do several divergences before getting in your desired direction (however, be careful not to jump in tilt mod. Know your win rate and R:R and keep your money management serious. You'll get blown out if you start tilting on this, especially if you trade reversals with divergences, as it's difficult to get the right timing every time).

3. RSI as a trend filter?

- I've found that in trending markets, when RSI's Exponential Moving Average (EMA) crosses above the 50 line, it's an indication of an uptrend and vice versa. However, this is less effective in ranging markets as there's more noise, hence more invalid crosses.
- I've found that in trending markets when the RSI line crosses above the EMA (I use a 12 period), it's an indication of an uptrend and vice versa. However, this is less effective in ranging markets as there's more noise, hence more invalid crosses.
- As an indication of the trend's direction, I don't find any value in using bullish and bearish control zones. The only use I can find them is when using them for divergence levels filters.


This is the end of the first post of this 2 parts series. There's just so much more you can discover about this indicator that it simply cannot be constricted to a few lines of writing. However, you are welcome to take a few of my findings and go test them out using replay and backtesting. See for yourself, and find your balance.
Most of my learnings have been made through screentime, trial, and error, backtesting, mistakes, and research.

Have a good day,

Arthur Girard