Trading Regime Analysis Using RSI

[Do Or Die 10]

Hi,

 

This article is a build up on the previous one- Trend Following Vs Mean Reversion: Trading Regimes. I will now try to clarify the meaning of Trading Regimes using RSI as an example, before we move to more sophisticated methods. This is a discretionary method, but conceptually strong, and can serve as an intro to quantitative methods. We will gradually shift to the crux, "Big moves are often preceded by contraction,and followed by expansion in volatility, this works across all time frames." There is also significant academic research on regime shifting, and you can find papers on SSRN. Two other relevant links are Relative Strength - Internal and Introduction to Understanding Volatility

 

Most trading strategies can be categorized into two different camps: Trend Following and Mean Reverting.

 

An example of trend following is buying when prices cross above a moving average. For trend following systems or strategies, simply buying and holding the market has it's benefits. First and foremost - It's simple and when markets catch a trend, it can run and run. However, Trend following relies on directional moves. Your strategy will lose if the market is sideways; and, the markets CAN remain in a sideways mode until you go bonkers. You can make a trend following strategy howsoever complicated using MACD, ADX, Parabolic SAR. But if the strategy is "trend following" it will not perform in sideways market.

An example of mean reverting will be- buying when a oscillator tuns upwards from a oversold stage. Mean reverting strategies relies on the assumption that market cannot continue in one direction for prolonged time; it will "revert" back to a theoretical mean. The "mean" is always shifting ofcourse. The moving average is a good example of a mean around which prices move around. Trading using price bands (eg. bollinger bands) and channels is also an example of a mean reverting strategy. These strategies lose money in trending markets.

The aim of these articles is to establish a method of switching between strategies as they blow hot and cold, so you don't continue to follow the strategies that aren't adapting to current market conditions.

Shifting Strategies using RSI

 

• A sharp and extended move to either the overbought or oversold area signifies the start of a directional move. Refer to Mega Overbought stage at Relative Strength - Internal. The opposite conditions qualify for a Mega Oversold stage which indicates the start of a new downtrend.
  
• An uptrend is marked by the fact that RSI will not reach the lower parts of its range. RSI typically moves in the range >40 in an uptrend while touching >70 area often.
  
• A downtrend is marked by RSI does not reaches the upper part of its range. In a downtrend it will tend to remain below 60 and move toward <30 area often.
  
• A sideways trend is marked by RSI moving >70 and <30 area.
  

 

When the market is trending, the Moving Average Crossover with price is used to generate signals. When the market is sideways, the oscillator turning up (down) from a oversold (overbought) stage is used to generate signals.

 

 

Note the period where all signals are discarded, whether they come from a trend following or a mean reverting method. This is because the regime shift is not clear, as per the definition above. From a historical perspective it looks more like a sideways market and should have given money from a mean reverting strategy.

 

The improvement by switching strategies should be very apparent in this example. The productivity of your trading system/methodology can improve dramatically by including a 'regime analysis' or 'adaptive layer' on top of strategies. I currently use a custom indicator for regime analysis, however, it can be achieved through many techniques.

 

Please post a comment- they encourage me to provide better examples.

[Tams 20]

thanks for the thread.

interesting observations, I will study them this weekend.

[Predictor 11]

Interesting application of RSI.. thanks

[carltonp 10]

Very interesting indeed.

 

I will also be studying it this weekend. I will provide feedback.

[umfan92 10]

Hi Do or Die,

 

Thank you for the thread. I will also follow along and I know I'll learn a lot.

 

  Do Or Die said:

Please post a comment- they encourage me to provide better examples.

 

Which method do you think would be more successful? Do you think it would be more successful to keep it simple and just use the RSI or are there better indicators to use for mean reversion? Also do you recommend this for a beginner trader like myself?

 

Thank you again for the thread.

[Do Or Die 10]

  umfan92 said:

Which method do you think would be more successful? Do you think it would be more successful to keep it simple and just use the RSI or are there better indicators to use for mean reversion? Also do you recommend this for a beginner trader like myself?

 

Hi Umfan,

 

As I previously mentioned, for beginner traders it is good to understand one technique/indicator before moving to another.

 

The other day I was discussing with Phantom about divergences- most oscillators behave in same way. You do not need to research each of them to learn about trading overbot/oversold levels. Studying one will be sufficient. You can stick to only one (say RSI) http://www.traderslaboratory.com/forums/technical-analysis/10120-relative-strength-internal.html Once you are fully comfortable with it, you can tweak them to suit your trading style (example Trading Regime Analysis Using RWI

[zdo 20]

Do or Die,

Why RSI length of 10? Thanks.

[Tams 20]

Do or Die,

Have you tried a length of 2?

Thanks

[zdo 20]

  Tams said:

Do or Die,

Have you tried a length of 2?

Thanks

 

 

:rofl:

[Do Or Die 10]

  zdo said:

Do or Die,

Why RSI length of 10? Thanks.

 

Same logic: http://www.traderslaboratory.com/forums/tech-analysis/10448-trading-regime-analysis-using-rwi.html#post124510

 

  Tams said:

Do or Die,

Have you tried a length of 2?

Thanks

 

Same logic :haha: http://www.traderslaboratory.com/forums/technical-analysis/10120-relative-strength-internal-2.html#post122239

[Kojak 10]

This looks to be interesting and one look forward to the more sophisticated techniques in identifying "regime change".

 

However on the basis of the examples that you gave, it seems that it identifying a trend is easier than identifying a mean reverting period objectively. Visually it may be quite easy. Bouncing off the +70 and the -30 range is not easily identifiable. Or by the time you identify it, the regime has played itself out.

 

Waiting to see the "pure technical analysis" way of identifying a side-wise moving regime from you, instead of the techniques like co-integration etc which are more in the realm of statistical arbitrage.

 

Jose Kollamkulam.

Chennai

[Do Or Die 10]

  Kojak said:

...

However on the basis of the examples that you gave, it seems that it identifying a trend is easier than identifying a mean reverting period objectively. Visually it may be quite easy. Bouncing off the +70 and the -30 range is not easily identifiable. Or by the time you identify it, the regime has played itself out.

This is a discretionary method... and like all other discretionary pattern methods it can be subject to hindsight bias and availability heuristics. This method will be helpful to only traders who use RSI regularly in their trading. (refer Relative Strength - Internal )

 

Similarly for people who trade using chart patterns will be able to better relate Trading Regime Analysis Using Chart Patterns- Part 1 and Part 2.

 

If you're using any of techniques talked about on internet (versus using custom indicators), and wish to tell me about it, I can give a better contextual reply.

[Do Or Die 10]

During bull markets, all indicators tend to trade with a bias to the high end of the range, while in bear markets indicators tend to shift to new, lower extremes. This is what I call indicator scale shift, and where truly major market moves are involved, it will happen.

 

In addition to gauges of breadth and volume, we also see the scale shift phenomenon at work in more classic gauges like the nine-day Wilder Relative Strength Index (RSI), which is a common momentum gauge. The classic RSI scale suggests that values above +70 are overbought and readings below +30 are oversold. In reality, in strong uptrending bull market moves, the nine-day RSI will often move up to record readings in the +80 area and end up a subsequent correction in the low +40 zone. This upward scale shift away from the traditional +70 and +30, to +80 and +40 parameters, is the bull market “80/40” rule for the nine-day RSI. Alternatively, in bear markets, we often experience a downward scale shift to parameters of +60 overbought and +20 oversold. This is the bear market “60/20” rule for the nine-day RSI.

 

--- Frank Barbera

[Do Or Die 10]

Here is most of the discussion plotted as an indicator.

 

 

It has a lag of 50 bars; but even with that it gives an edge if one takes the regimes into consideration for trading.

[Do Or Die 10]

  Kojak said:

Visually it may be quite easy. Bouncing off the +70 and the -30 range is not easily identifiable. Or by the time you identify it, the regime has played itself out.

 

Its rather easier to code the same thing without any future looking formula; I just did. Compare the char above with the one posted in first post. Works satisfactory even with a lag of 50 bars; though I would prefer going discretionary with this method.

Edited November 17, 2011 by Do Or Die

[Do Or Die 10]

I separated the above Trending component in bear and bull parts; now the results are even better. It works good with lookback period (lag) of 30 bars.