Funny question. What is the difference between an elephant and a hippopotamus?

Elephant and hippopotamus are both mammals. They belong to the same category of animals.

In a similar way RSI and CCI are both normalized momentum oscillators. They belong to the same category of indicators.

They show you the relative momentum over the selected lookback period, where the relative momentum is being processed through a mathematical formula.

A momentum oscillator has a negative value when the momentum points down, it has a positive value when the momentum points up.

A normalized momentum oscillator uses a scale which does neither depend on the instrument nor on the time frame.

A well coded normalized momentum oscillator should pass the coherency test (known as C-test) suggested my trader and mathematician William Eckhard.

In fact, both RSI and CCI pass the C-test without any problems, see chart below.

Now let us come to the differences.

The RSI uses a scale from 0 to 100. There is no value possible outside this scale. Mathematically this is achieved by summing up the close-to-close changes for positive days and for negative days. The sum of the positives is then divided by the total sum of positives and negatives. The sum of the positives is always a fraction between 0 and 1 of the total sum, therefore the indicator will always show a value between 0 and 100.

The CCI looks at the distance of price from a moving average calculated from the typical price, and then divides that figure by the mean absolute deviation of the data points over the lookback period. Afterwards the result is multiplied with a scaling factor of 1/0.015. This mathematical formula does not impose an absolute scale, but the division by the mean absolute deviation is good enough to make the indicator scale independent from selected instrument and time frame, and it therefore passes the C-test. Let us assume that the lookback period of the CCI is 14. In that case the distance of price from the SMA cannot be larger than 14 times the mean absolute deviation, and therefore the theoretical maximum that the CCI may take is 14/0.015 = 933.33. The scale is indeed limited, but the maximum and minimum values depend on the lookback period. However, most of the time the CCI will not go above 250 or below - 250.

The CCI is defined in a very strange way. The first thing that is astounding is that it does not used the standard deviation but the mean absolute deviation for normalizing the values. The second strange thing is the arbitrary scaling factor 1/0.015 = 66.67, which has been selected to improve the readability of the results.

It is much easier to use the z score (z stands for the normal distribution) instead of the CCI. The z score also measures the distance of a data point from the mean, but then divides it by the standard deivation of the population. If any scaling factor is added it should be 100. This would also make it easy to detect 3-sigma, 4-sigma or 5-sigma events, which have a known probability for the normal distribution.

Attached is a chart comparing Z-Score to CCI. It shows that both indicators pass the C-test, but that the z score has a superior ability to handle outliers, as it has an ability to dampen the impact of extreme values.

The following 11 users say Thank You to Fat Tails for this post:

Platform: "I trade, therefore, I AM!"; Theme Song: "Atomic Dog!"

Trading: EMD, 6J, ZB

Posts: 798 since Oct 2009

Williams RSI had been color coded, for well over 10+ years, and was deadly accurate to a fault, until too many traders started using it, and too many analysts started making it more publically aware.

CCI was made widely popular by the likes of Ken Wood, popularly known as Woodies' CCI.

Given the exposure, customization and moving away from center, or what the momentum indicators originally were designed for, @Fattails, how would you comments on how these two indicators have been used, with respect to which is more effective to utilize, or more consistent in reliable trading results?

From what I have learned, the CCI is far more accurate on long term charts and almost worthless on short term, intra-hour charts. Equally, from what I have learned, the RSI is highly useful and remains so, with intra-hour charts, and only coincidently accurate on longer term charts.

Wilder will certainly disagree with that statement. Infact the first sentence of his book with regards to the Relative strength index says: Rsi is a tool which can add a new dimension to chart interpretation when plotted in conjunction with a daily bar chart.

@kronie: I do not know any Williams RSI. There is a Williams %R which is a variation of the Stochastics. It just uses a different scale and is otherwise identical. The scale of the Stochastics is more convenient, therefore I have never used the Williams %R.

None of the oscillators is accurate, none of them is worthless. I do not see any particular reason that a CCI would work better on daily charts compared to a RSI or vice-versa.

The following 2 users say Thank You to Fat Tails for this post:

Platform: "I trade, therefore, I AM!"; Theme Song: "Atomic Dog!"

Trading: EMD, 6J, ZB

Posts: 798 since Oct 2009

I stand corrected, and no shame or embarrassment warranted either. Williams %R, thanks for clarifying that.

I find it interesting that oscillators, are accurate or worthless. What did you mean by that? Generally speaking the type of intraday trading that most of us discuss and need on these threads, are short term in nature, hence we curve fit towards the sub-one-minute (whether in renko, vol-bars, candles or otherwise) to the sub-hour intervals, hence the need for oscillators and momentum indicators that are sensitive enough to predict or report accurately (as close to realtime as possible, with minimal lag), what the smoothed average of prices are doing, or have just completed.

When doing longer-term analysis, beyond an hour, and towards the four hour blocks and daily / weekly bars, just about every indicator (which mathematically represent some smoothed form of an rolling average sampling) will be as accurate as that interval in predictive analysis. Hence, when a crossover of moving average lines on a 4hour bar shows, then its likely that further separation from those crossing points will continue in same direction in the next 4hour bar. Even when that is not the case, at least the person reading the chart is able to do a 50-50 analysis of what might occur next.

I see the word "Accuracy" batted around in this thread and I think you should clarify what is meant when you use such a term.

Indicators... to me anyway means "showing signs of"... not "this is a message from above".... there is no silver bullet when it comes to indicators...be it RSI, CCI or others.

Sometimes individually they show the way to riches.... sometimes they lead you down the garden path.

I use Indicators a lot. If you check out my journals you will find indicators form the cornerstone of my trading strategy. But do I make my decisions based on a single indicator....not on your life. They will all give misleading information or sometimes outright false directions....But if you choose a suite of indicators and draw a consensus from their indications...then you can get a pretty reasonable idea about what may happen.

All indicators are misleading....but they are not so insync with eachother... arriving at a consensus is possible.

This is also my objection to trading platforms that do the analysis for you.... the idiot light systems. I smile when as I heard on one presentation when they ran down indicators because they represent the past...but they have a forecasting system... based on what??? Historical data... is that not the past??? My objection to those systems is that you do not know (and they won't tell you) how those indicators (yes they use them) are interpreted.... usually they are based on nothing more than traditional settings IMHO.

And another thing.... let us say we have two RSI readings on separate charts. One chart has a RSI of 45 and another chart has a RSI of 55.... which RSI is the most bullish????

Answer: You don't know without more information about the RSI readings. The chart with an RSI of 45 could be the most bullish if the trend of the RSI was positive (the greater the +ve slope of the trend the more bullish it is becoming) and the chart with the RSI of 55 could be VERY BEARISH if the trend of that chart's RSI is negative (again the higher the -ve slope the more bearish it will be)

How many times have you heard a stock is "over-sold" and less than 30.... this makes it a buy??? Hockey pucks NO!!!... sure it might turn around and climb out of the mud .... but it might take months to do so. Similarly, the RSI is "over-bought" and over 70.... time to sell??? You are foolish IMHO if you do... it could remain there for months and you would lose a lot of profit potential.

Paying attention to only single numbers for any indicator is a recipe for losing money. At the very least, pay attention to the trend of that indicator, no matter what it is.

You really up your game when you start developing a consensus with a suite of non-similar indicators (eg. don't choose TRIX and MACD as they give you a similar look).... only then will you find indicators really useful.

There is no such thing as "accuracy" with indicators...it has no meaning... all indicators are accurate as far as the number that is calculated... most formulae are pretty simple but that means nothing in the interpretation of them.