**The next indicator that we will talk about is “Relative Strength Index”, also known as RSI. This indicator has many features in common with Stochastic, but is even simpler, because it consists of just a single line. This indicator is also scaled in the range of 0-100%. The major purpose of this indicator, at least how traders apply it – is for estimation of overbought/oversold conditions. RSI also has adjustable extreme areas – below 30% and above 70%. But you can appoint any area that you want – say, 25/75 or 15/85 etc.**

Commander in Pips:Commander in Pips:

**Ok, I see, and what the difference with Stochastic then, except the visual appearance?**

*Pipruit:***It has different math as a basis for how it works. It makes RSI more suitable for estimating overbought/oversold conditions.**

Commander in Pips:Commander in Pips:

**If you do not want to learn the math of RSI – you can skip it.**If you want to know it, then look:

****

RSI = 100-(100/(1+RS)), where

RSI = 100-(100/(1+RS)), where

**RS**= UpC/DownC;

**UpC**= moving average of n-period gains;

**DownC**= absolute value (that is, ignoring sign) of the smoothed moving average of n-period losses;

**Well, sounds too complicated…**

*Pipruit:***I see, let’s look how it works on simple example then:**

*Commander in Pips:*Assume that we calculate 14-period RSI. And during this 14 periods price has changed as follows, compares to preceding trading session, in pips:

**-0.0037, +0.0015, +0.0068, -0.0092, +0.0069,- 0.0029, -0.0001, +0.0060, +0.0056, -0.0031, +0.0029, -0.0009, -0.0001, +0.0023**.

So, we have 14 numbers for our calculation…

**Wow, wow, wait a minute Commander. Say, what does -0.0037 pips mean, in the beginning?**

*Pipruit:***It means that in this day (and it was 14 days ago) close price has declined from the previous close for 37 pips. On the next day it has closed 15 pips higher, on the next day 68 pips higher, and so on, until we will come to today trading session that has closed 23 pips higher, then yesterday’s one. Is it clear?**

*Commander in Pips:***Oh, yes. Now it is…**

*Pipruit:***Now we should split these numbers into two groups: one group should contain only upward changes, while another one only downward changes during recent 14 trading sessions:**

*Commander in Pips:***Upward changes: 0; +0.0015; +0.0068; 0; +0.0069; 0; 0; +0.0060; +0.0056; 0; +0.0029; 0; 0; +0.0023**

**Downward changes: -0.0037; 0; 0; -0.0092; 0; -0.0029; -0.0001; 0; 0; -0.0031; 0; -0.009; -0.0001; 0.**

**And where do you find so many zeros in our array?**

*Pipruit:***Nowhere. The point is that each array should have 14 numbers, but periods with opposite change replace by zero. For instance – see, the first change -0.0037, it negative and you can see it only in downward array. But it replaced by zero in upward array. The same is for any negative number – it replaced by zero in upward array. In downward array each positive number, in turn, is replaced by zero also.**

*Commander in Pips:***Oh, I’ve got it, and what then?**

*Pipruit:***Now we should calculate average value of each array. I hope you remember how to do that, so this is task for you. But we have to apply absolute changes – don’t pay attention to “+” or “-".**

*Commander in Pips:***Ok, That’s simple:**

*Pipruit:***= Sum (0; +0.0015; +0.0068; 0; +0.0069; 0; 0; +0.0060; +0.0056; 0; +0.0029; 0; 0; +0.0023)/14 = 0.0320/14 = 22.86 pips**

Upward changes average

Upward changes average

**Downward changes**= Sum (-0.0037; 0; 0; -0.0092; 0; -0.0029; -0.0001; 0; 0; -0.0031; 0; -0.009; -0.0001; 0) = -0.0200/14=-14.29 pips. But, since we need absolute value, so it will be just

**14.29**

**Great. These values are our**

*Commander in Pips:***UpC**and

**DownC**. Now you can get RSI number – just put that you’ve calculated in RSI formula.

**Cool! With pleasure:**

*Pipruit:***So, RS=22.86/14.29 = 1.6**

RSI = 100-(100/(1+1.6))=61.54

RSI = 100-(100/(1+1.6))=61.54

Excellent! But, Commander, how to estimate a new value of RSI when a new closing price will appear?

**Here you can apply two different approaches – calculate the value using exponential averaging method, or simple method. The most common and initial (W. Wilder – creator of RSI) method is exponential. Assume that price was rising during the new trading session for 19 pips. Then, using the exponential way of averaging is done like this:**

*Commander in Pips:***upC=((13 х 22.86)+19)/14=22.58**

downC=((13 х 14.26)+0)/14=13.24

RS=22.58/13.24=1.7

RSI=100 - (100/(1+1.7))=63

downC=((13 х 14.26)+0)/14=13.24

RS=22.58/13.24=1.7

RSI=100 - (100/(1+1.7))=63

If you will apply simple averaging method (Cutler RSI), then you will get a bit different number:

**upC=(15+68+0+69+0+0+60+56+0+29+0+0+23+19)/14=24.21**

downC=(0+0+92+0+29+1+0+0+31+0+9+1+0+0)/14=11.64

RS=24.21/11.64=2.1

RSI=100 - (100/(1+2.1))=68

downC=(0+0+92+0+29+1+0+0+31+0+9+1+0+0)/14=11.64

RS=24.21/11.64=2.1

RSI=100 - (100/(1+2.1))=68

First, the exponential method is the traditional one, and most software programs use this particular method. Anyway, as with moving averages, it’s impossible to say which one is better, because it’s very personal. You will have to estimate it for yourself by exploring RSI. Now, let’s look how it works.

**Application of RSI**

RSI uses for estimation of oversold and overbought conditions, and I might say that in general it makes better work of this than Stochastic, but still - it’s not perfect, at least not for me.

*Chart #1 | EUR/USD Weekly and RSI (9)*

**Commander, and what does overbought or oversold terms mean, in general?**

*Pipruit:***Well, it’s simple. Oversold condition is when there is a high probability that there are no sellers left on the market – sufficient to push the market lower. The same with overbought – there is a high probability that there are insufficient buyers on the market to push it higher. To estimate such conditions, we use different indicators. These conditions do not suggest that the market should definitely reverse, but it should at least turn to a sideways move or see some retracement.**

*Commander in Pips:***Thanks, I see.**

*Pipruit:***But still RSI is not good for me. And this is all because of it’s scaling in the 0-100 range. The same as with Stochastic – take a look in rectangle again. RSI has reached oversold, but the market has continued its move lower. Even more – during this move RSI has risen! Outstanding! Again, if you’ve taken this “Buy” signal from RSI – you will be really hurt. Due to RSI scaling, that follows from its math – if it reaches, say a value of 15%, but market accelerates down further – it has only 15% left for response to it. That’s why RSI shows poor results during a real thrusting market. This is absolutely unacceptable for us. In the meantime, RSI is simple, so you may experiment with it, maybe you will be able to optimize it according with your personal tasks on the market.**

*Commander in Pips:***And could we use it for trend identification?**

*Pipruit:***In general, yes. For that purposes we should use 50% border of RSI. If your analysis tells, that may be new uptrend is forming – check RSI. If it stands above 50% - this is a confirmation sign. For downtrend RSI should stand below 50%. According to RSI formula, it tells that during the recent N periods more than the half close-to-close price changes on market were negative. The same is for uptrend – when RSI is above 50% - more than the half of price changes were positive. This gives a bit more confidence with current trend**

*Commander in Pips:***Thanks. Cool. But when we finally will discuss what we can use for overbought/oversold estimation?**

*Pipruit:***Soon. Now we pass to ADX – Average Directional Move Index, and after that we will return to the topic of overbought/oversold estimation.**

*Commander in Pips:*