This has been taken from Larry Connors ConnorsRSI Pullback Guide Book.
ConnorsRSI combines three components, and as you might guess, they are all elements that our research has repeatedly shown to have significant predictive ability:
Price Momentum: RSI is an excellent way to measure price momentum, i.e. overbought and oversold conditions. By default, ConnorsRSI applies a 3‐period RSI calculation to the daily closing prices of a security. We will refer to this value as RSI(Close,3).
Duration of Up/Down Trend: When the closing price of a security is lower today than it was yesterday, we say that it has “closed down”. If yesterday’s closing price was lower than the previous day’s close, then we have a “streak” of two down close days. Our research has shown that the longer the duration of a down streak, the more the stock price is likely to bounce when it reverts to the mean. Likewise, longer duration up streaks result in larger moves down when the stock mean reverts. In effect, the streak duration is another type of overbought/oversold indicator.
The problem is, the number of days in a streak is theoretically unbounded, though we could probably place some practical limits on it based on past experience. For example, we might observe that there have been very few instances of either an up streak or a down streak lasting for more than 20 days, but that still doesn’t get us to a typical oscillator‐type value that varies between 0 and 100.
The solution is two‐fold. First, when we count the number of days in a streak, we will use positive numbers for an up streak, and negative numbers for a down streak. A quick example will help to illustrate this:
The closing price on Day 2 is higher than on Day 1, so we have a one-day up streak. On Day 3, the price closes higher again, so we have a two-day up streak, i.e. the Streak Duration value is 2. On Day 4, the closing price falls, giving us a one- day down streak. The Streak Duration value is negative (-1) because the price movement is down, not up. The downward trend continues on Days 5 and 6, which our Streak Duration reflects with values of -2 and -3. On Day 7 the closing price is unchanged, so the Streak Duration is set to 0 indicating neither an up close nor a down close. Finally, on Day 8 the closing price rises again, bringing the Streak Duration value back to 1.
The second aspect of the solution is to apply the RSI calculation to the set of Streak Duration values. By default, ConnorsRSI uses a 2period RSI for this part of the calculation, which we denote as RSI(Streak,2). The result is that the longer an up]streak continues, the closer the RSI(Streak,2) value will be to 100. Conversely, the longer that a down]streak continues, the closer the RSI(Streak,2) value will be to 0. Thus, we now have two components – RSI(Close,3) and RSI(Streak,2) – that both use the same 0-100 scale to provide a perspective on the overbought/oversold status of the security we’re evaluating.
Relative Magnitude of Price Change: The final component of ConnorsRSI looks at the size of today’s price change in relation to previous price changes. We do this by using a Percent Rank calculation, which may also be referred to as a ”percentile”. Basically, the Percent Rank value tells us the percentage of values in the look]back period that are less than the current value.
For this calculation, we measure price change not in dollars and cents, but as a percentage of the previous day’s price. This percentage gain or loss is typically referred to as the one]day return. So if yesterday’s closing price was $80.00, and today’s price is $81.60, the one-day return is ($81.60 – $80.00) / $80.00 = 0.02 = 2.0%.
To determine the Percent Rank, we need to establish a look]back period. The Percent Rank value is then the number of values in the look]back period that are less than the current value, divided by the total number of values. For example, if the look]back period is 20 days, then we would compare today’s 2.0% return to the one-day returns from each of the previous 20 days.
Let’s assume that three of those values are less than 2.0%. We would calculate Percent Rank as:
Percent Rank = 3 / 20 = 0.15 = 15%
The default Percent Rank look-back period used for ConnorsRSI is 100, or PercentRank(100). We are comparing today’s return to the previous 100 returns, or about 5 months of price history.
To reiterate, large positive returns will have a Percent Rank closer to 100. Large negative returns will have a Percent Rank closer to 0.
The final ConnorsRSI calculation simply determines the average of the three component values. Thus, using the default input parameters would give us the equation:
ConnorsRSI(3,2,100) = [ RSI(Close,3) + RSI(Streak,2) + PercentRank(100) ] / 3
The result is a very robust indicator that is more effective than any of the three components used individually. In fact, ConnorsRSI also offers some advantages over using all three components together.
When we use multiple indicators to generate an entry or exit signal, we typically set a target value for each one. The signal will only be considered valid when all the indicators exceed the target value. However, by using an average of the three component indicators, ConnorsRSI produces a blending effect that allows a strong value from one indicator to compensate for a slightly weaker value from another component.