ConnorsRSI explained

What ConnorsRSI measures

ConnorsRSI averages three 0–100 measures into one short-term overbought/oversold reading. It was published by Connors Research in 2012 with the default ConnorsRSI(3, 2, 100): a 3-period RSI of the close, a 2-period RSI of the up/down streak, and the percent rank of today's one-day return among the previous 100. Together they describe how far price has moved, for how many days in one direction, and how unusual today's move is.

How ConnorsRSI readings are interpreted

Low readings mean recent closes have been weak, a down streak has lasted, and today's return ranks low against the last 100 days; high readings mean the opposite. The guidebook describes 5-day returns rising as ConnorsRSI falls below 20. All three components react within days, so ConnorsRSI moves much faster than a 14-period RSI and reaches its extremes often.

Conventional ConnorsRSI levels

• Below 10: conventionally described as oversold (below 5 for a stricter reading)
• Above 90: conventionally described as overbought (above 95 for a stricter reading)
• Streak: +1, +2, … for consecutive higher closes; −1, −2, … for lower closes; 0 after an unchanged close
• Percent rank: the share of the previous 100 one-day returns that are below today's

Where ConnorsRSI works, and where it does not

ConnorsRSI was designed for short-term mean reversion on daily bars of stocks and ETFs, where a deep oversold reading inside a longer uptrend is the classic setup. In a strong, persistent decline the reading can stay low for many days while price keeps falling.

Indicators commonly read alongside ConnorsRSI

• 200-day moving average: Connors' published rules read oversold signals only above it
• RSI(2): the same short-term idea from the same author, built from the close alone
• A close back above a short moving average (for example 5 days) is the exit commonly paired with it

Known limitations of ConnorsRSI

• Large gaps and news days push the percent rank to an extreme in a single bar
• The 100-day rank needs 101 bars of history before the first value appears
• Readings near 0 or 100 are frequent; they describe the last few days and carry no forecast by themselves

ConnorsRSI in practice

• GU Analyser computes ConnorsRSI(3, 2, 100) by default; the three periods can be changed in a custom signal
• Both RSI components use Wilder's smoothing, the same calculation as the app's RSI
• The percent rank counts returns strictly below today's; an equal return counts as not below
• Commonly used on daily bars; on intraday bars the 100-bar rank covers far less than five months