Pro Access

Connors RSI Reversion is part of the expanded Pro strategy library. The guide stays public, but running it in backtest, compare, paper tracking, and source-code view is a Pro capability. Pro currently unlocks 8 additional pre-built strategies.

The Intuition

Connors RSI is a deliberately short-horizon composite indicator. Instead of relying on one oscillator, it averages three pieces of evidence: a very fast RSI on price, an RSI on the current up/down streak length, and the percentile rank of the latest one-day return inside a longer sample.

The motivation is practical. Short-term reversals are usually not explained by one feature alone. You want to know whether price is stretched, whether the move has persisted for several consecutive days, and whether the latest return is unusually large relative to recent history. Connors RSI packages those three checks into one score.

That makes it more nuanced than a plain RSI threshold. A market can be oversold on price but not especially unusual in streak or return percentile terms. Connors RSI tends to fire only when several kinds of short-term pressure line up at once.

The weakness is fragility. Because it combines multiple components, the signal is more parameter-sensitive than a basic oscillator. It can also overtrade during stressed markets where "extreme" short-term moves are no longer rare but simply the new regime.

The Math

Read this as a compact model summary: what the signal sees, what it ignores, and where fragility can creep in.

RSI_price(t)   = RSI(Close, rsi_period)
streak(t)      = consecutive up/down closes with sign
RSI_streak(t)  = RSI(streak, streak_period)
PctRank(t)     = percentile rank of 1-day return within the last rank_window days

CRSI(t) = (RSI_price(t) + RSI_streak(t) + PctRank(t)) / 3

Signal(t) = +1  if CRSI(t) < oversold
          = -1  if CRSI(t) > overbought
          =  0  otherwise

Parameters

ParameterTypeDefaultDescription
rsi_period int 3 RSI period applied to closing prices
streak_period int 2 RSI period applied to up/down streak length
rank_window int 100 Window used for one-day return percentile rank
oversold float 20.0 Connors RSI threshold for long entries
overbought float 80.0 Connors RSI threshold for short entries

Source Code

Source access for this built-in strategy is included with Pro.

Further Reading

  • Connors, L. & Alvarez, C. (2009). Short Term Trading Strategies That Work. TradingMarkets Publishing.
  • Connors Research (2012). Connors RSI indicator notes.
  • Chan, E. (2013). Algorithmic Trading. Wiley.

When It Works / When It Fails

Works
  • Range-bound equity markets with regular short-term reversals
  • Short-term reversal setups in liquid instruments
  • When single-component RSI produces too many false extremes
Fails
  • Strong sustained trends — composite pins at extremes
  • High-volatility regimes where all three components diverge
  • When the three CRSI components send conflicting signals

Regime Fit

Transition / Calm vol
Best conditions. All three CRSI components agree; composite extreme readings are reliable.
Transition / Normal vol
Workable. Composite still useful though components may diverge slightly more.
Bull or Bear / Calm vol
Borderline. Directional drift creates persistent component bias; reversal quality degrades.
Any regime / Stressed vol
Avoid. Components diverge significantly; composite breaks down and signal is unreliable.

Compared to Alternatives

vs RSI
Standard RSI uses one component; CRSI composites price RSI, streak RSI, and return percentile rank. CRSI has fewer false extremes but is harder to interpret.
vs Williams %R
Williams %R uses only range position; CRSI incorporates return ranking in the distribution plus a streak component. CRSI is more elaborate but %R is instantly readable.
vs Stochastic
Stochastic %K is a simple range oscillator; CRSI adds streak RSI and return percentile rank on top. CRSI is more complex but potentially more signal-selective.
Unlock This Strategy →