Indicator Redundancy: When More Indicators Means Worse Results

Why Redundancy Feels Like Confirmation

Human brains are wired to seek confirmation. Three green lights feel safer than one. So traders stack indicators: RSI below 30 AND Stochastic below 20 AND CCI below −100 = “triple oversold confirmation.”

The problem? RSI, Stochastic, and CCI all derive from the same input: recent price changes over a lookback period. When RSI says oversold, Stochastic almost always agrees — because they're mathematically correlated at 0.85–0.95 depending on settings.

True confirmation requires different DATA sources: price direction (trend) + trading volume (participation) + price range (volatility). These are genuinely different perspectives on market behavior. Three momentum oscillators are the same perspective repeated in three different fonts.

Indicator Categories — Pick One From Each

A strategy with one indicator from each category (4 total) covers more analytical ground than a strategy with 4 momentum indicators. The first provides diverse information. The second provides the same information four times.

The Correlation Matrix: Proof of Redundancy

Numbers make the case more clearly than intuition. A typical correlation matrix from BTC/USDT 4H data (2021–2025) shows how often indicator pairs agree:

RSI, Stochastic, and CCI show correlations of 0.84–0.91 — they move together nearly all the time. Meanwhile, EMA, OBV, and ATR show correlations below 0.50 with momentum indicators. Combining RSI + OBV + ATR gives you three distinct data dimensions. Combining RSI + Stochastic + CCI gives you the same dimension measured three times.

The Three Costs of Indicator Redundancy

1. Fewer signals, not better signals. Three momentum indicators must ALL agree. This dramatically reduces entry count — from 200 trades to maybe 40. But the 40 aren't higher quality. They're just the subset where three correlated tools happened to align on timing. You've filtered out valid trades without improving win rate.

2. Overfitting amplification. Each indicator adds parameters: RSI period (14? 21? 7?), Stochastic %K period (5? 14?), CCI period (20? 14?). Three indicators = 3+ tunable parameters measuring the same thing. Your optimizer finds the magic combination that works on historical data — and fails on new data. Pure curve-fitting disguised as strategy development.

3. Analysis paralysis in live trading. RSI says buy. Stochastic says neutral. CCI says sell. Now what? You freeze. Miss the trade. Or worse — override the system because “two out of three” seems good enough. Conflicting redundant signals destroy discipline.

“Simplicity is the ultimate sophistication.” — Leonardo da Vinci (and every profitable systematic trader who figured this out)

Building Non-Redundant Indicator Sets

Pick one indicator per category. Each should answer a different question. Three example sets that maximize information diversity:

Trend-Following Set: EMA crossover (trend direction) + RSI (momentum confirmation) + ATR (volatility-based sizing). Three indicators, three distinct data dimensions, zero redundancy.

Mean-Reversion Set: Bollinger Bands (volatility extremes) + RSI (oversold/overbought) + OBV (volume divergence). BB defines the setup, RSI confirms the extreme, OBV verifies fading selling pressure.

Breakout Set: Donchian Channel (range breakout) + ADX (trend strength) + CMF (money flow). Donchian signals the breakout, ADX confirms strengthening trend, CMF validates institutional participation.

Testing for Multicollinearity in Your Strategy

If you already have a multi-indicator strategy and suspect redundancy, run this diagnostic on StratBase.ai:

Step 1 — Baseline. Run your full strategy and record profit factor, win rate, and trade count.

Step 2 — Remove one indicator at a time. Disable each indicator individually and re-run. If removing an indicator changes trade count by less than 10% and profit factor by less than 0.05, that indicator is redundant.

Step 3 — Replace, don't just remove. After identifying redundant indicators, replace them with indicators from under-represented categories. Swap the redundant Stochastic for OBV (volume) or ATR (volatility). The replacement should increase trade count while maintaining or improving profit factor.

How to Test for Redundancy

Run these backtests on StratBase.ai:

Step 1: Strategy A = Trend (EMA cross) + Momentum (RSI). Note the Profit Factor, Sharpe, and trade count.

Step 2: Strategy B = A + second momentum (Stochastic). Compare metrics. If PF improved less than 0.1 and trade count dropped by 30%+ — Stochastic is redundant. Remove it.

Step 3: Strategy C = A + Volume (OBV filter). Compare. Volume is a DIFFERENT data dimension — it may actually improve results.

This “additive testing” approach reveals which indicators contribute unique information vs which just echo each other.

Real-World Example: BTC/USDT Daily

The redundant strategy killed 86% of trades for zero improvement. The diverse strategy lost some trades but genuinely improved win rate and PF by adding a volume perspective. That's the difference between redundancy and confirmation.

The Golden Rule

Build strategies on StratBase.ai with ONE indicator per category. Maximum 3–4 indicators total. Test each addition: does it improve Profit Factor by at least 0.1? Does it maintain at least 50 trades? If yes — keep it. If the improvement is marginal — simpler is better. Always.

Further Reading

• RSI on Investopedia
• Bollinger Bands on Investopedia
• Sharpe Ratio on Investopedia