Keltner Channel

Both are volatility channels, but they measure volatility differently. Keltner uses ATR (Average True Range) which produces smoother, more consistent bands. Bollinger uses standard deviation which makes bands more reactive and 'spiky.' Keltner is generally better for trend trading, Bollinger for volatility plays.

Both are valid. Breakouts have lower win rate (~40%) but capture larger moves. Pullbacks have higher win rate (~55%) but smaller wins. Beginners often find pullbacks easier psychologically (buying on dips feels safer). Choose based on your personality and risk tolerance.

Daily is best for most traders - clear signals, manageable frequency, less noise than intraday. Weekly works for position traders wanting fewer, larger trades. 4-hour for active traders. Start with daily and only move to shorter timeframes with experience.

Add filters: ADX > 25 (trending market), volume > 1.5× average (conviction), and wait for daily close (not intraday poke). No filter catches every false breakout, but these significantly improve odds. Accept some failures as cost of catching real breakouts.

Common options: 1) Below middle EMA (moderate), 2) Below lower band (wide), 3) 1.5-2× ATR below entry (volatility-adjusted). The ATR-based stop is most consistent for position sizing. Tighter stops have more failures but better risk/reward when right.

Plot both Keltner Channel and Bollinger Bands on the same chart. When Bollinger Bands contract INSIDE Keltner Channel (BB upper < Keltner upper AND BB lower > Keltner lower), you have a squeeze. This indicates very low volatility, often followed by a significant move when volatility expands.

Generally, keep 20 EMA / 10 ATR / 2× multiplier for consistency. Highly volatile stocks might use 2.5× multiplier for wider bands. Less volatile stocks might use 1.5× for tighter bands. Always backtest before changing defaults. The standard settings are robust across most instruments.

For breakouts: Only enter when ADX > 25 AND price breaks the band. For pullbacks: Still prefer ADX > 25, but can be more flexible. For mean reversion: Only fade bands when ADX < 20. ADX direction matters too - rising ADX confirms momentum, falling ADX warns of weakness.

Option 1: Trail below middle EMA - simple, gives room for pullbacks. Option 2: Trail with ATR - maintain 1.5× ATR below highest high. Option 3: Combination - use whichever is tighter. Most traders use the middle EMA for simplicity and because it naturally trails with the trend.

Weekly Keltner sets the bias (above weekly EMA = long bias only). Daily Keltner provides entry timing (breakout or pullback). Only take daily signals in the direction of weekly trend. Exit on daily signal but also watch for weekly trend change as major exit trigger.

Create inner channel (EMA ± 1× ATR) and outer channel (EMA ± 2.5× ATR). Entry: When price reaches outer band, then retreats to inner band, enter in original direction. This catches the 'exhaustion and reversion' pattern while staying with the trend. Exit at middle EMA or opposite outer band.

Squeeze (narrow channels) = buy volatility (straddles, strangles) expecting expansion. Wide channels = sell premium (iron condors) expecting contraction. Breakout = directional spreads in breakout direction. Pullback to EMA = better entry for directional spreads with tighter defined risk.

Score 0-10: Breakout signal (+2), ADX > 25 (+2), Volume > 1.5× (+1) or > 2× (+2), RSI confirms (+1), Above 50 EMA (+2), Squeeze release (+1). Minimum threshold of 6 to enter. Rank all qualifying signals, enter highest scores first. Track score vs outcome to refine.

Track rolling 2-year profit factor. Compare trending vs ranging regime performance over time. If both are declining while having normal trending periods, edge may be fading. Check if parameters need adjustment or if additional filters are needed. Review correlation with other systematic strategies.

Research shows 20 EMA / 10 ATR / 2.0 multiplier is robust and not overfit. Small variations (18-22 EMA, 1.8-2.2 multiplier) produce similar results. If small changes dramatically alter backtests, the system may be overfit. The standard parameters work well because they're based on meaningful time periods (roughly one trading month).