Donchian Channel

Donchian identifies price extremes, but not all extremes lead to new trends. Many breakouts are exhaustion moves (the end of a trend) or occur in ranging markets where price simply oscillates. This is normal - the system accepts many small losses in exchange for catching occasional large trends. The math works when you keep losses small and let winners run.

20-day is more common and produces more signals. It catches shorter trends and is good for swing trading. 55-day is for position traders who want to catch only major, extended trends - fewer signals but potentially larger moves. Many traders use both: 55-day for trend direction, 20-day for entries. Start with 20-day to learn the system.

Yes, but adjust the period to your timeframe. For intraday, use 20-period on hourly charts (20 hours lookback) or adjust accordingly. Be aware that shorter periods on smaller timeframes generate many signals with more noise. The Donchian principle (breakout to new N-period high/low) works on any timeframe.

Donchian uses fixed lookback for highest high/lowest low - bands only move when new extremes are made. Bollinger uses standard deviation around a moving average - bands expand and contract with volatility every bar. Donchian is pure price-based, Bollinger is volatility-based. Both identify breakouts but behave differently.

It depends on your exit rule. With 10-day exit, typical holds are 1-3 weeks. With 20-day opposite band exit, holds can be 3-8 weeks in trending markets. Let your exit rule determine duration - don't exit just because time passed. Some trends last months; others reverse quickly. The exit signal, not calendar time, closes the trade.

The last trade filter skips a breakout if the previous breakout in the same direction was profitable. The logic: if the last long breakout worked, that trend may be exhausted. However, this filter also misses extended trends with multiple profitable breakouts. Modern traders often skip this filter and take all signals, relying on other filters (trend, volume) instead.

This is a false breakout - it happens often. Your stop (usually 2x ATR below entry for longs) should limit the loss. Don't try to trade the reversal unless it triggers your short signal (closes below lower band). Accept the loss, log the trade, and wait for the next valid signal. Multiple false breakouts often precede a real trend.

The Turtles did pyramid - adding at each 0.5 ATR profit increment up to 4 units. This amplifies winners. However, it also increases risk if the trend reverses. If you pyramid, trail your stops on all units. Modern traders often skip pyramiding for simplicity, accepting smaller winners in exchange for simpler management.

Gaps are inherent risk in position trading. Your actual loss may exceed planned stop loss on gap moves. Size positions assuming occasional 2x stop-distance losses (for gaps). After a gap loss, execute at market - don't hold hoping for recovery. Gaps through stops are painful but manageable with proper sizing.

Donchian works on any trending instrument - stocks, ETFs, futures, forex, crypto. The Turtles traded diverse markets. Stocks work well, especially liquid large-caps and ETFs that trend. Avoid very illiquid stocks where breakouts may not have follow-through. The key is the instrument's ability to trend, not what type it is.

Use standard periods (10, 20, 55) that have economic rationale rather than arbitrary optimized values. Test that nearby periods (18-22) all work reasonably. Walk-forward optimization with out-of-sample testing is essential. If optimization shows 23-day beating 20-day by 0.5%, that's noise - stick with 20. Robust systems work across a range of parameters.

Track difference between signal price and actual fill for every trade. Calculate average slippage as % of ATR. If slippage is eating 0.5 ATR per trade on 2 ATR risk, that's significant edge erosion. Mitigate by: using limit orders slightly above breakout level, accepting slower fills, or using breakout-plus-pullback entries where you wait for retest.

Research suggests 20-30 diverse markets provides sufficient diversification for trend-following systems. Fewer markets means less diversification, more volatility in returns. More markets means capital spread thin, potential for over-trading, and diminishing marginal diversification. The Turtles traded about 25 markets. Include multiple asset classes for true diversification.

Monitor rolling metrics: 50-trade win rate, profit factor, average R-multiple. Compare to backtest expectations. Significant persistent deviation (e.g., backtest shows 38% win rate, live shows 28% for 100 trades) signals potential edge decay. Also watch for regime changes - if ADX across markets is consistently low (markets not trending), the environment doesn't favor breakout systems.

ML can classify which breakouts are likely to succeed based on features: volume relative to average, ATR percentile, ADX, prior false breakouts in that market, distance from MAs, etc. Train classifier on historical breakout success/failure. Use predictions to filter signals (only trade >60% probability) or adjust position size (higher probability = larger size). Validate rigorously on out-of-sample data.