Knowledge · Strategies · Donchian Channel Breakout (Turtle Trading)

Donchian Channel Breakout (Turtle Trading)

Richard Dennis & William Eckhardt — 'Turtle Trading' Experiment (1983)
Breakout Evidence: Strong position futurescommoditiesfxcrypto
8/10
Relevance for Botty
Long on a breakout above the 20/55-day high, short on a breakout below the 20/55-day low. ATR-based position sizing. Richard Dennis' original system.
A price breakout above the high of the last N days is a validated trend start. Combined with ATR-based position sizing (N = 2×ATR unit) and pyramiding, the system historically beats buy-and-hold by a wide margin.
Relevance Score 8/10
Hyperliquid BTC perp is an ideal Donchian market (high volume, clear breakouts). Botty could implement a Donchian-55 strategy with ATR stops and a 10-day exit. Problem: a 55-day high on intraday charts means very long holding phases.

Entry

  • System 1 (short): long when price closes above the 20-day high; short below the 20-day low
  • System 2 (long): long when price closes above the 55-day high
  • Pyramiding: an additional unit every +0.5×ATR (max 4)
  • Risk per trade: 2×ATR × unit size

Exit

  • Stop at 2×ATR (N) against entry
  • System 1: exit on a 10-day counter-breakout
  • System 2: exit on a 20-day counter-breakout
NameTyp. valueDescription
entry_period 20 or 55 Lookback for the breakout level
exit_period 10 or 20 Shorter channel for the exit
atr_period 20 N calculation (True Range average)
unit_risk_pct 1-2% Risk per unit on the account

Pros

  • One of the best-documented mechanical approaches
  • Clear rules, no discretion needed — ideal for bots
  • ATR sizing is standard today
  • Profitable across commodities/FX/crypto historically

Cons

  • Low win rate — psychologically demanding
  • Many false breakouts in ranges
  • Weaker since the 2000s in some markets (crowded)
  • Barely works on individual stocks
cagr
The Turtles made $175M in 4 years (original experiment)
notes
Win rate below 50% — profit comes from a few large trends. Very robust in commodities/FX, weaker on stocks.
win rate
~38%
max drawdown
~20-30%
profit factor
~1.5-2.0
Perfect — purely mechanical, no questions of interpretation. ATR sizing and pyramiding cleanly codable.

This is what the strategy looks like on real BTC 4h candles — exactly the configuration deployed on Wallet 1 (donchian 20 + adx_rising + ATR trailing 1×/3× + partial TP), all trades from a real live-sim backtest 2025-2026. Interactive: zoom/scroll, click a trade in the list → the chart jumps to it, dotted terms explain themselves on hover. Open in fullscreen

The story

In 1983, Richard Dennis (then one of Chicago's best-known commodity traders) started a bet with his partner William Eckhardt: trading rules are teachable. He recruited 23 beginners with no prior knowledge, taught them his system in two weeks, and gave each real capital. Result: the 'Turtles' together made over $175 million in profit in 4 years.

The system was kept secret into the 1990s and only made public by ex-Turtle Curtis Faith.

The rules

System 1 (short)

  • Long: price closes above the high of the last 20 days
  • Exit: price closes below the low of the last 10 days (or 2N stop)
  • Skip: if the previous signal would have been profitable, skip (anti-whipsaw)

System 2 (long)

  • Long: price closes above the high of the last 55 days
  • Exit: price closes below the low of the last 20 days
  • No skip rule

ATR-based sizing

  • N = 20-day ATR of the asset
  • Unit = 1% of the account / N (in contracts)
  • Maximum 4 units per market; add every +0.5N
  • Stop: always 2N against the average entry price

Why it works

The actual edge lies not in the breakout signal but in risk management:

  • Win rate only ~38% — most trades are small losses
  • But: the few winners run 5-20× as far as the losers
  • Pyramiding turns medium winners into large ones
  • ATR sizing ensures consistent risk-per-trade across different vol regimes

Modern relevance

Donchian breakouts still work today, but weaker on stocks (too many whipsaws in single names), still strong on commodities and crypto. A typical modernization: an additional trend filter (e.g. only long breakouts when above the 200-MA) to reduce sideways false signals.

Relevance for Botty

Botty trades BTC perp intraday. A Donchian breakout on 4h or daily would be a natural add-on: simple, mechanical, ATR-stop-compatible with Botty's existing risk infrastructure. The challenge: position holding time with the 55-day system is weeks long, which does not fit Botty's current intraday loop — a 20-day system on 1h bars would be more practical.