Knowledge · Strategies · Pullback-to-EMA (Holy Grail)

Pullback-to-EMA (Holy Grail)

Linda Bradford Raschke & Laurence Connors - 'Street Smarts' (1995)
Trend Following Evidence: Moderate intraday to swing futuresequitiescryptofx
9/10
Relevance for Botty
In a strong trend (ADX > 30): wait for a pullback to the EMA-20, then enter on the first continuation candle. Linda Raschke's classic.
Core concept
Strong trends (ADX > 30) continue after pullbacks at an above-average rate. The EMA-20 acts as a dynamic support/resistance zone - many traders trade off it, which is reflexively reinforcing. Enter only after a confirmation candle, not preemptively.
Relevance for Botty
Relevance Score 9/10
Botty already implements the setup in strategies/holy_grail.py. Core logic: ADX > 30, pullback to EMA-20, confirmation. Well integrated with the existing infrastructure (ATR stops, trailing stop, EMA filter).
Rules

Entry

  • Trend filter: ADX(14) > 30
  • Direction filter: price > EMA-200 (for longs)
  • Pullback: price touches or comes close to the EMA-20
  • Confirmation: the following candle breaks the high of the pullback candle
  • Entry: on the breakout above the pullback high

Exit

  • Stop: below the pullback low (or 1-2x ATR)
  • Target 1: 2R trailing stop at an ATR-scaled distance
  • Alternative: exit on an EMA-20 breakdown against the trend direction
Parameter
NameTyp. valueDescription
adx_threshold 30 (25 for crypto) Minimum trend strength
ema_pullback 20 Pullback target
trend_ma 200 Higher-level direction
Pros & Cons

Pros

  • Intuitive and visually clear
  • Entry only after confirmation - filters out reversals
  • Positive R-multiple thanks to the natural stop below the pullback low
  • Works across multiple timeframes (1h to daily)

Cons

  • Less mechanical than breakout systems - the definition of 'pullback' is discretionary
  • In weak trends (ADX < 25) many false signals
  • Misses direct trend moves without a pullback
  • Problematic on altcoins with V-shaped movements
Typical performance
notes
Less systematically backtested than Connors/Turtle; heavily dependent on clean trend detection. In crypto it works relatively well on 1h-4h.
win rate
~45-55% with clean rule application
r multiple
on average 1.5-2R on winners
Bot suitability
Very good - if the rules are coded strictly mechanically (ADX threshold, EMA-touch logic, confirmation candle)
Background

Origin

Linda Raschke, a hedge-fund manager with 30+ years of experience, described the setup in Street Smarts (1995, with Larry Connors). The name 'Holy Grail' is meant ironically - not a magical system, but a simple, reliable trend-continuation setup that has proven itself over decades.

The setup (long)

  1. Trend check: ADX(14) > 30 - the market is trending strongly
  2. Direction: price above the EMA-200 (higher-level trend)
  3. Pullback: a correction to the EMA-20
  4. Confirmation: the next candle breaks above the high of the pullback candle
  5. Entry: stop-buy above the pullback high
  6. Stop: below the pullback low
  7. Exit: trailing stop or EMA-20 break

The short version is mirror-imaged (ADX > 30, price < EMA-200, pullback to the EMA-20 from below, breakdown below the pullback low).

Why it works

  • Strong trends have persistence: statistically, ADX-30+ trends continue disproportionately often after short corrections.
  • EMA-20 as a magnet: many traders watch this line; their buying/selling there reinforces the bounce.
  • Confirmation filters reversals: without confirmation you would often be buying into an incipient trend reversal.
  • Asymmetric risk-reward: the stop below the pullback low is usually tight (~1x ATR), a target at 2-3R is realistic.

Typical mistakes

  • Entering too early: 'catching' the pullback preemptively without a confirmation candle -> often a trend reversal instead of continuation
  • ADX threshold too low: below 25 it is not a trend; the setup loses its edge
  • Stop too tight: below the exact pullback candle it is often too tight, a 1.2-1.5x buffer is better

Variants

  • Raschke original: ADX > 30, EMA-20, first-bar signal
  • Adam Grimes: a similar setup with a 50-EMA instead of 20
  • Jesse Livermore (indirectly): the line-of-least-resistance concept is conceptually related

Relevance for Botty

Already implemented as strategies/holy_grail.py. Potential improvements:

  • Test whether the ADX slope (rising ADX) matters more than the absolute threshold
  • Refine the confirmation-candle definition (close vs. break of the high)
  • Systematically optimize the pullback tolerance (how close must price come to the EMA-20?)