Sunday, September 8, 2024

Toby Crabel’s Bull Hook Trading Strategy Tested | Ali Casey

As an algo trader, I value patterns for their ease of programming and testing, which allows for the development of robust trading strategies. Today, we'll explore bull and bear hooks, patterns that can vary in details but generally serve to catch traders on the wrong side. Toby Crabel, Joe Ross, and Thomas Bulkowski, among others, have variations of these patterns.

Toby Crabel's original definition of the Bull Hook pattern:
» A Bull Hook occurs on Day 2. A Bull Hook is defined as a day with a higher open than the 
previous day's high followed by a lower close with a narrowing daily range. The next day (Day 1), 
a trade is taken on the initial move off the open, preferably to the upside. «
 
Toby Crabel's original definition of the Bear Hook pattern:
  » Bear Hook is a day in which the open is below the previous day's low and the close 
is above the previous day's close with a narrow range relative to the previous day. As implied by 
the name there is a tendency for the price action following a Bear Hook to move to the downside. «

The Bull Hook pattern has two main forms:

Bull Hook 1: In a downtrend, the pattern is identified when today's bar is an up bar with a smaller range than the previous day and is an inside day (high lower, low higher than the previous bar). We buy with a stop order above the high of this bar.
Bull Hook 2: Here, today's bar is a down bar with a smaller range than the previous day, opening above the previous high and closing below the previous close. This pattern involves just two bars.


For testing, I used TradeStation with S&P 500 e-mini futures data. The backtest for Bull Hook 1 was disappointing, showing a loss with only 15 trades, which seemed unusual given its pullback nature. A deeper analysis suggested that the specific conditions, particularly the inside day and green bar requirements, were limiting trades. By removing some conditions, like the inside day and green bar, and focusing on a simpler pullback strategy, the results improved significantly with about 200 trades and positive performance metrics. For Bull Hook 2, the test also yielded fewer trades than expected, which might be attributed to its breakout nature, not performing well on the S&P 500. Simplifying the conditions here also improved the results somewhat, though it remained less effective. The Bear Hook pattern, when flipped for long trades, performed better but still had a low trade count. Removing some conditions and simplifying it increased the trade count and improved performance. While both Bull Hook patterns had potential, their effectiveness was highly dependent on specific conditions and the number of trades generated. Simplifying the patterns often led to better results.