Tactical Trading with Mechanical Systems, 09/03/05
Tactical Trading with Mechanical Systems In the examples below, mechanical trading signals are used to provide tactical depth to discretionary trading decisions.
Four different signals, from four different and uniquely purposed trading systems. In two of the examples, one system is used to set-up or back-stop another system, providing extra depth to trading decisions and risk management.
All of the signals are generated by systems that trade the SP futures contract and were tested from 1997 to present. The first signal testing criteria is statistical stability (high z score) the second criteria is a stable ratio of expected wins to actual wins over time. These measures help weed out signals that worked in the past but are no longer working or have currently decreased effectiveness. The goal of the testing criteria is to help generate signals that are the most likely to perform in the future as they have performed in the past.
1) Signal 1: Long at 11:30, exit MOC (Example 7/19/2005)
Buy at 11:30, exit MOC
102/164, 62%, avg 2.0 pts
This signal has occurred 164 times since 1997. It has been correct 62% of the time and averages 2 pts per trade. The average winning trade is just shy of 8 pts.
For the discretionary trader, this is an example of a signal that says, "Start looking for an entry, the market is likely to be higher on the close." See signal 2 for a potential set-up trade and signal 4 for a potential back-up trade.
2) Signal 2: Long on close, exit the following open (Example 7/18/2005)
339/549, 62%, avg 1.2 pts
This signal has occurred 549 times since 1997. It has been correct 62% of the time and averages 1.2 pts per trade. The average winning trade is 4.5 pts.
The discretionary trader can use the signals from signal 2 to enhance entry and exit decisions. For entries it can be used to enter a trade early, such as anticipating signal 1. For exits it can be used to help determine whether to exit a trade on the close or hold it overnight.
3) Signal 3: Long on the close, exit on the first profitable close or after three days (Example 8/5/2005)
188/221, 85%, avg 4.5 pts
Signal number three is from a high probability trading system, the signal has occurred 221 times since 1997. It has been correct 85% of the time and averages 4.5 pts per trade. The average winning trade is 7.75 pts.
The discretionary trader can 'trade around' this high probability signal, taking profits on runs up and buying back on declines. Entries and exits can be enhanced w/ Signal 2 and back-stopped with Signal 4.
4) Signal 4: Long on the open, exit MOC (Example 8/8/2005)
82/126, 65%, avg 3.0 pts
This signal has occurred 126 times since 1997. It has been correct 65% of the time and averages 3 points per trade. The average winning trade is 10 points.
Signal number four is not only a great trade but it does double-duty as a back-stopping trade because it is often set-up by failed long signals from other systems. Knowing that there is a high probability trade behind an existing signal makes it easier to stay with an existing position.
On 8/8/2005, Signal 3 began to fail and Signal 4 backed it up, allowing the discretionary trader to either scale into Signal 4, add leverage on a scale and take both Signals 3 and 4, or simply stay with Signal 3.
For the discretionary trader, mechanical signals are like war-time reconnaissance: they show a general enemy troop strengths, aid in campaign planning and give the general an edge. For the system designer, trading systems that bridge the gap between automation and tactical support for the discretionary trader are the result of conscious multipurpose design.