Saturday, September 05, 2009

Developing a Trading System to Support Discretionary Traders

My recent post explained how backtested signals from trading systems can provide useful decision support for discretionary traders. In that post, I illustrated how unaided discretionary judgment can lead traders astray, as buying strength and selling weakness has led to consistent, short-term underperformance.

To move my illustration forward, I modified the system described in that prior post and incorporated the modified version as the first in three triggers that produce buy and sell signals.

Trigger 1 occurs when stocks close above or below a defined distance from their moving averages. This captures relative strength or weakness. Trigger 1 serves as a heads up that a signal is likely to be generated in coming days. (This system uses daily data only; it is a swing trading system).

Trigger 2 occurs when stocks lose upside or downside momentum after the first signal has occurred and after stocks have continued their move to new price highs or lows. This captures a loss of momentum to the upside or downside. Trigger 2 is based upon a patterning of the proprietary Demand/Supply indicator that I post each morning before the open via Twitter (follow here). Once the second trigger occurs, a trade signal is imminent.

Trigger 3 is an execution signal based upon an intraday configuration of the NYSE TICK. The signal looks for buying that makes a lower high (if one is going short) or selling that makes a higher low (if one is going long).

The conservative exit occurs when Trigger 1 is no longer in force. More aggressive exit criteria consist of weekly price targets that are adjusted for recent market volatility. The longer the period between Triggers 1 and 2, the more aggressive should be the price target.

The stop loss occurs when Trigger 3 is violated; i.e., stocks make a higher high after having seemed to have topped out. No position can be entered unless the move to the conservative exit makes at least as much money as the potential loss if the stop is hit. That means that there is very good risk:reward on trades to the more aggressive price targets.

The average holding period for the trades is 2-3 days. On paper, backtested several years, the system appears to be particularly successful for the following instruments: SPY, IWM, QQQQ, XLF, XLE, XLB, XLK, and XLP.

Interestingly, the system has not been successful for GLD or USO, suggesting that commodities follow different short-term patterns than stocks. However, the system has been quite effective with EEM, suggesting that the short-term patterns may extend to markets outside the U.S.

Long-time readers will recognize that this is a systematic implementation of the transition pattern that I've described in numerous posts. This is all work in alpha stage; I need to firm up the last two triggers before being anywhere near ready for prime time. If the subsequent development looks promising, however, I will post all signals to the blog and then illustrate how they can be used in discretionary trading.

Similarly, I will be looking for signals from Henry Carstens' systems (follow here) to serve as decision support for discretionary trading. Hopefully, I'll be able to cajole him to post useful information regarding how his system signals might inform short-term discretionary trading. Knowing his commitment to traders and their development, that shouldn't be too difficult.