Having determined the objectives for your trading system and with a thorough understanding of what it is you want to achieve with your ‘edge’ it is now time to look at the entry and exit signals you will use. Specifying exit signals is a must! Too many discretionary traders and those lacking a disciplined approach to the markets spend countless hours of their time agonising over entry signals but pay little attention to exit signals. Exit signals need to be considered for both losing and winning trades. The first to clearly define stop loss levels on trades that simply do not work right from the outset, and the second to provide a set of rules for extracting ourselves from profitable trades – this may be either with the use of a trailing stop for medium to long term trend followers, or the use of a profit target mechanism for shorter term traders.
Entry and exit signals
The purpose of designing a system is to define a set of unambiguous entry and exit signals that:
• have a positive mathematical expectancy
• are repeatable into the future
• are practicable for you to engage the market according to who you are and what you want to achieve
The design process outlines how to achieve these objectives.
Concepts
Once you have stated your objectives, next you need to decide what concepts you will use to research your buy and sell timing criteria. You may also include other timing points like profit stops, or rules for when to accumulate, pyramid or lighten any position. You must be able to use the concepts in live trading with the tools you use on a daily basis.
Typical concepts – and remember, there are others – you have at your disposal for filtering and for entry and exit signals include:
• candlestick and bar formations and patterns
• price breakouts—prices making new timeframe highs or breaking out of a range or a channel
• price exhaustion—entering on troughs (support) for long trades or peaks (resistance) for short trades
• volume action linked to price action
• support and resistance (within a trend channel or range trading)
• momentum concepts using typical momentum indicators such as moving averages, Rate Of Change, MACD, RSI, ADX, Over Bought/Over Sold, or derivatives of these and other indicators
• parabolic
• Average True Range (ATR), Daily/Weekly range, Standard Deviation and other volatility calculations
• Fibonacci retracements and projections
• Elliot Wave and fractal swing charts
• Steidelmeyer
• Comparative Relative Strength
• inter-market analysis
• fundamental analysis
• combinations and permutations of the above
Of course, the system designer also has the option of designing his own ‘indicator’. There are many excellent sources for ideas and concepts for back-testing.
Some concepts lend themselves better to different timeframes. A designer needs to research which concepts match certain timeframes. Whilst texts that accompany various trading systems on sale around the world will contain statements about how their system rules transcending timeframes, in the majority of cases this has not been my experience when researching the stated rules of these systems. However, some concepts do transcend different timeframes.
Each concept has different variables which can be researched such as time periods, extension levels, smoothing values, etc. Variables for each concept can also be researched. Which variable you choose for a particular concept depends on the objectives you have in mind for your system. For instance, it is unlikely that a nine day RSI will be useful for a medium to long term trend following system that requires average hold periods of, say, five months.
Furthermore, the listed concepts can be used with different methods of signalling, such as:
• crossover with a moving average(s)
• crossover of a zero line (e.g. MACD)
• change in direction
• an indicator diverging from raw price action
• crossover of an overbought or oversold zone (e.g. RSI)
• reaching a particular retracement level such as 61.8%
• above or below a particular indicator level for a minimum specified period
• a count of bars/candles on intra-day, daily, weekly or monthly charts
• percentage movements in price or indicator values
• multiples of indicator values (e.g. 3 x ATR)
• plus many more—this is just a small taste
There are a huge number of parameters and interrelationships that you need to explore and experiment with during this stage of the system design process. This is perhaps the most time consuming phase as it requires hours of work to research, implement, check and cross check these combinations of relationships. Only by experimenting with these can you arrive at the combination of tools that you believe will combine to provide you with a system that gives you a profitable edge over the market. Much of this is arrived at thoroughly back-testing your raw system on the available historical data. The process of back-testing will be covered in detail in next week’s blog.
Response to Comment by P. Allan:
RE Rober Colby, not bad for a bloke who once got paid to be a fundamental analyst!! He is one of the many converts to technical analysis which, not that long ago, was considered voodooism, hocus pocus, off-with-the-fairies stuff. Many acedemics still consider technical analysis in this light.
See the beginning of an interview with Robert Colby in the Dec 2006 S&C magazine at this link: Interview with Rober Colby.
Regards
Gary
Not a comment.Coud give me the portfolio managemet performance for the last 10 years(each year)?
A very useful coverage by Gary of how to build a trading model.
Robert Colby’s excellent “The Encyclopedia of Technical Market Indicators” (McGraw Hill 2003) back tested 127 technical indicators over 101 years (1900 – 2001)using the DJIA.
He says that when using both long and short trades the sweet spots for moving averages are 5 days, 44 days and 120 days for exponential moving averages (EMA) and 5 days, 66 days and 126 days for simple moving averages (SMA).
He has tested each EMA and SMA against a buy and hold strategy and found that even without enhancing them with momentum oscillators and other bells and whistles they beat buy and hold by a big margin. Based on the daily closing prices for the DJIA from 1900 to 2001, the 5-,3- and 2-day EMA would have produced maximum net profits in excess of six billion dollars, but all EMA crossover stratgies of all lengths up to 300 days would have been profitable and beaten a buy-and-hold strategy by at least 69% assuming one started with $100 in 1900(page 264).
The 5 day EMA crossover strategy was found to be the best and would have beaten buy and hold by 78 million percent over the test period, but its trading frequency would have been hyperactive with on average one trade every 5.88 calander days (page 265).
He also found that only a minority of long or short trades using moving averages were profitable, but their net gains far exceeded the net losses from unprofitable trades. In other words the ‘edge’ of moving average strategies depended not on a majority of trades being wins, but on the average profit of each successful trade far exceeding the average loss of each failed trade.
The longer EMAs would have been less profitable, but still beaten B+H by a wide margin.
We at http://www.marketiming.com.au will leave active trading to SPA3 which is an execllent model. Our focus is on share investors who want to do only a few transactions a year to protect their capital from share market corrections and crashes. Such investors dont want to be active traders focusing on the short term, but prefer instead the slower pace of catching or avoiding only medium to long term market waves.