Login

User Name:  
Password::  
Forgot Password?
Sign up with

Join now (it's free)

User Name:
First Name:
Last Name:
Email:
Password::

Six Stages of Strategy Design

Every successful developer needs a blueprint or plan. Whether he is an architect, engineer ,or a budding strategy developer, the plan visually communicates the overall design of the project. It displays how all the parts logically interconnect together to form the house, machine, or strategy. If there is no visual sense of the logical interrelation between parts and whole, there is no way forward, and the result is the construction of a broken home, machine and strategy.

Novice strategy developers embark on strategy development without any sense of design or plan. They take stabs at strategy development in piecemeal fashion, without any form of method. They typically become fascinated with a holy grail indicator that they have seen work enchantingly well on a chart, at least for the period of time they had been watching it. They then think if the indicator’s trigger conditions can be coded into buy and sells of an EA, they will have the holy grail EA. They confirm the strategy “excellent” and “fit” for automated trading after it generates a “good performance” in a one or two year Strategy Tester backtest. They then become shockingly surprised to discover that their good performing EA performs disasterously in a forward test.

Because they did not have a proper method of development and evaluation to begin with, they do not know what to blame for the failure: they may blame the idea to code translation, the problems with MT4, the differences in broker datafeeds, or the unreliability of backtesting. In truth, the bulk of the blame does not rest with their method or, more correctly, their lack thereof. They embarked on strategy development with no method and plan, no steps and rules of procedure, no criteria for systematic backtesting and evaluation, and so they were doomed to fail before they proceeded – regardless of the potential of the indicator or idea they had sought to codify into an EA.

Let us start with a visual overview of the strategy design process. The design of a successful strategy (EA) involves a careful progression through six distinct and interrelated stages, as illustrated below.

Order Stages
1
Selecting Strategy Type (+ Market Type / Timeframe)
2
Design: Developing the Setup (Entry Conditions)
3
Design: Developing the Entry (Optional)
4 Design: Developing the Exit 
5 Design: Money Management
6 Backtesting, Optimization & Evaluation (on eachy stage)

What follows below is a breakdown of these six stages into short definitions, explanations and illustrations. More exhaustive explanations and illustrations can be found in the separate articles detailing each stage.

Step #1: Select the strategy type for the market and timeframe you want to trade.

What is a strategy type? Because markets can be trending, directionless/ranging, or volatile, there are strategy types aimed to exploit each one. There are the trend-following strategies that seek to ride the trends, counter-trend strategies designed to “buy low, sell high” the lower and upper channels of directionless or lateral-ranging markets, and breakout strategies engineered to capture the market thrust through market ranges on high volatility.

As a strategy developer, you should first select the strategy type you want to trade (trend following, countertrend, and breakout) that corresponds to a market type (trending, directionless, and volatile) you want to exploit. You can then go to look for an indicator that works best for that strategy type.

For instance, you might see that the EUR/USD is a currency pair that has long and sustained trends on higher time-frame charts (4 hour, daily and weekly). You would like to be able to ride these large trends. You would then seek to experiment with different setup techniques with trending indicators that can get you into the large trends early on, so that you can enjoy their full ride. Alternatively, you notice that the GBP/USD has long periods of directionless /ranging activity, and you want to experiment with setups using overbought/oversold indicators that can get you to buy long and sell short the bottoms and tops of these lateral ranges.

It is important that the strategy type and timeframe suits your style and personality. You might like the idea of trend trading, but you might not have the patience to wait for a trend to develop on a large timeframe, or the discipline and steady nerves it would take to stay true to the system in the midst of one its corrective phases. Perhaps you would be more comfortable with trend trading on smaller timeframes, which means that you miss the larger profits available on larger trend / timeframe, but you will not have to be as patient waiting for the next trend or as fearful about the next correction.

Selecting a strategy type and time period is critical and you should evaluate the alternatives.Your personality will direct you to the strategy type that is right for you. Only you know what type of person you are and what type of trading would be best for you.

Step #2: Designing the Entry Setup (Entry Conditions)

The entry setup is the condition or set of conditions for determining the entry you want to take on the strategy type you have determined. It is the indicator or group of indicators that once triggered, tell your EA to take the trade. It could be the fast moving average crossing the slow moving average as a condition for entry for trend following strategy, or the RSI moving into oversold territory (below 20) or into overbought territory (above 80) as a condition of entry on countertrend strategy. Numerous indicators and conditions can be used as setups for the strategy type you have selected, and you may have to test out several before you find the ones that may work for you.

The entry setup is the most critical stage of the process because it alone determines the future potential of the strategy. If a setup that does not  demonstrate statistical worth on its own, in its most basic form (without special treatment from advanced entry and exit techniques), there isn’t any point in moving forward with the other stages. Strategy development is time-consuming, and if you can see early on that your setup is getting poor results, you should abandon it and move on. Most strategy setups by themselves will actually show negative results in backtesting—so don’t feel so bad if the indicator that you had been enchanted with does not work out in backtesting. If you see your indicator setup has glimmer of hope after backtesting it you may consider combining with another indicator or with an advanced entry type.

Step #3: Designing the Entry Technique (Optional)

An entry trigger is the signal by which the strategy enters the market once the rules for the setup have been met. The entry trigger is dependent on the strategy and setup you have previously selected. For instance, you might have a moving average cross setup that indicates you should be long the current market trend, and the entry trigger could be a break above the high of today’s bar to confirm that the market is still bullish. The advantage of having such an entry trigger, versus none at all (market order generated from setup), is that you may avoid some false signals and consequent losses. Entry triggers might be based on price action, like the preceding example, or they can be based on interesting twists of other indicators. All new entry trigger ideas must be thoroughly backtested to see if they improve the original setup idea. Entry triggers are also dependent on the types of orders used (market order, stop entry orders, and limit orders), and these entry types need to be also separately experimented with to see which ones are more advantageous.  

Many novice developers trade only the entries, and these are usually not as effective and profitable as strategies that use both a setup and entry combination. Strategies based only on entries have too many trades and a low percentage of profitable trades.

Step #4: Designing the Exit

Most traders think that if they come up with the best setup and/or entry, their work is done, and strategy just needs to be outfitted with an optimized stop loss and profit target. Their work transfers to the optimization engine for finding the right stop loss and profit target level, and they do not feel the need to think creatively about further exit conditions. However, selecting the proper exit for your strategy is as critical as the setup and entry. The exit is the other side of the entry equation and deserves as equal amount of time and creativity.

Initially, when you are first deciding on the setup and entry, you might be using standard exits, such as the reverse conditions of the setup (the short setup conditions exit you from the long trade and vice versa) or stop loss and profit target levels that you estimated upon, based on the strategy type and time frame. It could be that your stop and reverse condition is the most appropriate exit and that your hunches on stops and profit targets were right on. You will not know this until you have tested other possibilities.

You need to examine the charts and experiment with different indicators as possibilities of an exit setup that might allow you to get out of the trade with more profit/less loss than your reversed entry setup. In effect, you need to turn good entry setup combinations inside out, reversing the buy long conditions into exit short conditions and sell short conditions into exit long conditions. This will take you on a whole new path of creative thinking, experimentation and backtesing. For instance, you might be a trend trader that uses the moving average setup to get you into the trade, but you find that using the overbought/oversold indicator of stochastics is the best way for you to get out of the trade (exit long when overbought, exit short when oversold).  

Of course, you should also experiment with all general types of exits: stop loss, profit target, trailing stop, and breakeven. All these forms of stops can be calibrated to be a dollar value, pip value, percentage value, or average true range (ATR) value. The optimization engine can certainly help you calibrate the right form and value to have as a stop loss and profit target. It can also help you determine the appropriate trailing stop and breakeven or none at all. While the profit trailing and breakeven can help you protect your profit, it can also restrict your profit. Likewise for the profit target, which you might not need if the reversal exit setup discussed above allows your strategy to reach for more profits.  The tougher general exit to decide to use or keep is the stop loss. Stops also interfere with market action: if they are too close, we end up getting stopped out to early, and if too far, you can take on more damage. In the end it is a comprise and balancing act. It is generally safer and more comfortable to use a stop loss of some sort, instead of relying totally on the stop and reverse exit setup. You never know when that freak market event can wipes you out because you did not have a stop loss in place.

The most difficult thing about trading is accepting the losing trade. For novice traders the losing trade means that something is not right and they have made a mistake. For experienced traders, losses are just a cost of doing business. You should even be prepared to love losses – you will be spending a lot of time with them.

Step #5. Selecting the Money Management Scheme

Most wise traders emphasize importance of sound money management – that it alone makes the difference between success and failure.  They might even go so far as to suggest that money management trumps in importance the design and execution of the strategy. They would argue that successful traders are not worried about tweaking their indicator or setup, they are worried about the risk on each trade. I would not go that far – for every part or stage in the design is important enough and each is interdependent on the other (without a good setup even the best of money management schemes could not save the account). However, I do agree that money management is highly important, and that that importance can be overlooked by naïve traders.

Working with an effective stop loss is one step towards proper money management. The stop loss is the worst case scenario of any given trade – and it can help you determine the percentage of the account that should be risked for each trade. Bear in mind that if your strategy is going to open several positions concurrently, you need to know the combined risk of all open positions relative to your stop loss and account size –and imagine the worst case scenario if all open positions hit their stop loss levels.
Another part of money management is scaling/pyramiding – knowing how to scale into a potential big move and scale out as the market moves in his direction. The focus is on the value of pyramiding to maximize the leverage of his open equity.

Step #6. Optimization, Backtesting and Evaluation

All three concepts are interrelated and important in each stage of the design process. In each stage of the design process, you will backtest your ideas using standard variables, optimize the variables, select some optimized variables to backtest further, all the while evaluating each backtest and optimization according to key statistical measurements.

The four numbers that are statistically important and reflect the viability of the strategy are: total number of trades, the average profit per trade, the largest winning trade, and the profit factor. Other numbers reflect your comfortability trading it: MAXID, Percentage Profitable Trades, Maximum Consecutive Losers.  All these numbers must be within acceptable parameters.

There are risks to backtesting and opitization that are well known, but we seek to minimize those risks instead of discarding backtesting and optimization altogether as invalid. Future prices may be different from the past – but there are degrees of similarity that justify the use of historical data. We not looking for exactness – the markets never move in the exactly the same manner twice and no one can predict them – and instead we are looking for degrees of similitude. The real questions we ask is "what are the risks of backtesting and over-optimization, how do we see them, and how we can minimize them?" 

We will spell out the answers more thoroughly in the appropriate article but suffice it to say when we look at optimization and backtesting, we must keep in mind the ideas of integrity, simplicity, surrounding parameters, along with the testing of optimized scenarios on multiple currencies and time frames with the key numbers in sight. Backesting should be conducted on a statistically relevant sample size – which includes how many years back, the number of trades, and the number of bars. Historical data should be manovered into two periods: one representing the control group (or period) and the other representing the experimental group (period). The experimental period receives the strategy testing and optimization, and the optimized strategy is then tested to see if it works the same way on the control period.

Summary

The object of this article was to sketch out the design process in six stages, and subsequent articles will move deeper into each stage.  You will be learning how find a proper strategy type that fits your personality, market, and timeframe. You will learn how to discover one or a combination of indicators appropriate to that strategy type. Next, you will learn how to retool those indicators into the proper setup and/or trigger, which can be different or the same for your entry and for exit. Lastly, you will learn how to enhance a promising strategy with an equally promising money management mechanism.

Throughout all these stages you will learn how to thoroughly and rigorously backtest each component of the strategy, in isolation and in total, on a statistically relevant sample size using the most relevant criteria.

Top
An e-mail with your verification code has been sent to your e-mail address. Please access your in-box and use the verification button or verification code to complete your registration.
You already have an account linked with this E-mail (it maybe standard or social login). Please, sign in with it.
Please, provide us your e-mail so we can verify your account.
Email: