The RBT Approach
What are the ingredients of a Rule-Based Trading approach?
1 – A setup, or a precondition, is the starting point. Matters not if it is a simple bar pattern such as an IP or Key Reversal, or a more complex one such as a triangle or head and shoulders formation. The same basic concept applies. Indicators can also be part of the mix or even Market Sentiment. The event could be based on phases of the moon, front page articles in a magazine or any seemingly out there idea. The point is that the event or setup needs to be quantifiable - black and white. This is the 'Look' part of the my equation.
2 – Naturally, there needs to be a level at which you enter the market based on your chosen 'setup' or event. This may be on the close of the period that completes the setup, or on the next open or a break of the current bar's extreme or one 3 bars ago … or a set number of points beyond the next open ... this part is limited only by your imagination and ability to have your idea expressed in computer language. This is the 'Investigate' part of the L.I.V.E.
3 – The next part is to decide on where you will admit that all is not well and you need to cut your loss. The Initial Stop Loss (ISL) may be on the break of the opposite extreme of the bar that creates your setup or may be a fixed number of points beyond that bar or indeed a fixed number of points beyond your entry level. Again there are many alternatives to consider here. As the entry and initial exit levels are known, the risk can be quantified. The 'V' part of my equation does that and a bit more.
4 – Next step is to determine how the position will be managed on an ongoing basis. Will it be exited on the close of the bar of entry (as many daytrading systems do as opposed to intra-day trading systems) or 2 or 'x' number of bars later or at a fixed profit target. Will some sort of Trailing Stop Loss (TSL), Break-Even Stop Loss (BESL) or even a Time-Based Stop Loss (TBSL) be used as an exit strategy. Maybe even adding a Common Sense Stop Loss (CSSL). This is covered in the T.M. part. Another consideration is whether the entire position will be liquidated or the exit staged as popularised by Walter Bressart way back in the 1980's.
5 - Position sizing is part of the Money Management (MM) component of coming up with your own ruleset or system. This aspect is usually done after you have come up with a profitable set of rules (points 1 to 4) that fit your goals, risk tolerance, time constraints, abilities and personality. Whilst a change in the MM component should not turn a losing ruleset into a profitable, I have found that a change in MM can certainly improve both the results (profit) and the path (the drawdown in both duration and depth).
Before getting too carried away with your raw results a reality check is needed in terms of those results. Are they real or purely a numbers game. In other words, have you factored in the very real cost of the spread and cost of doing business such as slippage and broker? These can make a dent in otherwise shiny numbers! A real trap for many players.
The icing on the cake comes when applying the awesome power of compound to the results. Even this aspect has some wrinkles to deal with. They relate to the concept of dependency and ‘depending’ on the sequential results of your ruleset you may be better served to apply the conventional version of compounding. Alternatively your results may benefit handsomely by applying my non-conventional approach to compounding.
What data is needed to test a set of rules?
Preferably you use the same data that you broker uses or gives you via the trading platform. This is the ideal and rather rare to say the least. There are many sources of both End of Day (EOD) and intra-day day available for most, if not all, markets. Needless to say, the data should be clean and accurate, else GIGO applies. It is vital that you know what the data represents … is it the bid or the ask or a midpoint price. In the case of the futures market is it of a particular delivery month contract or a hybrid price of a ‘perpetual’ contract.
How much data is needed?
Now that is an interesting one. There is no one size fits all answer in my mind. There needs to be enough data to represent the various phases a market evolves through … and there are more phases that first come to mind:
Trend up – trend down – sideways congestion
These are the obvious ones. Initially less obvious is that a trend up may be a gradual affair, almost like a sine wave or a direct or violent one. Equally the same can be true of a trend down as well as sideways movements. This makes the answer market specific as opposed to a generalisation.
Never use all the data you have to create and test your working hypothesis. This leads to inevitable curve-fitting, the death knell of most attempts to create a set of rules that actually go on to be profitable in the real market. Curve-fitting is the adding of rules to achieve a desired outcome. It stands to reason that the desired outcome is a better result - either better profit at the end or smaller drawdown on the way through or a myriad of other desired outcomes.
Most people, given enough time, resources and imagination, could probably come up with a set of rules that had zero losing campaigns and a profit way beyond belief. In doing so they would have introduced so many rules and exceptions to the orginal base set of rules that it would be hard to find the original rules. This results in a phenomenon known as 'Freedom Loss'. Much the same as in our every day life when we find ourselves bogged down by endless rules imposed by our elected governments and companies we deal with.
In the markets a consequence of curve-fitting and the resultant freedom loss is that when the same rules are exposed to 'out of sample' testing, the results will probably be far less spectacular. In fact, commonly they would not be profitable at all. Out of sample simply means that you have not exposed the data to the rules whilst you were initially testing the rules … hence out – of – sample.
Throughout the arrticles and vidoes and blogs on this site you will find many clues as to what may serve you well in creating your own ruleset and what may serve you otherwise. You will also begin to see what makes up the L.I.V.E.T.M. approach and how the various elements fit togther.