subject: Our Necessasary Tips To Robust Trading Systems [print this page] Having pcs as powerful as they are these days it is easy to optimize a trading system causing it to look extraordinary, however an optimized system is not a dependable system. Just simply because a trader can train a computer to have 20/20 hindsight will not mean that future performance will be anything like the past.
The primary dilemma with optimizing past performance is that markets transform. A low-volatility market abruptly turns into a high-volatility market. A market inclined to trends becomes a choppy directionless market or, a market that previously had high leverage becomes a market with low leverage. The list is endless.
What tends to occur is that market X will tend to start behaving like market Y, and market Y will tend to begin behaving like market Z. If a investor has extensively optimized his system to trade market Z, then he will be in hardship when it begins to trade like market X! This is a problem with many systems, typically stock index systems that are likely to be optimized to one market or sector. In spite of their occasional remarkable looking results, there's some toxin in their mixture.
Compare this last scenario with one in which the systems model works nicely with most the markets, A thru Z. Now, it will not matter when market Z starts to behave like market Y or market A starts to act like market P. They can change as many times as they want since the systems design will be globally robust with most ALL the various markets. Once again, the market traits can reshuffle countless times and the system works like a Swiss army knife that has proven throughout historical testing it can cope well with most all those situations.
There tend to be a few tip offs to an optimized system.
1. Unlikely looking performance
2. Just trades one market or sector well
3. Utilizes unique rules (algorithms) for each market
4. Utilizes completely different inputs for each market
5. Uses different rules or inputs for entering buys and sells
6. Does not calculate in realistic transaction costs (slippage & commission)
7. Utilizes money management methods that do not include market normalization (like single contract performance only)
8. Utilizes static amounts for all markets like a $2000 stop or $5000 profit target (some markets could strike those in an hour and others could take weeks).
An important feature of a robust system is that it must weight every market equally. The testing should be carried out in a way that normalizes the difference between the markets. For example, natural gas changes an average of a few thousand dollars a day for every contract; but, Eurodollars move an average of a few hundred dollars a day for each contract. Investors need a means to balance and normalize this difference in testing.
The reason traders want to do this is that what if the system fulfills most of the previously mentioned non-optimized rules, but it is trading one natural gas market contract for every one Eurodollar contract. The system will appear best if it had numerous natural gas winners, however what if natural gas begins to have many losing trades and the Eurodollar begins to have many winning trades? Will a couple of, hundred dollar winning trades in one Eurodollar contract be enough to offset a few THOUSAND dollar losing trades in a single natural gas contract?
When a trader is trading 20 markets, it is to have diversification, however if he is trading them all on a single contract basis then he is not diversified. Traders might have 25% of their portfolio generating 90% of the profits and losses! The issue is that moving forward they will be reliant in those markets. It is much better not to be dependent on any given market inside the portfolio. They should all be of the same weight and significance.
In summary, a robust system should do the following.
1. Trade a portfolio of Just about every commodity market
2. Trade that large portfolio over a lengthy test interval
3. Use the same rules for each market
4. Utilize the exact same input values for every market
5. Have the identical (inverse) logic for getting into buys and sells
6. Calcualte in reasonable transaction expenses
7. Normalize the markets for risk
After all this, the last phase would be to perform some walk-forward testing. This would mean, test and create systems on data up till year 2000 (for example). Then after doing all the testing, see how it would have done from year 2000 till today. This would help prevent numerous advantages of hindsight. These are all things we perform in the trading systems creation process here at DH Trading Systems.
