|
|
||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Developing Trading Systems - Part 3Cho Sing Kum This is the last of a 3-part series which I wrote in a financial forum on 3rd June 2001. Lightly edited.
This is probably the
most difficult part to write. But what I am going to introduce you to
today will certainly bring you closer to profitable trading. No, it is
not a secret trading system. Neither is it the rediscovery of some
tightly kept secret indicators. But if you were to pick up from here and
explore it further, I can assure you that you will be much more
systematic and disciplined than before and most importantly much more
profitable in your trading. You will also find out why the Holy Grail is
not be found in the indicators or trading systems. That is the wrong
place to look in the first place. Inspite of the difficulty in putting
the subject matters across, I will try to be as clear as possible. I am
also learning with you as I write this. Part 3 requires that you have followed
Part 1 and 2. So before we go on to the article proper, I shall provide
the links to Part 1 and 2 and the supplementary postings. Please read
through them if you have missed them or if you need a refresher. I like to repeat here the Benefits of Strategic Trading that is so helpful in rectifying the haphazard methods of trading that so many of us are doing. Without these it is not possible to go on to the next step to evaluate your trading systems as one portfolio, not piece meals. Benefits of Strategic Trading
Combining trading systems into a
Portfolio In Part 3, we will bring trading system
one BIG step forward by combining them into a portfolio. There are a few
ways to create a portfolio. This is also referred to as portfolio
format:
I will continue with the trend system
introduced in Part 2 and build on from there. Taking the same system, I
will apply it on multiple markets - five to be exact - in the short-term
interest rate futures market of Canada Bankers Acceptance, US
Eurodollar, UK Short Sterling, EU Euribor and the EuroSwiss. Since I will be applying the same system
to five markets, I will be using format 1. This format is often
considered to be one of the more robust applications of system testing
and trading. This is because if the same system is applied to several
markets generating profitable results, this may be considered to be more
statistically significant than results obtained from applying the system
to only one market. Limitation is that optimal results are sacrificed to
find a system that is viable across multiple markets. Let's proceed. Applying The Trend System to the Five
Markets The markets are spread across different
countries hence the profits and losses are also reflected in different
currencies. In order that these profits and losses be combined and
evaluated in total, I have to apply the respective exchange rates to
convert the tick value and the trading system parameters to show in US$.
The base currency will be US$. The exchange rate used are: US$ 1 = CAD 1.5 The parameters for the trend system are
the same for all markets. They are 10 ticks for Money Management Stops,
40 ticks for BreakEven Floor and 100 ticks for Trailing Floor. Trailing
retracement is 20 percent. The parameters for the technical indicators
are also constant. The moving averages are 5 and 20 days while the slow
stochastic is 5 day. Account size is the same as in Part 2 which is
US$20,000 equally allotted to all five markets at US$4,000 each.
Basically all these are unchanged from Part 2. For continuity of historical data, I am
using the backward-adjusted 3rd delivery contract month for all the five
markets. Usually for short-term interest rate futures, the 2nd contract
delivery is the most active but it is always the further delivery months
where trends are more pronounced. However, they would not be as active
as the front delivery months. This will provide the data for trading evaluation. I am using data starting from Jan 1996 as this will give at least five years of data, which should be sufficient for testing and evaluation. So while in Part 2, I used the outright Eurodollar Sep 2001 contract, now with the backward adjusted 3rd delivery contract the result is bound to be different. With this in mind, let's see what are the results thrown up by the five markets. The Individual Market Results I am sure you are anxious to know how they performed. The following are the respective Equity Curves for 1994-1996. As you will see, three are profitable and two are not.
The Net Profit column is self-explanatory. The Win, Loss and % Profitable columns show how reliable the trades are. Naturally a higher % Profitable number the better. Sharpe Ratio Average monthly return (%) minus the risk-free rate (in this case 5%) divided by the standard deviation of monthly returns. The higher the number, the greater the return in relation to the risk. This calculation is based on the last 36 months. A number of 1 indicates a good system and a higher number will be better. Only Euribor has a good number of 1.11. Even though Eurodollar and EuroSwiss are profitable, the Sharpe Ratio are 0.89 and 0.85 respectively. Certainly an area to improve to make the system better. Needless to say, the numbers for Bankers Acceptance and Short Sterling are not good. RINA Index This is a proprietary index of RINA
System. It combines select net profit, time in the market, and drawdown
calculations into a single reward/risk ratio. The larger the number the
more efficient the system. This performance measure is a trade-based
statistic as opposed to equity based measures of performance such as the
Sharpe Ratio. Look for a system with an index of 200 or more. The
formula is: RINA Index = (Select Net Profit)/((Average Drawdown) x (Percent time in the market)) Select Net Profit This figure artificially adjusted the system's results by removing all positive and negative outlier trades. The final figure presents a net profit devoid of aberration trades. Systems that are heavily dependent upon outlier trades will have dramatically different select net profit results than the systems that do not. A trade is considered to be an outlier if it's profit/loss is greater than three standard deviations away from the average. A trader may also want to pay attention to whether or not the model is attempting to systematically capture returns from "outlier" or rare events when using this information to make value judgments about systems. Average Drawdown The average maximum open loss (whether realized or unrealized at the time) of all the trades. Percent in the Market Divides the test period by total time in
the market to produce the percentage of time spent in the market. While this index does provide some very useful input, I think for the trend system the index is likely to be low for two reasons. The first is that the system try to catch trends and stay the course (letting profit run). This means that the Percent in the Market number will be big. The second reason is that the occasional big profits from trend moves are treated as outliers and excluded in the Select Net Profit. Still, this is a good Index to have. Perhaps we can try to make the number bigger somehow by improving the trading system. One area that is immediately obvious is the Average Drawdown. After you understand Entry and Exit Efficiency that are explained below, you will know that by improving the Entry and Exit Efficiency to push the Average Drawdown number smaller, the RINA Index number will increase. We never know if we don't try, isn't it? Entry Efficiency Exit Efficiency is defined as a maximum
possible realized difference in prices from a trade that has the trade
exit price expressed as a part of the total profit potential during that
trade. Exit Efficiency shows how well a system exits a trade. If a trade
is long - how close an exit to the highest point within the trading
period, if a trade is short - to the lowest point within the trading
period. The following formula is used to compute Exit Efficiency for a
trade. Entry Efficiency = Maximum Possible
Difference in Prices For This Entry/Profit Potential. Maximum Possible Difference in Prices For
This Entry is the difference between the Highest Close Price (for Long
Trade or the Lowest Close Price for Short Trade) and Entry Price. That means For Long Trades For Short Trades A good Entry Efficiency number is at least 60 percent or higher. Only EuroSwiss meet this criteria at 68.17 percent. Even the most profitable Euribor is below this marker at 58.43 percent. Obviously improvement in this department is required. Exit Efficiency Exit Efficiency = Maximum Possible
Difference in Prices For This Exit/Profit Potential. That means For Long Trades For Short Trades Similarly to Entry Efficiency, a good number is also 60 percent or higher. The trend system fare very badly here. The biggest number is a lowly 26.27 percent in Euribor while the Banker Acceptance number is well…6.08 percent. Definitely a lot of work must be done in this area to improve the profitability. Evaluating the Portfolio Now you may be wondering what is the
performance result of all five markets combined in total as one
portfolio. The individual performance summary reports generated by
TradeStation 2000i were imported into and combined by Portfolio
Evaluator 2000 on a bar-by-bar basis in order to calculate values such
as Maximum Equity Drawdown for the portfolio, Sharpe Ratio, RINA Index
and many others, just like in the individual reports and more. The results of certain risk measures for
the portfolio, once combined, may be smaller or greater than the
individual reports. Therefore, portfolios have the potential to yield
higher or lower risk than any of the individual markets that comprise
the portfolio. Analysis of the reports are done bar-by-bar which is the
only means to understand precisely the effect of combining systems or
markets or both. For example,
let us look at the Maximum Equity Drawdown of trend system in the five
individual markets. See Table 2.
Table 2. Max Equity Drawdown
Correlation Analysis With the advantage of bar-by-bar
analysis, we are able to find out whether the five markets tend to: 1. offset each other in regards to
drawdown Figure 8 is
the correlation analysis of the five markets.
Portfolio Profit Contribution So where do
the profits come from. Of course it is easy to visualize since there are
only five markets and only three return profits. If the portfolio
comprises of a lot more markets, say twenty or more, then a pie chart
like the one in Figure 9 will be very helpful.
Table 3. Portfolio Performance
Table 4. Portfolio Maximum Drawdown
We can see that when a trading system is
applied on separate individual markets with the appropriate portfolio
and money management tools, you can identify the various different
numbers that are thrown up. Some may be good while some poor. Also
revealed are the system weakness and strength. Combining the separate
systems and/or market into a portfolio, we can take advantage of the
bar-by-bar uncorrelated movement to good advantage. When we may suffer a
loss, another system within the portfolio is generating profit in an
uncorrelated manner. Overall, we may find that the risk can be lower
trading a combined portfolio than to trade and monitor the systems or
market separately. In other words risks are not incrementally added up
but cleverly sieved out by the use of proper portfolio and money
management techniques. I stated in the beginning of this article
that this is probably the most difficult to write and now at the end of
this article, I still hold the same view. What I have touched is only
very superficial. There are a lot more numbers to look at. I would very
much like to include Underwater Equity Curve that is so very important
but that will make the article too long. I would like to thank both James Goldcamp
and Leo J. Zamansky of RINA Systems who have given me useful and
invaluable pointers in portfolio and money management techniques that
help me a long way in writing this article. As a result, I have decided
that it would be helpful to you if I included in the combined portfolio
some markets that the trend system is not profitable to give an idea on
what real-life trading is like and how it will show up. I have come to the end of this series of
three articles.
Back to Let's Learn Technical Analysis Index page for more articles...
Important: These charts and commentaries are provided as an education on how technical analysis can be used. Technical Analysis & Research is not an Investment Advisor and we do not claim to be one. These charts and commentaries are not to be construed as investment advice or any other investment service other than the originally intended purpose as stated here.
|
