• Performance
• Services
• Company
• Learn About
• My Attain

The R-Mesa 5 Drawdown - When should you stop trading?

February 21, 2006

Poor R-Mesa 5? The system which hung its hat on consistency since its release in 2002 is in the midst of its worst drawdown since release. Since Attain began tracking the actual client fills on the system in August of 2002, R-Mesa has had an uncanny ability to make new equity highs - hitting 10 new equity highs in the first 27 months after it was released.

The pattern seemed to be a new equity high, then about 3 or 4 months of losses down to a drawdown low, then another 2 or 3 months up to a new equity high. In fact, when excluding the months where the system made new equity highs in consecutive months - the average time between equity highs was just a little over 5 months.

But here we are sitting at 16 months since the last equity high and a new post-release Max DD of -$28,200, or 94% based on the developer's recommended initial balance of $30,000; leaving investors asking what is going on with R-Mesa.

The numbers don't look all that good - R-Mesa is down -35% YTD (-46% after today's $800 winning trade), and is in the midst of an 16 month, 94% drawdown which started in early November of 2004 and hit a new low last week on Thursday.

These numbers are far from the average annualized gain of 41% the system sported coming into 2005, and investors want to know…..Is it time to worry? Is it time to stop? The short answer is it is definitely time to worry, and very close to being time to stop. The rule of thumb is to always expect the worst drawdown will occur in the future. But this begs the question - when faced with a new max drawdown, when should you stop trading this or any other system?

Setting a "Line in the Sand"

At the risk of sounding too vague, the question of when to stop trading a system is a personal one. Each investor utilizing R-Mesa or any system for that matter, should have a "line in the sand" which they will not go past. This is not the Nasdaq in 2001. We're not interested in trading something all the way down to zero, thus it is extremely important to know what your max loss point is BEFORE getting involved with a trading system.

But where do you set that line in the sand? Usually not a the pre-tested Max DD, as we can expect that a new Max DD will occur in the future. But then how far past the old Max DD. What is a normal excursion beyond the previous max drawdown, and what level of new drawdown might signal that the system is "broken" and should be reevaluated?

Let me first say that system's don't "break". No system starts throwing out rogue trades and losing more than it was designed to because of some broken logic or flawed code. (At least no system executed for clients at Attain). I believe what many investors refer to when saying a system "breaks" is when the risk profile moves severely outside of what was expected in first getting into the investment. If one of the premises for investing in the system was a Max DD of only 40%, for example, then the possibility of future drawdowns of 75% should cause the investor to reevaluate.

So if systems don't break, what do they do? They simply become much more risky. So reevaluating your system investment doesn't necessarily mean stopping it, rather analyzing whether you're comfortable with the added risk. The new profile of the system may simply mean a higher allotment of capital to the system or a reduction in exposure to the system (i.e. 2 contracts down to 1). If the system's overall metrics remain very good, stopping a system simply because it has eclipsed its max DD could do more harm than good. For example, the Compass system was in similar territory in the middle of 2004, only to come roaring out of the Drawdown to post new equity highs three times in 2005.

Measuring Future Drawdowns:

But let's return to the question of where should you put that line in the sand? How far past the old drawdown is too far past? Attain Capital has several statistical methods to calculate the "line in the sand", but setting the bar at 1.5 times the predetermined, tested, historical drawdown is a good starting place that gives a system plenty of "breathing room".

The 1.5 times the past historical DD admittedly has issues in so far as it has no statistical basis. (Why not 1.35 times, or 2.4 times, etc.), and for that reason Attain uses several other measures to measure expected future drawdowns. Again, we should always expect a new max DD in the future, thus want to be prepared for that eventuality when it arrives by calculating an estimated future drawdown.

The first method of measuring a possible future max DD was to run a Monte Carlo simulation using monthly R-Mesa data. The simulation, run 5000 times over 1200 months (100 years), randomly shuffles existing data to give probabilities of future drawdowns exceeding certain levels. We plugged in all of the hypothetical backtested results prior to the system's release, and our simulations told us that 99% of the simulation trials saw drawdowns less than 101.21% or ($30,363). This tells us that the current drawdown of ($28,200) is still below the "line in the sand" generated using a Monte Carlo engine. For more information how Monte Carlo simulations work, see our past newsletter on the subject using this link: http://www.attaincapital.com/alternatives/alt_feb2805.htm#Topic.

The next method was to run an empirical test on available data using the mean, standard deviation, and skew of monthly returns to calculate a drawdown estimated to occur with a 1 in 100 year frequency. These tests allow us to calculate both a Max DD in dollar terms and an estimated max drawdown duration. The empirical test showed we are slightly past the 1 in 100 year levels, with the estimated future max drawdown (or stop trade point) having been hit about $800 ago.

Finally, we put the six sigma test into effect. To achieve Six Sigma confidence, a process must not produce more than 3.4 defects per million opportunities. Six Sigma standards are common in the production of microchips and high level security software, where the margin for error is razor thin, and in short insures there are six standard deviations between the mean and the nearest specification limit. In trading systems, the mean is your average drawdown, and the specification limit is the max DD you never want to see. The Six Sigma "line in the sand" for R-Mesa 5 is whopping Max DD of -124.62% or ($37,386), which gives the current drawdown another $9,000 or so to go.

So here we stand, with R-Mesa having lost significantly last year, and starting off much the same this year. By all accounts it is a time to worry. But, as shown in the table below, the system has another $2,000 or so to go before reaching the average drawdown levels we use as potential "lines in the sand" for stopping trading.

While the room, however small, between the current drawdown level and the stop trade "lines in the sand" is a good sign in so far as it means R-Mesa is operating within its tested parameters, I am guessing the specter of an R-Mesa DDs as high as 100% of the initial account balance will catch more than a few people off guard. Could it be that the starting point, the developer's recommended initial balance of $30,000 is too low for R-Mesa SP. Definitely.

With the future Max DDs of R-Mesa averaging over 103% of the $30,000 initial balance, it becomes abundantly clear that a conservative investor should use a value greater than $30,000 per contract for their initial investment in the R-Mesa trading system. Attain Capital recommends starting with at least $50,000 per full size S&P contract for those investors trading the system on its own without any diversification amongst other systems.

The table below shows the current R-Mesa Drawdown as it compares to the pre-release Max DD, and the estimated stop trade/reevaluation levels (lines in the sand) as derived from each of our four methods discussed above, with the average of the three shown. As can be seen, the 1.5 times tested max DD is a relatively valid measure, as it quickly and easily gives an estimated future drawdown close to the more mathematically advanced methods. Investors can choose which method they wish to ascribe to, or use the average of the four to measure what point they should stop trading R-Mesa or reevaluate their trading.

R-Mesa Stop Trade Calculations

What about trading R-Mesa differently, R-Mesa 3?

The R-Mesa drawdown has led many investors, including Attain, to research ways and methods to improve its performance.

With the ever-shrinking volatility in stock index futures - one idea was to eliminate the filter that keeps R-Mesa 5 from trading on the 3rd and 4th Fridays of every month (option expiration days) and on FOMC meeting days. The thinking went - if R-Mesa is struggling because of a lack of volatility, why filter out those days which tend to have more volatility. The results did not improve things, unfortunately; with the "no exceptions" R-Mesa having greater losses in the low volatility years of 2004 and 2005.

Another method was to look at R-Mesa 3, which was released well prior to R-Mesa 5 and in theory has a base logic that is more pure and not fit to the higher volatility of the "bubble burst" era. The testing on R-Mesa 3 showed little improvement, however; and underperformed R-Mesa 5 over the past two years as well.

Finally - Attain applied some of its own filters to R-Mesa, including a Rate of Change filter which would only have the system enter in line with the predominant trend in the market. These filters cut the drawdown in half, but cut the profit by an equal amount, leaving the system nearly the same on a risk adjusted basis. Moving the filter onto shorter and shorter time frames increased the risk adjusted profitability, however, and we continue testing this filter.

At the end of the day - we continue to believe in R-Mesa 5, and will do so for another $2,766 in losses. If the system does eclipse that line in the sand, we will no longer recommend the system and advise clients to stop trading it - but at this point, stopping the system is not advised. Stopping at this point would lock in much bigger losses only to protect another $2,700, making the risk much smaller than the reward at this point.

Low risk, high reward "Trade" Opportunity

For those investors not currently trading R-Mesa 5, this drawdown level represents an excellent "trade" opportunity. This is an equity curve "trade" where you are buying the system at an extreme discount. The current level represents an option play of sorts that costs $2,700, with an unlimited upside. The system is paid for out of profits, and the downside risk is just $2,700 to the stop trade point; meaning you only have to risk $2,700 for the chance to make over $40,000 (if the system makes back the $28,000 drawdown plus earns the $15,000 average annual return it saw before the current drawdown). Making over $40,000 on a $2,700 investment represents a return over 1,300%.

- Jeff Eizenberg