Turtle Trading Rules

Chapter 19 Turtle Trading System (2)
The smooth line that closely matches the price action is the short-term moving average, and the smooth line at the bottom of the graph is the long-term moving average.As can be seen in the chart, a long-term upward trend is in progress, so only longs can be done.The two staggered curves at the top and bottom of the price information are the breakout levels.Whenever the market makes a new high, the highest point will be refreshed directly, so the curve of the upward breakthrough level closely matches the price trend.Note that since the price is trending up, the downside breakout level (previous low) does not match the price action as well.

According to this chart, you should go long on April 4, as the price crossed the previous high of $10 on March 3 on this day.Note that the market tried and failed to break above this high in late March.This is a good example of drag in action.When the price rose to this level for the second time, it finally broke through the resistance, and it rose by 7 cents in one breath, and there was no obvious correction until $0.6802.The breakout was successful because there were no longer traders willing to sell at this price, but there were traders willing to buy at a higher price.

Timing to exit the Donchian Trend System
The Timed Exit Donchian Trend System is a variation of the Donchian Trend System that uses a timed exit strategy instead of a breakout exit strategy.It exits after 80 days with no stop loss of any kind.There are many traders who claim that the entry point is not important, but the exit point.This system is my answer to them.When we compare the performance of the various systems later, you will see that this very simple exit strategy outperforms those complex ones.

Dual Moving Average System
This is a very simple system that only buys or sells when the 100-day moving average crosses the 350-day moving average.Unlike other systems, this one stays with the market, whether long or short.The only exit point is when the short-term moving average crosses the long-term moving average: at this point, the trader exits the previous trade and starts a new trade in the opposite direction.Figure 10-4 is the moving average of the dual moving average system.

The daily moving average is more in line with the price action, and when it crossed the 7-day moving average upward in late July, traders took long positions.You may find that your system is a fairly long-term trend-following system that trades less frequently than most other systems.

Triple Moving Average System
这个系统使用三种移动均线：150日、250日和350日均线。交易者在150日均线穿越250日均线时买入或卖出。最长期的350日均线扮演的是趋势过滤器的角色。只有150日和250日均线位于350日均线的同一侧时才能交易。如果两者都高于350日均线，只能做多；如果两者都低于350日均线，只能做空。

The top one is the 150-day moving average, the middle one is the 250-day moving average, and the bottom one is the 350-day moving average.It can be seen that all three lines rise slowly as prices rise over the same time period as in Figure 10-4.If the top moving average turns around and crosses down through the middle moving average, the system exits the trade.

Before we move on to the next section, take a guess as to which of these systems fared better during this test period.How much worse does a timed exit system perform than a normal breakout exit strategy?Which two systems have the highest MAR ratio?How much better is the triple moving average system than the dual moving average system?
Test Results
我用同样的数据（同样的资金管理法则、市场组合和测试期）和我们自己的交易模拟软件Trading Blox Builder对6个系统作了测试。这个软件模拟了每一个系统在1996年1月~2006年6月的每一次交易，并据此生成了业绩统计数据。表10–1列出了每一个系统的一些最主要的业绩指标值。

ATR通道突破49.5%1.241.3420642.2%39.9%8.3布林格通道突破51.8%1.521.5213054.6%34.1%7.8唐奇安趋势29.4%0.800.99183239.7%36.7%27.6唐奇安定时57.2%1.311.3574658.3%43.6%12.1双重均线57.8%1.821.5521039.5%31.8%8.3三重均线48.1%1.531.3718142.5%31.3%8.5当我第一次测试定时退出策略时，我大吃一惊。它们的表现比我想象的好得多，甚至比突破法退出策略还好。如果说一个系统的威力在于它的退出策略，这就是最好的证据。这说明，一个系统的赢利能力并不完全依赖于一个有优势的入市策略。

Note that the Donchian system does not perform as well as the other systems.This shows that the breakthrough method has lost some of its advantages in the years since Project Turtle was implemented.I think this is largely due to the trader effect.I'm going to talk about this effect.

Another surprise is the excellent performance of the double moving average system.Note that this system outperforms the comparable but more complex triple moving average system.This example proves that more complex systems are not necessarily better, and there are many more.

These are fundamental systems.Among them, the three systems of double moving average, triple moving average and Donchian trend do not even have any stop loss exit standards.This means they violate one of the most revered adages of trading - always have a stop.Yet their risk-adjusted returns are comparable to, or better than, other systems.

Do you want to add a stop loss point?

Having no stop loss exit criteria at all would make many traders uneasy.How would the performance of the dual moving average system change if we added exit points?A lot of people like to think about these kinds of things.They take these questions to ask their friends or more experienced traders.

For me, I prefer to have faith in an idea if I get a solid yes after testing it.Confidence is good for you.Figure 10-6 reflects the MAR ratio under different stop loss standards, and the horizontal axis represents the stop loss standard, with ATR at the time of market entry as the unit.

Note that the MAR ratio is highest when the stop loss criterion is zero, that is, when there is no stop loss point at all.In fact, this conclusion holds for all indicators: average compound growth rate, MAR ratio, Sharpe ratio, decline, decline duration-every one.The same goes for the triple moving average system: if you add a stop loss, every indicator will go bad.The stop loss test of the timed exit Donchian trend system also came to almost the same conclusion, with one exception: for large stop loss standards above 10ATR, the system performed similarly to when there was no stop loss point.This of course is very contradictory to the general belief that "there should always be a stop loss point".Why?Isn't the stop loss point very important to preserve capital?Now that we've added a stop loss, why doesn't the decline level go down?
Many traders believe that the most important thing to worry about is a series of losses.This may be true for short-term traders whose trades last only a few days, but not for trend followers.For trend followers, declines also come from trend reversals, which usually occur after a major trend ends.Sometimes the market following a trend reversal can be very volatile and tricky to maneuver.

The Turtles know that giving up some of the profits accumulated in the trend is only part of the game.We know that sooner or later we are going to experience a huge decline.But it's a real pain in the ass for some turtles, especially the ones who can't stand losing money the most.Watching newly earned money evaporate into thin air is one of the most unbearable aspects of our trading style.

Therefore, the decline of trend followers does not come from the risk of entry, but from the return of profits.We'll discuss this in more detail, but let's go back to the test results for the six systems.

As mentioned earlier, our testing period ended in June 2006.As I write this, many more months have passed, and you may be curious about the situation during this time.

Which system would you choose if you were based on data as of June 2006?If you could choose two systems, which two would you choose?To give you some inspiration, I modified the test cut-off period and added new data up to November 6. The updated results are shown in Table 2006-11.

ATR通道突破45.9%1.151.2721643.1%40.0%8.3布林格通道突破49.2%1.441.4713653.7%34.1%7.8唐奇安趋势27.4%0.750.94190138.7%38.7%27.6唐奇安定时57.1%1.311.3477359.1%43.6%12.1双重均线49.1%1.041.3422236.9%47.2%8.3三重均线41.2%0.971.2118641.9%42.3%8.5稍微看一眼CAGR和MAR比率就能发现，趋势跟踪系统在2006年的最后几个月中整体表现不佳。有趣的是其中所发生的变化，表10–3列出了CAGR和最大衰落的变化百分比。

ATR通道突破45.9%49.5%–7.3%40.0%39.9%0.3%布林格通道突破49.2%51.8%–5.0%34.1%34.1%0.0%唐奇安趋势27.4%29.4%–6.8%38.7%36.7%5.4%唐奇安定时57.1%57.2%–0.2%43.6%43.6%0.0%双重均线49.1%57.8%–15.1%47.2%31.8%48.4%三重均线41.2%48.1%–14.3%42.3%31.3%35.1%这是怎么回事？测试结果为什么会发生这么大的变化？我们的最佳系统的衰落幅度为什么会上升50%之多？为什么使用最简单的定时退出策略的系统在最后的5个月中没有什么变化，但其他系统却如此糟糕？一个交易者如何建立起更符合期望值的系统？换个角度说，你怎么才能让你的期望值与一个系统的潜力更加吻合？

These questions serve as an introduction to the next chapter.We will explore these questions to help you better understand the difference between historical test results and actual trading results, and to better understand the various factors that cause this difference.

(End of this chapter)