Dennis' Turtles | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Richard Dennis in Modern Times 3 replies

I need a mechanical system like the turtles 2 replies

Attachments: Dennis' Turtles
Exit Attachments

Dennis' Turtles

  • #1
  • First Post: Jun 22, 2008 4:07am Jun 22, 2008 4:07am
Hi all

Some time ago I downloaded a pdf which one of the original turtle traders put together and posted online for anyone to download. The idea was to stop one of the ex turtles, who failed trading the turtle system, from selling the system. This PDF has a step-by-step method for how they entered and exited trades. It's a simple trend following system.

Has anyone else downloaded the pdf, and has anyone tried to use the strategy? I haven't tried it yet.
  • #2
  • Jun 22, 2008 4:19am Jun 22, 2008 4:19am
Yes and yes. The whipsaw is horrible... but it does work over large enough time frames. I was told that using a momentum filter helps some, I haven't really tried that tho.
  • #3
  • Jun 22, 2008 7:53am Jun 22, 2008 7:53am
Remember that the system was designed largely for the commodities markets back in the 80s.

You'd just need to change the period on the donchian channels in order to accommodate for the forex market.
  • #4
  • Jun 22, 2008 5:58pm Jun 22, 2008 5:58pm
Quoting Ronald Raygun
Remember that the system was designed largely for the commodities markets back in the 80s.

You'd just need to change the period on the donchian channels in order to accommodate for the forex market.
What do you change the period on the donchian channels to?
  • #5
  • Jun 22, 2008 6:00pm Jun 22, 2008 6:00pm
Wasn't it based on a breakout of 20 days and or 55 days?
  • #6
  • Jun 22, 2008 6:05pm Jun 22, 2008 6:05pm
Quote
Wasn't it based on a breakout of 20 days and or 55 days?
Yes, with an ATR-based lot sizing. According to Van Tharp, 40 day variants still work pretty well... I've never had much luck w/ any of those approaches in forex. I do use a 10-day high/low as a "hey, dude, there might be trend here" signal, but I don't trade it the way they do.
  • #7
  • Jun 22, 2008 6:17pm Jun 22, 2008 6:17pm
I think the system referred to in the PDF does not work so great anymore but the ideas in there are useful -

Amazing that these guys were trading such huge size with a very simple volatility breakout system. The most complex element is the position sizing that you mentioned earlier.

Worth noting that both Richard Dennis and Curtis Faith had major blowouts in their careers.
Attached File(s)
File Type: pdf turtlerules.pdf   271 KB | 864 downloads
  • #8
  • Jun 22, 2008 6:42pm Jun 22, 2008 6:42pm
over at tsd there is an ea based on the system
  • #9
  • Jun 25, 2008 7:43am Jun 25, 2008 7:43am
Quoting dimitrius
What do you change the period on the donchian channels to?
I've been using 40 for the forex market. The main reason for doing so is to detect the establishment of a trend. I figure that a two month high is some sort of trend with some false breakouts fit for a turtle.
  • #10
  • Jun 25, 2008 8:39am Jun 25, 2008 8:39am
So 40-bar has worked for you too? Do you actually trade that way?

FYI for those that don't know: Donchian channels are just an X-period High/Low indicator. Found an MT4 indi, attached it here. Basically all you'd do is set it up and wait for a bar to break out above/below the channel and take that trade... at least according to the Turtle rules.

.
Attached File(s)
File Type: mq4 Donchian Channels.mq4   3 KB | 769 downloads
  • #11
  • Jun 25, 2008 8:57am Jun 25, 2008 8:57am
Quoting Rabid
So 40-bar has worked for you too? Do you actually trade that way?

FYI for those that don't know: Donchian channels are just an X-period High/Low indicator. Found an MT4 indi, attached it here. Basically all you'd do is set it up and wait for a bar to break out above/below the channel and take that trade... at least according to the Turtle rules.

.
They're working for now, but I've only had 7 entries in 3 months, so I can't be absolutely sure. Backtesting shows turtle-like order accuracy. (long string of losers, one large winner. But given the frequency of signals, I'm not committing cash yet.

On a side note, EURUSD seems to be the best pair for this type of trading. As long as it's trending. One of my entries happened during the major rally in Feb.
  • #12
  • Jun 25, 2008 9:26am Jun 25, 2008 9:26am
Quote
They're working for now, but I've only had 7 entries in 3 months, so I can't be absolutely sure. Backtesting shows turtle-like order accuracy. (long string of losers, one large winner. But given the frequency of signals, I'm not committing cash yet.

Nod. I've noticed that most trends don't occur in a vacuum, but rather seem to center around a particular currency. For instance the EURUSD's trend was due to USD weakness, which led to trends against the USD for quite a while.

The same thing is happening to JPY-based pairs right now. Have you tried this system w/ JPY pairs any?

My biggest problem w/ the turtle method is the rather blunt entry approach, it's simple but leads to really big stop losses... which ends up leading to deeper periods of drawdown. I suspect if the entry was refined a bit and taken intra-day the S/L could be tightened a lot and the drawdown could be reduced or maybe even eliminated.
  • #13
  • Jun 25, 2008 9:38am Jun 25, 2008 9:38am
Quoting Rabid
Nod. I've noticed that most trends don't occur in a vacuum, but rather seem to center around a particular currency. For instance the EURUSD's trend was due to USD weakness, which led to trends against the USD for quite a while.

The same thing is happening to JPY-based pairs right now. Have you tried this system w/ JPY pairs any?

My biggest problem w/ the turtle method is the rather blunt entry approach, it's simple but leads to really big stop losses... which ends up leading to deeper periods of drawdown. I suspect if the entry was refined a bit and taken intra-day the S/L could be tightened a lot and the drawdown could be reduced or maybe even eliminated.
This system was developed to capture huge trends. You saw the results for yourself, and you should know the terms under which the turtles achieved those results. Why change that?

As for other pairs, I have done visual indicator-based backtesting. They need something other than a 40 period donchian trend largely because some pairs are more responsive than others.

Remember, consider the trading situation about 25 years ago. Volatility was pretty even across markets, so the standard 20-day donchian channel could apply across markets. Nowadays, that's not always the case.
  • #14
  • Jun 25, 2008 10:00am Jun 25, 2008 10:00am
Quote
This system was developed to capture huge trends. You saw the results for yourself, and you should know the terms under which the turtles achieved those results. Why change that?

Well because there aren't very many huge trends compared to small trends, and a lot of the turtles washed out. Today most of those traders don't even use the same system, which tells me that changes need to be made somewhere. It does perform over very long periods of time, years and years, but try running the system against less trending pairs, or during non-trending patches of good pairs, and it'll get eaten up.

Psychologically, large and long periods of drawdown just don't fit well into my trading. But I do know of ways to eliminate large chunks of that while turning profits on smaller trends.

Quote
As for other pairs, I have done visual indicator-based backtesting. They need something other than a 40 period donchian trend largely because some pairs are more responsive than others.

Nod, other pairs certainly behave differently. That leads to curve-fitting, trying to make the system fit the pair. IMO a good system should be robust enough to operate w/o any curve fitting, which is yet another reason why I look for ways to improve it.

Quote
Remember, consider the trading situation about 25 years ago. Volatility was pretty even across markets, so the standard 20-day donchian channel could apply across markets. Nowadays, that's not always the case.

Well, I don't know if that's true or not. There have always been patches of volatility across sectors. I don't know if you can relate this system's success to volatility as much as you can relate it to trend persistency, and I don't have a standard metric for measuring trend persistency over time.

But the forex market is a whole different beast than stocks, forex is far more volatile. Techniques that work great on stocks and commodities (that have a fixed period day) don't always work so well in forex, and those that do tend to need a lot more micro-management of risk. Trend persistency in forex is a lot more predictable than equities too, it's almost always linked to changes in core fundamental metrics (interest rate, budget deficit, trade deficit, inflation, employment). It's also easy to verify, since there are so many pair combinations it makes currency strength/weakness easy to spot.

To that extent I suspect (strongly) there may be ways to reduce drawdown and improve responsiveness while eliminating the need for curve fits. So that is why I would change it.
  • #15
  • Dec 27, 2008 8:55am Dec 27, 2008 8:55am
FWIW, I added a Turtle EA to this thread: http://www.forexfactory.com/showthread.php?t=140805
  • #16
  • Dec 28, 2008 11:12pm Dec 28, 2008 11:12pm
Thanks longtobefree.
  • #17
  • Dec 28, 2008 11:14pm Dec 28, 2008 11:14pm
Quoting dimitrius
Thanks longtobefree.
Not a problem. It was fun to write.
  • #18
  • Last Post: Dec 30, 2008 6:24am Dec 30, 2008 6:24am
Quoting Rabid
Well, I don't know if that's true or not. There have always been patches of volatility across sectors. I don't know if you can relate this system's success to volatility as much as you can relate it to trend persistency, and I don't have a standard metric for measuring trend persistency over time.
Exactly.

More volatility may not = more false breakouts.

The real beauty in these sorts of systems is that they WILL capture trends...but identifying trends will always be the tough part, particularly if the trader would like the satisfaction of joining the trend *early*......
0 traders viewing now