Why isnt "Turtle System One" (1984) profitable?

[sluggo]

(Moved from a different thread)...

Jas-105 posts: What do you think has made Turtle System One lose its profitability , how have the markets changed ? Is it just more short-term volatility or "noise" or has it been an increase in trading systems that pick-off longer-term systems , or something else ?

Anthony Garner has a nice article in the Feb 2010 issue of Active Trader magazine which discusses this: "Tuning up the Turtle System". He suggests a few modifications to the system, which result in a much improved equity curve post-1993. His modifications include

• Widening the Donchian (breakout) channels
• Widening the initial stop
• Enlarging the portfolio
• (To go with the bigger portfolio): install Max Sector Risk and Max Portfolio Risk limits
• Switching from scaling-in ("pyramiding") to scaling-out ("profit target exits")
• Adding a Very-Long-Term-trend filter which only permits trades in the direction of the VLT trend

Unfortunately, it appears Active Trader makes .pdf reprints available on their website approximately 3 months (!) after publication in the magazine.

There have been lots of discussions here about "what happened to the Turtle System's performance" and/or "have the markets changed". Here are a few of the threads to get you started. You can find many more by performing a little Google-Fu (in Advanced Search, fill in the bottom box "Search within a site or domain" with tradingblox.com)

viewtopic.php?t=5428

viewtopic.php?t=3637

viewtopic.php?t=3779

viewtopic.php?p=40810

[DeanoT]

If I were creating a trading system to stand the test of time, I would want to create one which had the most substantial edge, or positive expectation possible. This would be my best bet against a system eroding to the point of break even, or worse.

When you review trend following trading systems over the past 30 years, you notice that the ultra-long term systems have continued to perform well, while the short to medium term systems have begun to fail. This tells me that trend following remains profitable today, but that your timeframe is key. The shorter your timeframe, the smaller your edge, and the less safety margin you have in your favour.

This, in my opinion, is the reason for the failure of the Turtle system over the past 15 years. Richard Dennis even alluded to the failure of medium term trend following systems in Market Wizards with his comment "be as long or as short term as you like, but avoid the middle like the plague". I don't know if Dennis thought of the Turtle system as short or long term (I assume he meant short), but in any case, it has become clear, by my standards at least, that traditional short term trend following systems are a thing of the past. And if Dennis knew the middle was finished in the late '80s, the the only thing left is the long term.

Maximise your expectancy, would be my best advice to warding off the potential for system death.

[AFJ Garner]

I agree with all Deano T has to say..................except..........

Let us assume the date is 1970 and we are running back tests with dinosaur computers and we have enough data and enough markets. Maybe we would NOT make the same conclusions we are making today? I don't know. I am playing devil's advocate. I am saying that I have spent many, many hours back testing and cogitating over the past decade and as of today's date I deem it meet and right to use very long term filters and to adopt a very long term system.

If I had been around and done the same thing in 1970 what would I have found and what conclusions would I have come to?

[nodoodahs]

Hmmm ...

I read about an old European "hedge fund" manager who passed along this "system" to a n00b many years ago. Basically, he invested in countries only if both their bonds and their overall stock index were "in trend" defined as the 50dma being above the 200dma. He used relative strength to decide which countries were preferable out of the ones that met criteria.

From the timeframe of the book and the author's experience, I would judge that the old guy probably had a backtest ending the 1970s.

I would venture that such a "system" is not TOO unlike something you'd enjoy ...

[PaulZ]

I put the .tbz file in my Import folder but when I start TB I still don't see the system show up in the list. I am running Builder 2.2.12. Is that the problem?

[sluggo]

Must be. I don't have a 2.2.12 lying around, but I did try it on 2.2.9 and got the same result as you.

[AceofAce]

[quote="AFJ Garner"]I agree with all Deano T has to say..................except..........

Let us assume the date is 1970 and we are running back tests with dinosaur computers and we have enough data and enough markets. Maybe we would NOT make the same conclusions we are making today? I don't know. I am playing devil's advocate. I am saying that I have spent many, many hours back testing and cogitating over the past decade and as of today's date I deem it meet and right to use very long term filters and to adopt a very long term system.

If I had been around and done the same thing in 1970 what would I have found and what conclusions would I have come to?

[/quote]

Like many others i believe that increased noise of the late 90s and 2000s contributed vastly to the STTFs and MTTFs becoming obsolete or less profitable whilst LTTF appears to stand as yet (scarred though)

Could there ever be a case when STTFs becomes profitable again and at the same time LTTFs become obsolete or it the former a subset of the latter.

Any opinions are most welcome...

[Chris67]

Absolutely there could be that situation - and my personal view is we are on the verge of it
I think those s/t systems that performed so well like the orginal turtle systems thrive in markets where you have a high inflationary environment or ballistic moves - given the fact that we MUST have one of those 2 imminenly or the World is in big trouble - then those systems will prob work again - BUT the L/T systems will prob do well too so why not cover yourslf and run L/T systems !!!
This whole are of discussion boils down to the fact we dont have enough data to make any conclusions. 30 or 40 years of data is sod all in the big picture - 250 years would be more persuasive to enable calls about what systems failed and what havent ?? Unfortunately - for us - will never happen that we have that much data but in the meantime I think we should be less arrogant and accept the fact we cannot make big calls about the permanent failure of any type of trend following system
C

[LeviF]

I doubt there will be a time that long term fails and short term prevails. And if long term fails most of us are going to have to find another line of work.

[AceofAce]

I guess that could happen if for many years most trends become very short lived (i.e. Mostly Frequent reversals) to the point that STTF can capture some Profits but not LTTF!?

[innertrader]

JUST A THOUGHT: To answer your question I would suggest checking the correlation of the marketing of the Turtle Method by Russel Sands and it's downfall. Also, as a professional trader, I and others began "hunting" for the stops... so to speak.
There was also a time in the 70s and 80s when I used the Pivot Point system... but "then" there were only TWO of us using it; today, it's mostly useless.
By the way, I read a book printed in the 1930s that virtually described the so called original Turtle trading method to a "T"... makes one wonder how many times it's been used, abused and abandon.