The turtle system made many turtles rich in the 1980s. Today, it has been said that the trading system no longer works. Which parts of the strategy still work and which parts have failed? Why has the system malfunction today?
I had a quick look at http://www.metastocktools.com/downloads/turtlerules.pdf (been a few years since I did). It's a very simple system, and all of it makes sense. Without backtesting it's hard to say but I would imagine this would still work okay. I'd make the following changes if I was to update the system: - reduce position size by 4-8 times, as it's too aggressive - change the capital rescaling accordingly - Add more markets, assuming you have enough capital - Trade a wider range of entry signals, targeting different frequencies - Simplify the system by removing the exits, but keeping the stops (or vice versa) This wouldn't change performance very much, but would make it safer. GAT
We know how many it made rich but we might not know how many it made poor. This and similar systems took advantage of serial correlation that decreased after the 1990s significantly due to arbitrage algos.
The part where you identify a system that works and stick to it through thick and thin, using proper money management/sizing, yes, that still works. There are many systems that work, but few people who have the psychological bent to follow them. An understanding of probability is helpful. Most nonprofessional gamblers, a category into which may traders fall, do not understand probability. For some, they don't want to understand it.
Newedge (now SocGen) publishes a trendfollowing index based on published, mechanical rules. It won't track any given trendfollower exactly, but it's a quick way to find out how the wind is blowing. Yearly summary, from data from the linked Excel sheet... 12/29/2000 14.2% 12/31/2001 -4.0% 12/31/2002 35.0% 12/31/2003 13.1% 12/31/2004 -0.8% 12/30/2005 -1.6% 12/29/2006 6.1% 12/31/2007 -1.5% 12/31/2008 31.4% 12/31/2009 2.9% 12/31/2010 7.9% 12/30/2011 1.5% 12/31/2012 -16.0% 12/31/2013 -17.7% 12/31/2014 39.1% 8/13/2015 -7.8% ... up ~10% in the seven years post- 2008 crash (2009- 15) versus ~80% in the seven prior years (2002- 08) http://www.newedge.com/en/newedge-indices/ The Newedge Trend Indicator is a market based performance indicator designed to have a high correlation to the returns of trend following strategies. Bloomberg: NEIXTRND Index Downloads: Research Note | Historical Returns
This is usually (but not exclusively) due to a low winrate. Many trading pundits dismiss winrate because it is not solely determinative of expectation. But winrate does matter psychologically. It's not easy sticking with a system that loses more often than it wins. Unless you're an android or a Vulcan. Which the vast majority of traders are not. AFAIK the turtle system even in its heyday was/is a low-winrate system. Takes a lot of discipline to stick with it thru thick and thin.
one problem I had with that interesting article was this: Their developers falsely attributed the apparent autocorrelation in the equity markets that basically turned profitable any method based on buying on strength to some predictive capacity of these methods. Essentially,they were fooled by randomness. Its amazing how many people read that book or see that title and now wish to feel everything is random. Perhaps... traders made markets auto correlated in the 60s through 80s and electronic trading allowed other traders to erase the autocorrelation. In fact having daytraded through the change over I can tell you... things shifted dramatically as the qqqs and naz 100 and the s&p electronic markets got liquid. All of sudden the spreads on our favorite daytrading NYSE stocks got much tighter and their intraday trends got much shorter. At the same time all these traders started talking about reversion to the mean trading and raising money to do stat arb. What happened to the markets was not a return to random. It may have looked that way... but it was traders and computers arb trends out of mature markets. IMO.
Jem, check out his new Momersion indicator. P.S. Actually I thought he meant that the patterns were random, not the price action. Since he talks about +autocorr he could not have meant that price was random.