The last 7 years have been rough for LTTF

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
sunyata
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Sat Jun 06, 2009 4:27 pm
Location: Charlottesville, VA

The last 7 years have been rough for LTTF

Post by sunyata »

This probably isn't new to many of you, but I'd thought I'd share nonetheless. The following study uses historical performance data beginning in 1960 for a Bollinger Band Breakout system trading the standard "All Liquid" portfolio. Starting equity = $2M and max portfolio risk = 50%. No bells and whistles.

I computed the linear trend line for the monthly equity changes for different time periods. (e.g., see attached graphs.)

For time periods of 25 years or more that DO NOT include the time period from 2005 to the present, the slope of the linear trend line is slightly positive. No matter what the time period, if you include 2005 to the present, the slope turns negative.

Youch. Maybe the future will be better. Maybe not.
Attachments
Montly performance.jpg
Montly performance.jpg (223.76 KiB) Viewed 12727 times
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo »

Perhaps a better title might be "The last 7 years have been rough for this one system, trading this one portfolio, using this one particular set of S&C cost assumptions"
sunyata
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Sat Jun 06, 2009 4:27 pm
Location: Charlottesville, VA

Post by sunyata »

Yeah, that would be a much more accurate title.
sunyata
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Sat Jun 06, 2009 4:27 pm
Location: Charlottesville, VA

Post by sunyata »

Here is another illuminating graphic for the same system, trading the same portfolio. The animation depicts over time the change in the R3 statistic (y-axis) for Average Close (in days) parameter values between 120 and 510 (x-axis). I conducted each test on 25 years of data beginning in 1960 and stepped one year forward per test. For this system the R3 statistics for all Average Close parameter values decrease over time and the decrease rapidly accelerates for test-end dates after 2004.
Attachments
Bollinger Breakout - R3 Statistic per Lag over Time.gif
Bollinger Breakout - R3 Statistic per Lag over Time.gif (191.08 KiB) Viewed 12662 times
bobsyd
Roundtable Knight
Roundtable Knight
Posts: 405
Joined: Wed Nov 05, 2008 2:49 pm

Post by bobsyd »

Thanks for that Sunyata - very cool animation which tells the story well - even if the story isn't great. Would you mind sharing how you did the graphing animation and how long it took to do (after you finished the TB simulation runs)?
sunyata
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Sat Jun 06, 2009 4:27 pm
Location: Charlottesville, VA

Post by sunyata »

As sluggo reminded me it's a individual story given the system and portfolio traded. However, for anyone researching a LTTF system without dynamic portfolio selection, dynamic risk modification, and other enhancements, they probably encounter similar historical performance.

I found a free GIF animator online. To make the graphs in Excel took probably 30 min. You can save an Excel workbook as a webpage which converts the graphs to GIFs, which is really convenient.
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner »

This leads us back to the most fundamental (and probably unanswerable) questions we have been discussing for some years about markets and trend following. The original Aberration was first marketed in 1993 and at that date, the no bells and whistles version back tested very nicely over many, many years. Indeed, on the original parameters and on many starter portfolios the original “no bells and whistlesâ€
sunyata
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Sat Jun 06, 2009 4:27 pm
Location: Charlottesville, VA

Post by sunyata »

I need to drill down to market behavior for the past 50 years. Since I don't have any futures trading experience, this is pivotal. That will help me answer your first question at least and may give me insight into the viability of an adaptable system.
rhc
Roundtable Knight
Roundtable Knight
Posts: 464
Joined: Tue Nov 18, 2008 8:46 pm
Location: Oz

Post by rhc »

AFJ Garner wrote: Now you might like to make some predictions for the future! :D
If we assume that the markets will do what is necessary to confound the maximum number of participants then here’s a possible prediction for the future.
Market action will consist of volatile choppy moves. Bull moves that begin slowly and steadily will collapse quickly. Bear moves will also begin nicely then all gains given back in a few bars.
This could make trading in general and trend following, in particular, very difficult & will satisfy the above condition of confounding the maximum number of system trading market participants.

I recall that volatility breakout type systems performed very well in the early 90’s and became very popular. They were all the rage at that time. (Remember these systems? . . VBS, Tradefinder, DollarTrader, Professional breakout system etc )
If I recall correctly, at that time, the Futures truth top 10 systems were mostly Vol. Breakout type systems each with their own little twist.
The popularity of these types of systems led to their poor performance in subsequent years. (Trend following systems performed nicely though)
Again, the majority of system traders at that time were confounded.

So here’s another prediction;
VBS type systems to make a comeback as market action is better suited to them.

Final prediction (not market related);
Regardless of whom you vote for in the upcoming US elections . . . some *$%# politician will still end up winning.

Caveat:
Really, as far as the future is concerned I have zero idea.
The rule of predictions states that one should make lots of forecasts and they should be as outlandish & farfetched as possible. In this way if anyone of them should come true they will be remembered and you will look like a Guru. Then you can move some product and make tons money regardless of what the market action is.
Attachments
bold predictions.jpg
bold predictions.jpg (45.73 KiB) Viewed 12558 times
Last edited by rhc on Mon Oct 29, 2012 7:08 am, edited 1 time in total.
Aaron01
Roundtable Knight
Roundtable Knight
Posts: 113
Joined: Thu Jun 02, 2011 1:21 pm

Post by Aaron01 »

rhc wrote: I recall that volatility breakout type systems performed very well in the early 90’s and became very popular. They were all the rage at that time. (Remember these systems? . . VBS, Tradefinder, DollarTrader, Professional breakout system etc )
If I recall correctly, at that time, the Futures truth top 10 systems were mostly Vol. Breakout type systems each with their own little twist.
The popularity of these types of systems led to their poor performance in subsequent years. (Trend following systems performed nicely though)
Again, the majority of system traders at that time were confounded.
(Emphasis mine)

Fortunately it would appear that this quandary is not unique to LTTF, though it appears some people like to believe otherwise.

There are a wide variety of strategies who have seen their performance deteriorate as we trend off towards the future and the ever growing number of hands causes their edge to diminish.

The best course of action would be to pause for reflection and assess what it is that we're witnessing:
Strategy death: More and more people chasing the inefficiencies ultimately eliminate them from the market altogether and thus their profitability. More and more HFTs will eventually arb out nearly all of the bid/ask spread in all of the markets. The spread on many futures markets will likely resemble that of the Spot FX world. Perhaps this sort of death is most common with arb type strategies, but it should not be ignored for our situation.

Poor testing data: Or simply what we're witnessing is that the inferences we drew from our backtesting results and thus our expectations were inaccurate. We have been grossly underestimating the duration that certain unprofitable market conditions can exist for. We all know to expect DDs in the real world to exceed those we've encountered in backtesting, and that our real world returns will not be as high as those we've seen in our tests. Some people may not be aware that the length of our DDs will likely be longer. But even for those who do, perhaps we've been underestimating the magnitude by which they shall...
Most all of us are testing on data going back to the mid 1960s, but throughout that whole period, the period from 2008-onwards is the first available data series on a deleveraging recession/depression. Never in our testing history have we encountered such a phenomenom, the closest comparable US environment was the 1930s, how many among us have tested with such data?

Rhetorically, we all are aware of the structural differences with this recession (Why is this night different from all other nights?) but yet, paradoxically, we expect our systems to comport with the returns of very different environments. People here speak in longing tones for the wild inflationary trends of the 1970s, yet we have been living through the exact opposite economical conditions.

We can mock Harding (Winton) for gathering data back to the 1800s, but perhaps he's ahead of (us and) the curve again, in that he not only figured out this conundrum, but a way to solve it. Or maybe it's just a sales pitch...

Summarily, I don't think LTTF is dead just that we are working through economic conditions that may not be most conducive to it. Whether or not such conditions will ever return to the US, the UK or the EU is in some respects irrelevant. The ever expanding worldwide market technology and liquidity combined with declining transaction costs in the form of comms and spreads will allow a more true ability to "trade the globe."
As long as there are people existing in the market place making decisions to buy or sell, typically irrationally, there will be price discrepancies (trends) for us to exploit. The unique characteristics with which we do may have to be adjusted, but at the end of the day, we're still riding the bucking bronco...
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

unfortunatley most people have no idea of the risks currently being traded in the markets with the proliferation of "bedroom" traders, HFT's and other fully automated approaches
I recently visited a friends house - 10 years ago he was a Global Macro trader - now he has an automated FX system that trades short term - he rarely has any input and he takes advantage of those 5 decimal places in some G7 major currencies now

If people cannot see the risks associated with the above then they may want to have a long hard look at them because I believe the risks are growing by the day and "little instances" like May 2010 and Knight Capital 2012 point very firmly as to where the futuremay lie i.e. some big risks currently being under priced by market participants ( thats a first isnt it !!)

It may well be that trend following goes through a 5 year period of nothingness - as we are currently potentially seeing - but that in no way means it has failed - it means it aint working for a while

I think its true that inflationary conditions bode well for trend following and unless the laws of physics have changed then we may well see a return to ( and way beyond) the inflationary conditions of the 1980's

People are so used to current market conditions its truly frightening - every day I hear about a gigantic 1 cent move in the Euro or a huge 20 Basis point increase or decrease in some countries interest rates

One day when the Euro moves 20 big figures in a day and U.S interest rates are at 17% people may finally be reminded that being suckrered into a given set of market conditions is a trait most will continue to succumb to
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

p.s Anthony - even if you backtested over 30 years the most optimised system you can find and start with 10 milion usd you wont come anywhere close to the current level of Global Debt hehe
Algonquin
Roundtable Knight
Roundtable Knight
Posts: 158
Joined: Sun Jan 28, 2007 9:38 pm
Location: Greenwich, CT

Post by Algonquin »

The notion that LTTF has had it rough is not up for any reasonable debate. Just look at the public returns of trend followers since 2008. Most have not made money in years.
sunyata
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Sat Jun 06, 2009 4:27 pm
Location: Charlottesville, VA

Post by sunyata »

Did a quick-and-dirty analysis of historical market participation for 14 different markets. Attached is an animation that includes all the charts. Each chart is displayed for 20 seconds.

Also attached are the Coefficient of Determination (R-squared) values for linear and exponential regressions for the OI and volume of each market. On average, we are seeing exponential growth in OI and volume. Some markets, such as the grains, are seeing faster growth in market participation than others. Moreover, some of the individual R2 values may be understated due to very sharp increases in the past few years. The currencies and metals are prime examples.

The period after 2005 sees a new round of speedy growth for OI or volume, or both, following a lull in the 90s. C2, S2, KC, SB, CL, CT, DX, BP, HG, and SI2 all exhibit this pattern. Perhaps, it is HFT entering the market around this time coupled with the emergence of very large LTTF funds, such as Winton.

The effect of such increased participation is marked amounts of noise. If you look at a chart of C2 at the widest angle view, it seems very quiet--and nearly flat--for most of 1947 - 2004. Once 2005 or so hits, the noise increases dramatically and LTTF profits seem to decrease. Profits for other strategies, such as for swing trading, could increase however.

Next step is to look into Kauffman's Efficiency Ratio and how it correlates with R-multiples per trade.
Attachments
R2 Values.jpg
R2 Values.jpg (74.65 KiB) Viewed 12459 times
Historical Volume and OI Study Animation.gif
Historical Volume and OI Study Animation.gif (464.46 KiB) Viewed 12461 times
rajivm
Roundtable Knight
Roundtable Knight
Posts: 131
Joined: Fri Jul 25, 2003 12:50 am
Location: New Delhi, India

while you waiting for the LTTF returns to come back

Post by rajivm »

Well,
while you waiting for the LTTF returns to come back you can learn this guy's method of identification of trend continution and stalling to make money from micro moves of markets :D
Attachments
Very simple view of market
Very simple view of market
halftime.jpg (142.09 KiB) Viewed 12379 times
sunyata
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Sat Jun 06, 2009 4:27 pm
Location: Charlottesville, VA

Post by sunyata »

Hahahahahaha! Wow
trackstar
Roundtable Fellow
Roundtable Fellow
Posts: 76
Joined: Sun Mar 09, 2008 11:03 pm
Location: Houston Texas

Post by trackstar »

take that back to elitetrader.....
Macro
Roundtable Fellow
Roundtable Fellow
Posts: 70
Joined: Sat Dec 03, 2011 11:22 am

Post by Macro »

trackstar wrote:take that back to elitetrader.....
Well said.
rajivm
Roundtable Knight
Roundtable Knight
Posts: 131
Joined: Fri Jul 25, 2003 12:50 am
Location: New Delhi, India

Post by rajivm »

trackstar wrote:take that back to elitetrader.....
Why so!!!!!!!!! IS this method of losing money any different from systematic LTTF way of losing money ???? Both have similar Equity curve
:lol:
Post Reply