Long term trend following on equities a fool's game?
Long term trend following on equities a fool's game?
Should someone starting out testing lttf systems NOT start with stocks?
In other words, are stocks the hardest instruments to trade profitably using lttf methods? If so, then futures or currencies?
The reason I ask is because no matter what parameters I use I don't get good results. Using the most liquid us stocks.
Thanks for the input.
In other words, are stocks the hardest instruments to trade profitably using lttf methods? If so, then futures or currencies?
The reason I ask is because no matter what parameters I use I don't get good results. Using the most liquid us stocks.
Thanks for the input.
This is a the equity curve of a system we are building for Canadian stock,
So far the result are better on stock system than on Future. This is a Long term system, scanning around 1400 stocks every day. The system hold a maximum of around 10-15 stocks at any time.
So I guess the result is yes lttf work in Stocks.
So far the result are better on stock system than on Future. This is a Long term system, scanning around 1400 stocks every day. The system hold a maximum of around 10-15 stocks at any time.
So I guess the result is yes lttf work in Stocks.
- Attachments
-
- lttfstock.jpg (86.66 KiB) Viewed 20710 times
Very convincing data
Thank you both for sharing. I appreciate it.
Maybe the question should be does LTTF work on liquid stocks (e.g. SPY 500) which perhaps have a much higher correlation with each other than small cap. names.
I'm not sure how feasible any system looking at 1400 Canadian names will be. There are only perhaps 200 Canadian names tradable in any size. Beyond that slippage is going to be an issue and many names will be impossible to short.
Also, if memory serves me correctly, the Black Star Funds study is long only and covers pretty the 1990s onwards which was obviously a pretty strong bull market. I would be curious to see how the strategy performed during the 1970s.
I'm not sure how feasible any system looking at 1400 Canadian names will be. There are only perhaps 200 Canadian names tradable in any size. Beyond that slippage is going to be an issue and many names will be impossible to short.
Also, if memory serves me correctly, the Black Star Funds study is long only and covers pretty the 1990s onwards which was obviously a pretty strong bull market. I would be curious to see how the strategy performed during the 1970s.
-
- Roundtable Knight
- Posts: 115
- Joined: Sun Nov 20, 2005 12:02 pm
- Location: Westport
-
- Roundtable Knight
- Posts: 126
- Joined: Mon Nov 07, 2005 3:39 am
- Location: Silicon Valley, CA
- Contact:
For those interested in this topic there is a lot of research in this area especially with momentum strategies which is a form of trend following.
Earlier this year, February 2008, ABN AMRO released their global investment yearbook with a study of the largest 100 UK stocks and momentum trading showing that on average about 11% yr return based on a long/short strategy. The years covered for that part of the study was 1956-2007. For 1900-2007 the average return was ~10% - yes a 108 year study of the top 100 UK stocks. Your mileage and markets may vary. Other markets are covered in the study too - notice there are -40% down years so this method is not without a lot of risk.
I have attached a synopsis of the report - the full yearbook for non-customers is ~150GBP.
Earlier this year, February 2008, ABN AMRO released their global investment yearbook with a study of the largest 100 UK stocks and momentum trading showing that on average about 11% yr return based on a long/short strategy. The years covered for that part of the study was 1956-2007. For 1900-2007 the average return was ~10% - yes a 108 year study of the top 100 UK stocks. Your mileage and markets may vary. Other markets are covered in the study too - notice there are -40% down years so this method is not without a lot of risk.
I have attached a synopsis of the report - the full yearbook for non-customers is ~150GBP.
- Attachments
-
- GIRY2008Synopsis.pdf
- (205.27 KiB) Downloaded 1229 times
The start date was dictated by data integrity & availability. Getting accurate dividend, deslisted stock, and index total return data before the early 1990's was challenging. That being said, I have results for the time period Jan-1970 to Dec-1979.arb wrote:Also, if memory serves me correctly, the Black Star Funds study is long only and covers pretty the 1990s onwards which was obviously a pretty strong bull market. I would be curious to see how the strategy performed during the 1970s.
--Long only all time highs + 10 ATR stop--
CAGR: 4.3% , St_Dev: 14.3%, Max_Drawdown: -27.2%
--S&P 500--
CAGR: 3.9% , St_Dev: 15.9%, Max_Drawdown: -46.2%
Results are not great, but the 1970's was a dismal decade.
If by "filters" you mean fundamental inputs the answer is no. Many adhere to the long-value / short-growth trade because it has worked over the past 15 years. But it's just an observation. The next 15 years could easily yield the opposite.levijean wrote:Hi Eric,
Have you discovered any good filters for stocks to improve returns?
An all time high entry is the best filter I can think of. It filters out everything that isn't working and forces you to invest in everything that is working, regardless of what fundamental paradigm happens to be in vogue.
Check out the book "What Works On Wall Street" by O'Shaunessey if you haven't already studied it. It is full of test results of strategies based on various fundamental and technical stock factors that can be used as filters. It is all empirical research from computerized back-testing so the approach should appeal to people on this forum. Highly recommended.
-
- Senior Member
- Posts: 38
- Joined: Wed May 28, 2003 2:17 am
Trend following in stocks often puzzles those who wonder why it doesn't work like it does in futures markets, but maybe it's not such a big mystery. Richard Dennis reduced this to stocks being "more random," but there is a simpler explanation: stocks, especially stocks of a single country, are correlated. They represent their parent asset class too much to provide the type of diversification that trend following requires. Soybeans and T-bonds are separated by more degrees of freedom than IBM and GE.
A market in a futures portfolio can go 20 years without a big trend and then "wake up" to provide a winner that makes up for that entire period. The overall stock index is susceptible to this kind of cycle and so are the individual stocks making up the index. There's no reason to expect a trend following system trading a basket of stocks to perform much better than one that trades a few stock index futures contracts.
The correct approach then is exactly what Blackstar has done: represent a trend following method within the world of stocks and then compare the risk-adjusted returns with the S&P rather than to a traditional futures portfolio. Their study shows the way to be a 100% mechanical trend follower in the stock market (and is a great piece of work).
A market in a futures portfolio can go 20 years without a big trend and then "wake up" to provide a winner that makes up for that entire period. The overall stock index is susceptible to this kind of cycle and so are the individual stocks making up the index. There's no reason to expect a trend following system trading a basket of stocks to perform much better than one that trades a few stock index futures contracts.
The correct approach then is exactly what Blackstar has done: represent a trend following method within the world of stocks and then compare the risk-adjusted returns with the S&P rather than to a traditional futures portfolio. Their study shows the way to be a 100% mechanical trend follower in the stock market (and is a great piece of work).
-
- Roundtable Knight
- Posts: 2071
- Joined: Fri Apr 25, 2003 3:33 pm
- Location: London
- Contact:
This is indeed a most frustrating task which is greatly compounded if you want to broaden your horizon outside the US. https://www.globalfinancialdata.com/ looks to be a good source but at very considerable expense (at least for the private investor). I have not tried the data but have looked at their menu very extensively. In many, many cases while they have total return data for a given index, that data is monthly rather than daily data back beyond a certain point. And if you are going to use monthly data (ugh) then you may just as well download MSCI Index data for free at http://www.mscibarra.com/ – it goes back to 1969. That pre-supposes of course that you are happy to use MSCI Indices rather than, say, the Hang Seng or the S&P.ecritt wrote:Getting accurate .....................index total return data before the early 1990's was challenging.
You could take the possibly rash and certainly inaccurate decision to use yield data calculated from comparing MSCI TR and Price indices for the relevant country and then apply that data to a non-MSCI price index. This does at least give a general indication of dividends/yield for a country index but would of course be hopeless for a sub-index (utilities for instance).
If you are happy to use Standard & Poors World Indices, they have a Global Stock Markets Factbook which apparently supplies dividend or at least yield data going back (they say) to 1975 for many world markets. The data is available also in digital ASCII format I am told. If anyone has bought S&P data I would welcome their comments – so far, I have not.
What is so curious is that anything pre 1980 now seems to qualify as stone age history as far as data vendors are concerned. I remember in Hong Kong in 1986 we had dividend data going back a long way and used to produce charts accordingly with PE bands and suchlike. I have no doubt the data is still available somewhere but perhaps in non-digital form and probably at considerable cost, not to mention effort.
Again, any thoughts or comments appreciated from those who have had the patience, skill and/or luck to obtain such data.
The randomness argument doesn't hold water for long term trend following (in my opinion). See pages 3 and 4 in the following document:Eric Winchell wrote:Trend following in stocks often puzzles those who wonder why it doesn't work like it does in futures markets, but maybe it's not such a big mystery. Richard Dennis reduced this to stocks being "more random," but there is a simpler explanation: stocks, especially stocks of a single country, are correlated. They represent their parent asset class too much to provide the type of diversification that trend following requires. Soybeans and T-bonds are separated by more degrees of freedom than IBM and GE.
http://michaelcovel.com/pdfs/TrendingSt ... Market.pdf
Also, the high degree of correlation (especially in declining markets) can be used to your advantage in the context of long/short. A highly correlated hedge is a hedge you can count on. I've preached against shorting in the past. It's impossible to accurately simulate shorting of individual stocks because of their availability. But simulating short index futures; that's realistic. It took me some time to get over the fact that shorting is pretty much a sure bet to lose money, on a stand alone basis. But it sure changes the portfolio level results for the better. I would advise anyone not to dismiss stocks because of correlation without first considering how high correlation can be beneficial.
"A stock cannot start at $10 and finish at $200 without making new highs along the way. Regardless of the path taken, above average positive lifetime returns (adjusted for dividends) cannot result without a series of new all time highs. Buying that first all time high and staying invested in stocks that continue to appreciate is trend following…..on stocks."ecritt wrote:The randomness argument doesn't hold water for long term trend following (in my opinion). See pages 3 and 4 in the following document:
http://michaelcovel.com/pdfs/TrendingSt ... Market.pdf
Also, the high degree of correlation (especially in declining markets) can be used to your advantage in the context of long/short. A highly correlated hedge is a hedge you can count on. I've preached against shorting in the past. It's impossible to accurately simulate shorting of individual stocks because of their availability. But simulating short index futures; that's realistic. It took me some time to get over the fact that shorting is pretty much a sure bet to lose money, on a stand alone basis. But it sure changes the portfolio level results for the better. I would advise anyone not to dismiss stocks because of correlation without first considering how high correlation can be beneficial.
Eric, have you tried any other method other than buying new all time highs as an entry point to invest in stocks that "continue to appreciate"?
Yes, we've tried everything from 10 day highs to 3 year highs to all time highs, in 5 day increments. Performance is nearly the same from 2 year highs to all time highs. It starts to deteriorate when you move to the left of 1.5 year highs. The deterioration is due to opportunity cost at first. When you buy a stock at a 1 year high the buying power has to come from somewhere…in this case it would be at the expense of the other stocks in the portfolio (stocks at all time highs). As you go faster and faster (6 month highs, 3 month highs, etc.) you get mired in the natural noise of the market, turnover skyrockets, transaction costs pile up, and profits come way down.alp wrote:Eric, have you tried any other method other than buying new all time highs as an entry point to invest in stocks that "continue to appreciate"?
And what if you cannot buy all the stocks making all time new highs at the same time?ecritt wrote:Yes, we've tried everything from 10 day highs to 3 year highs to all time highs, in 5 day increments. Performance is nearly the same from 2 year highs to all time highs. It starts to deteriorate when you move to the left of 1.5 year highs. The deterioration is due to opportunity cost at first. When you buy a stock at a 1 year high the buying power has to come from somewhere…in this case it would be at the expense of the other stocks in the portfolio (stocks at all time highs). As you go faster and faster (6 month highs, 3 month highs, etc.) you get mired in the natural noise of the market, turnover skyrockets, transaction costs pile up, and profits come way down.