Do markets change? Trend Following.

my idea would be that while we have seen but a sliver of possible market outcomes, i would hypothesize that the outcomes that we have seen give us some idea of what the majority of future market outcomes might look like. in developing a model i am not only interested in how profitable it was in the past, but how likely that profitability will continue in the future. does the market need to do very specific things or very broad things for me to make money in the future? there are also certain facts (like the effects of costs on trading) that are not dependent on empirical evidence (that i would call an assertion or theory). by aligning the most facts, using the fewest variables, perhaps i can maximize my chances of future profitability. i would argue that any model that "stopped working" never really "worked" in the first place (arb situations aside. they can certainly become unavailable due to competition). by observing all the available data (beginning in roughly 1972) and using the same parameters (under 5) on 30 different markets across every possible "sector", one can observe over 1000 market years. this is very different than creating a model in the early 80's based on 10 futures markets or developing an equity model based on the NASDAQ from 1985-1999.

to mr. ganngalt:

i believe our profits are based on the non-normal distribution of price moves which i think comes from our species not fully recognizing magnitude vs. frequency in its dopamine reward system. this leads to our being risk seeking with losses and risk adverse with gains. this i don't think will change. how this will manifest itself in market prices, i think is boundless. if the price move distribution is normal, we will lose costs. if the tails are thin, we will lose more. certain methods of capturing this anomaly will prove to be inefficient. i must always keep in mind that i might be wrong and therefore accept the financial damage if i am.

 

If you use few enough if/then statements in your testing, show enough consistency between the creation data and the testing data, show little change as your variables have small changes, and you have shown the discipline to follow rules, then i think you have a usable model.

It would be difficult to have a method that loses >50% of its trades and has an average winner < its average loser, yet still shows a profit. Of course a model may have some of both, but most either win via a high W% or a large avW/avL ratio.

 

Solid stuff. I've got nothing to comment but say, thanks for clarifying.

Hrrrmmm... What you mentioned reminds me of someone... Do you run a CTA in Chicago (specifically in BOT)? The person I have in mind has a 'S' in their initials and in his AM's name.

 

I am happy to think that such a durable model may be possible. It is certainly what I have tried to design for myself and, using futures market data going back to 1970, I have made sure that my models are profitable in every time period during that time frame. In 2050 somebody trading such a system will be able to look back and verify whether or not my design (and my prediction of continued profitability) has been a success. I hope my prediction is correct and that my models remain profitable. Who can tell at this point in time? In any event, I shall continue research and development and ensure that if and when changes to the models become necessary, I make them.

I was not seeking to argue with your mathematics – just the polarity of the statement “ market making versus the rest”. As you now say, a model may have some of both. You can design a long term trend following system with a win/loss ratio greater than 50% where the average winner is larger than the average loser. There are profitable trading methods other than market making where a high win loss ratio and a low average win to average loss ratio are possible. I am probably just being pedantic and had misconstrued what you meant.

I would not seek to argue with your basic premise – as I stated above, I hope you are right and that an analysis of sufficient past data may give us a sporting chance of designing a simple system with few parameters which lasts well into the future. But markets of course go back far beyond 1972 and JW Henry stated at one time that they analysed price movements on commodities going back to the 19th Century and beyond. Admittedly beyond a certain date futures markets did not exist and daily prices, let alone daily OHLC, are not obtainable.

But my point is that there have been plenty of people analysing plenty of data well before we started in business – even if they did not have the computing power readily available today. Perhaps the likes of JW Henry, Dunn and Eckhart messed up their analysis; I would not know. What I do know is that some of the systems they designed seem to have perished (at least in their original form) – notably the Original Turtle Rules. JWH and Dunn looked as if they were going out of business recently and were forced to make alterations to their models – despite Dunn’s reputed contention that he is trading the same way as when he started out in business. You may have a point – perhaps (as per your definition) their systems never worked in the first place. But I for one have difficulty in believing that a system can be designed to be eternally profitable. I very much hope I am wrong.

 

Just a quick question to all here, are you guys all system traders, or is there a discretionary trader among you? I only ask because it seems that you all are system traders.

The great thing about the markets is that there is a tremendous amount of diversity in the manner that people trade. Due to this diversity, there will always be money available to be made in the markets.

Do i trend trade? Never. I attempt to pick bottoms and tops, what almost every book i've ever read tells you NOT to do. Of course i don't get in the way of a run away bull market and attempt to short, that's not what i mean.

Do i think people can make money trend trading? I think the right person can make money trading in ANY manner of trading. It depends on the individual and what rational they are using. Of course personally i would think that trend trading is the worst way to trade. The risk is far too high and markets move in ways that stop out these big risks at just the wrong time.

The markets do change all the time. This is a fact. It cannot be denied. They change in the way that will hurt the majority of traders. At least that's my opinion. They do this because it is a logical certainty that they must. What the majority know/do must fail, because everyone knows what they are doing and can side step the situation before it happens.

 

i might suggest that the original turtle rules were created in the early 80's and may not have had the necessary body of data.

many turtles and mr. eckhardt continue to have very profitable trading careers on a very large scale. that would suggest that while the tactics may have evolved, the strategies and philosophies probably have endured.

trendfollowing has been remarkable in its persistence. it has been well known since the mid-80's, has been written off for dead many times, yet many of the major CTA's still use it as a base strategy:
Man ADH
Winton
Aspect
Transtrend
Campbell
Milburn
Chesapeake
Abraham
Eckhardt etc.
that's around 50B under management. think about how much $$ they need to make and have made after fees, slippage, commission etc. it is a truly staggering number. this group made 20+ last year. that's 25 before fees +2 management, +5 comm/slippage. only managing say 30B that is 8-9B taken out of the zero sum futures markets. someone must lose that money.

while there certainly is a wild survivorship bias, there is another bias that i believe makes these models appear as though they deteriorate more than they do over time. the survivors almost by definition perform to the right of their expectancy in their first five years or else they would not attract any money. in a field of 500-700 at least 50 groups are performing 2 std dev. to the good. someone performing at expectancy stands no chance of gathering assets vs. those. those that gather assets perform at expectancy which includes some years 1-2 std dev BELOW expectancy. this gives the appearance of a disintegrating model when in reality it is well within expected bounds. given the longer term nature of trend following, the relative infrequency of trades, and the relative rarity of the large winners that produce the profits, both the good an bad performance can last a long time, 3-5 years. this could mask a good program, a bad program or make a solid program look tremendous and then disintegrating.

 

Your design approach and your analysis approach sound typical.

Your comparisons to others and to specific periods establish the scope and bounds of what is possible for you.

I would assume that when you go about interrelating with other professionals in your environment, most things and considerations fit into the range of approaches and the range of analysis that is done. The operating products probably also fit into the milieu of how surviving businesses perform. A lot of businesses come and go as you point out.

Naturally, you are fairly certain as to why various operations arrive and survive while others that are established finally drop out of the picture. There are many chronologies of the given mixes at times and many narratives of the lives individual enterprises.

Scales that establish spectrums of the character of most of the facets of approaches, analysis, models, mathematical tooling, sizing, and performance abound. From this there emerges a sweet spot which has moved around over time in your estimate of the history of success in trading.

I like reading this thread and its contributions, since it shows, in a defined way, the biases of the contributors. I associate these biases with limitations that prevent people from looking into other parts of the many spectrums that measure the relationship of traders and the markets.

The "small slice" comment was founded on some limitation, apparently. I can see that a lot of potential efforts for enlarging the slice are not affordable. This industry does not follow the model for most industries where growth has a relationship to knowledge. Not being able to afford the premium for knowledge, especially in a financial construct, is worth examining.

What people can't, won't and don't do is may be where the answer to your current question lies.

My time offset from yours is not more than a decade or two and I do not see the need for such a long interval to be able to draw conclusions. As an architect and as an adjuct professor in that field, I observed two primary streams of creation: A. developing one theme through iterative refinement and B. looking at many themes, and letting them compete until one is processed to completion as the best fit.

B is the modus of the contributors here and the caveat is the character of the market over time. The 80 year period you espouse is a determinant for finding out if some selected approach has durability.

Buffett's theme as we all know has the word durability in it. He is able to chose a different horizon than you and deploy his theme through rotation of capital as conditions permit opportunities and other opportunities fade.

You use the bottom line as the survival measure. Our culture does otherwise; it uses Dr Zeuss's measure, "biggering". We have, at long last, arrived at paying the social price of "too big to fail" after a long period of preventing cartels and the dominance of single enterprises (AT&T)

So I have just surronded this conversation with more items to include in the consideration of trends.

Finally, it is a good idea to dedicate a little time to looking into constructs and punching up the next refinements that will shift them along the spectrum. This is a rational exercise that can lead to rethinking basic constructs.

For example, in the dairy and cattle industry, the focus is production which has two dimensions: peaking and holding to the peak for milk and rate of weight gain for beef. Antibiotics (slow release) was the rage at one point since it shifted the performance a percent or two for each. For 50 cents a head per month, it was determined for over 14 million head, that at least seven times the anitbiotic's shift in peformance achievement and dropping the antibiotics did result on a rotating annual 12 month performance measure. The industry shifted its approach as a consequence.

The contributors have several common problems. There are no pooled efforts to address these problems. The solutions to these problems are used elsewhere. For the dairy and cattle the solution was simply stress reduction. Lo's research in the big sector of trading discovered and documented what were some major causes of degradation in trading performance. None of it had to do with the market's operation and the market's evolution over time into another kind of market. The general performance examined was not related to the connection between taking the market's evident offer and trending in markets.

As I mentioned previously, I found that using induction (right at the beginning of over 50 years of trading) was inappropriate. This statement is not a casual one (many many casual statements, bordering on beliefs have been made here). The consequences were that I do not operate in the sweet spot the thread defines nor do I have to deal with the down side issues the thread has defined.

 

Perhaps not - but the 1980s were not the dark ages. JWH made great play a few years ago about tracking futures markets and analysing them right back to their inception a couple of hundred of years ago.

The Chicago Board of Trade was formed in 1848 and standardised futures contracts were introduced in 1865. The predecessor to the CME was formed in 1874.

I do know from painstaking research that digitised data on even stock markets is hard to come by for dates prior to 1980. I also know however that daily data is available further back, albeit in non-digitised format.

I suspect that with time and effort (and perhaps money) a considerable body of data could be put together for both stock indices and the futures markets.

Perhaps JWH’s claim to have analysed data that far back was mere advertisers’ puff, BS. But they certainly did make such a claim!

 

Thank you for your post Jack. I am happy with both my design approach and my analysis. I do indeed find it useful to look at the very considerable achievements of the better known Trend Followers over many years and see nothing wrong in so doing: Dunn, JWH, Abraham, Druz - they have all done a magnificent job. I am not criticising them - far from it. I am impressed with their flexibility and their adaptability.

I am happy to say that I have lived entirely from trading the markets since the early 90s. Probably rather longer than many other people on this forum. I have had to change my spots quite a few times over this period – I have changed what I trade and how I trade quite a number of times. But I have survived and prospered – in some degree of style.

That is quite good enough for my purposes.

I am quite happy however to listen to and to enjoy other people’s points of view. They may well be right and I may well be downright wrong on any number of occasions.

 

Thanks for your response.

Throughout my trading I have used just one approach. As new markets were invented, it was fun to apply the approach to the new market.

The advent of the PC was particularly interesting. When data switched to real time, electronically, after the ticker tape period (which was more like a telegraph) it was possible to spend more than 15 minutes a day doing trading.

At some point, printers switched to color and graphs were possible. Before that (since 1957) I just made charts in pencil on brownlines to do trend analysis. "Trend following" would have been a very humorous terminology at that time; it still is actually. We shared doing charts and blueprinted them as "B" size drawings. The brokers were totally mystified by us pulling a year's profits in a week or so.

I have always just been an amateur.