Page 1 of 1

Let's List the smoothest trending markets

Posted: Mon Apr 28, 2008 5:12 pm
by TrendsCatcher
A wise man said "From a trend-following perspective, the best trends are not those that move the most, but those that move the smoothest. It is the choppiness of the trends, and the number of false movements that determine the profitability of a market."

I Totally agree! A smooth trend has high Signal to Noise ratio, affords tighter stops without compromising accuracy. Therefore, by trading a smooth trend, you are more likely to achive higher R-multiple trades. The other important thing is: does a market trends frequently (such as soybeans)? If it does, it derserves more attention and a more thorough study.

So, may I suggest traders here list the markets that in your opinion, trend most frequently and when they do trend, trend smoothest? Yes, all markets trend well at one time or another, but some just trend more frequently, and when they do trend, they display smoother, more tradable trends.

I'll give it a shot first:

Most trending market:
Grains, Oilseeds (soybeans trends a lot every few years), Coffee

Somewhat trending markets:
Metals, Energy, meats, most softs, FX, interest rates.

Hardly trending markets:
Equity indexes

Your opinion?

Posted: Mon Apr 28, 2008 6:05 pm
by sluggo
The beginnings of a list appears on pp. 217-218 of Way Of The Turtle, for those who may be interested.

Posted: Mon Apr 28, 2008 6:12 pm
by TrendsCatcher
Yes, that certainly is a good start. But in that book, he didn't discuss which market would trend more frequently. Also, other authors (eg, in Art Colins' "Market Beaters") mentioned that the financials trend the best (which is not my impression, which was also not the wise man's impression based on his book). Also, Jake Bernstein mentioned that some European markets display "absolutely beautiful trends". So I still would like to know the opinions of the traders here, and European fellow traders are welcome to share your experience trading the various European markets, which I have absolutely zero experince but would love to trade if the markets are highly tradable, and if standing stop orders are allowed


Posted: Mon Apr 28, 2008 7:51 pm
by Mark Johnson
Suppose you had historical data for ALL futures markets, all around the world, for the last 40 years. Suppose you had a Blox license and a huge fast computer to run it on. Suppose you had a huge team of Blox programmers working for you.

What tests would you perform, what experiments would you run, to find out which markets have trended the most smoothly in the past? (Sadly you can't use historical data to find out which markets will trend the most smoothly in the future because, oh shoot, what's that expression ?, past performance may not be indicative of future results.)

Would you use all 40 years of past history? Or only a more recent slice?

Would you look at indicators? If so, which indicators? What settings of parameters for those indicators? (for example, if you're looking at ADX, you need to choose a parameter "n" so you can calculate ADX(n), the ADX over n days)

Rather than looking at indicators, would you instead look at backtest results of mechanical trading systems? If so, which systems? What settings of parameters for those systems? (for example, if you're looking at a Dual Donchian Channel Breakout, you need to choose parameters "j" and "k" (with k < j), such that entries occur on j-bar breakouts and exits occur on k-bar breakouts the opposite way).

Would you look at both indicators and backtest results of mechanical systems? If so, how would you weight each of them? Would you look at several indicators and backtest results of several systems? If so, how would you weight the various indicators and the multiple systems?

Would you make a distinction between tolerable non-smoothness (i.e. non smoothness when making profits) versus intolerable non-smoothness (i.e. non smoothness when losing money)? Or would you say that all non-smoothness is intolerable, even when making money? Is a large gap up (a non-smoothness) in the middle of an uptrend, just as intolerabe as a large gap down in the middle of an uptrend? They are equally non-smooth but not equally profitable.

It seems to me you would have to make a lot of psychological decisions before you were able to tell your team of programmers, what to code into Blox and test.

Posted: Mon Apr 28, 2008 9:25 pm
by TrendsCatcher
Mark, I liked your reply, very profound insight! I particularly like the concepts of "tolerable non-smoothness" and "intolerable non-smoothness". I never heard of those expressions, but I immediately know what you mean. Yes, once I am in a trade, I would logically "hope" that the trend gets more volatile, as long as it doesn't breach my entry point - tolerable non-smoothness!

I need to ponder on your ideas more.

Would you elaborate on these concepts more? Eg, one would naturally assume that tolerable non-smoothness is highly correlated with non-tolerable non-smoothness, which is highly correcteld with non-smoothness, tolerable or non-tolerable? In other words, as long as the trend is jumpy (of course, up and down, not just a serious ups, in that case, it's not non-smooth), it is more or less non-tolerable from a trend-following point of view? Psychologcially, I do like smooth trend because when I get in, it is very easy to set a tight stop loss, and when I get out, it is also clear when I am going to get out (when the trend start to get un-smooth in the opposite direction - failure to be normal, failure to be smooth. Whereas in a choppy trend, you cannot setup a tight initial stop because you don't wait the noise to whipsaw you out, on the exit, you have setup a wide trailing stop because you don't wait the "normal" big nosie to kick you out before the trend is over too. I never backtested this (and I admit I don't know exactly how to do a legimate backtest), but just by looking at the CRB year book charts, some markets, such as Coffee and Soybeans, do tend to spike up and down more often than other markets, and when you look the weelky charts, they do seem to trend better, with shallower, shorter corrections, and long long up bars. Those seem to be the best money-making opportunties. What do you think?

Posted: Tue Apr 29, 2008 5:54 am
by nodoodahs
Anyone care to QUANTIFY a measure of "smoothness" that could be programmatically searched for?


Posted: Tue Apr 29, 2008 7:46 am
by sluggo
nodoodahs wrote:Anyone care to QUANTIFY a measure of "smoothness" that could be programmatically searched for?
The most obvious candidate from textbooks is Linear Regression R-squared.


I would amplify upon the thought mentioned earlier: before you launch a huge computer effort to go calculate "X", you might wish to pause and ruminate upon the question of what "X" you truly seek.

The question "what is it that we really desire in trading system behavior" has been extensively discussed, argued over, researched, calculated, and hotly disputed. It is the topic of an entire chapter in WOTT, for example (ch.7). Personally I find the discussion in Jack Schwager's book MTMT (link) to be among the best available. He summarizes the reasons why some people think it is important to distinguish between welcome non-smoothness and unwelcome non-smoothness, and thus why some people use "semideviation" (also called "downside deviation") as a proxy for undesirability.

Once you have formed your own opinion about what it is that you truly desire in trading system behavior, your search can commence. You can go looking for characteristics of futures markets that have given your desired behavior in the past. It may be smoothness of trendiness, it may not. Just because some bozo named Mortimer or Randolph or Winthorpe says he desires "X", or worse, just because he grandly proclaims that all people in all places ought to desire "X", doesn't mean it's true for you. Think for yourself.

Posted: Tue Apr 29, 2008 8:02 am
by nodoodahs
Ah, trust me, Sluggo, I do think for myself. :lol:

My point was philosophical, as was yours. You point out that we should question the desirability of X before looking for it with programs. I point out that we should DEFINE X before talking too much about it. There has been a lot of talk about smoothness' implications but little talk about what "smoothness" IS.

Moving on to some definitional discussions ...

I wouldn't necessarily say that R-Square was the most obvious candidate overall, although perusal of statistics textbooks may lead one in that direction.

Personally, I'd try to experiment with ratio calculations, where the numerator is some measurement of average price change from endpoint to endpoint of time, and the denominator is some measurement of between-endpoint volatility of price. For lack of knowledge of any predefined terminology for this concept, I call it "efficiency of movement." TradeStation had something similar which they called "Choppy Market Index," but it is very simplistic in calculation and that creates possible errors in application.

Posted: Tue Apr 29, 2008 12:43 pm
by sluggo
nodoodahs wrote:I'd try to experiment with ratio calculations, where the numerator is some measurement of average price change from endpoint to endpoint of time, and the denominator is some measurement of between-endpoint volatility of price.
Kaufman's Efficiency Ratio uses this very idea. Although, as you said yourself (link), unfortunately the most "efficient" way for a stock to move is to teleport (gap).

Posted: Tue Apr 29, 2008 3:01 pm
by nodoodahs
Ah, fond memories!

I added a gap filter to my efficiently-moving stock scans and got some meaningful information out of that combination, though. I used a ratio of average range (which is intraday) and average true range (which incorporates gaps) to throw some of the nasty teleporters out of the scan.

The thing with futures (and index ETFs) is that they don't have extreme gaps like stocks can, so looking for efficienct movers in those classes won't wind up with finding a bunch of acquired companies and biotechs ...

Good find on Kaufman's Efficiency Ratio. Now I have a name and reference when discussing the concept. :D

Posted: Tue Apr 29, 2008 3:13 pm
by RedRock
sluggo wrote: Mortimer or Randolph or Winthorpe
I love when he stuffs the fish into his santa suit at the xmas party. Ha

Posted: Tue Apr 29, 2008 4:00 pm
by Algonquin
Here's a thought: does smoothness of trends equate to smoothness of equity curves? If, because of your market selection rules, you are only trading five "smooth" instruments, I would think not. In fact, I would go so far as to say that five highly noisy instruments would provide a better risk adjusted return.

The essential goal is smoothness of the equity curve. In my experience, the best manner for achieving that end is the use of broad diversification across multiple facets of the trading system. Just my $.02.

Posted: Tue Apr 29, 2008 6:16 pm
by nodoodahs
And another thought: does the smoothness of a trend say anything about its statistical odds of continuing?

Posted: Wed Apr 30, 2008 8:39 am
by Mark Johnson
I called my old pal Vardebedian and asked him to make me a list of global futures markets, sorted by how well they work with trend oriented mechanical trading systems. He sent back an Excel spreadsheet (attached) with 128 markets. For some reason he didn't include any markets listed on exchanges in Dubai, or India, or Mexico, or China on his list. Sorry about that.

He generated the list by running Blox a few times and making some type of aggregate "Score" across the different runs. I didn't ask for specifics of his method, except to verify that it was 100% computerized, with no human discretion used to "juggle" the scores or the rankings.

It's fairly interesting that big brother, little brother pairs (like fullsize S&P and eMini S&P) don't wind up with exactly identical scores. In some cases, they are surprisingly far apart. I'll have to ask Vard about that one day.

Posted: Wed Apr 30, 2008 5:11 pm
by TrendsCatcher

Very interesting results. Thanks for the effort in doing this.

I am suprised that the VIX futures are ranked high (No. 13). My impression is that the VIX doesn't really trend, it is arguably the few true-mean reverting instruments.

Any thoughts?

Posted: Wed Apr 30, 2008 6:47 pm
by sluggo
Why rely on "impression", when the actual historical data of the actual VIX futures contracts is available for free download on the CFE website?

A downtrend plus some serious amounts of contango in the VIX futures, made for very pleasant trends to the short side. Too bad nobody was aboard!