Page 1 of 1

Transtrend way of measuring market "trendiness": T

Posted: Mon Feb 28, 2011 10:36 am
by Jez Liberty
Transtrend have come up with their method to calculate the "potential profitability of a trend following system", by measuring the market "trendiness"

They called it Trendpot, which is explained in this presentation.:
Trendpot is a backward looking measure that indicates for an individual market whether a medium to long-term trend following system could potentially have been profitable when applied on that particular market during the preceding month.

Trendpot is a number between 0 and 1:

* 0 indicates a very low chance for a trend following system to have been profitable.
* 1 indicates a very high chance for a trend following system to have been profitable.

I found the concept pretty interesting, if only to use as a monitoring tool (ie check your system performance against a "trendiness benchmark measure": if the system performs badly when the "trendiness benchmark" scores highly, your system might be "broken")

In 2008, for example, the trendiness measured via Transtrend Trendpot calculation was the highest of the past 20 years, explaining the great performance of most Trend Followers. In 2009 though, the Trendpot value was at the lowest of the last 20 years. Not surprisingly, it was not the best year for Trend Followers: it is not that Trend Following did not work, it is just that the markets did not offer many good trends that year.

Transtrend insist that "Trendpot is a measure, not an indicator: No predictive value!", but I wonder (and plan to test) whether this is true, ie could future Trendpot values be a function of past values (in which case it could be used to help decide which markets/timeframes, etc. to trade in the future).


Edit/Update: I uploaded the blox I wrote to implement the Trendpot calculation in the Marketplace on this thread

Posted: Thu Mar 03, 2011 7:03 pm
by M20J
It's been a while since I looked at it, but Automated Trader Magazine had a sister website called Alphability which produced a range of similar statistics.

http://www.automatedtrader.net/articles/alphability/

From what I remember, you can see quite a lot of the website just by registering. I considered using some of their ideas to build a regime-switching strategy, but it's still some way down my research-to-do list.

Posted: Sun Feb 05, 2012 4:30 am
by Macro
Seems interesting, however, why would they give away such a 'measure' for free?

Posted: Sun Feb 05, 2012 6:05 am
by Chelonia
Thanks Jez, where do I read the TP value after testrun?

Posted: Sun Feb 05, 2012 11:39 pm
by Jez Liberty
It is implemented as an indicator so you can use it in blox code (it's computed for the last x days) or also outputs its value to the Print Output file

Posted: Wed Feb 08, 2012 2:32 pm
by rabidric
Macro wrote:Seems interesting, however, why would they give away such a 'measure' for free?
Because it doesn't actually tell you anything really?

Posted: Wed Feb 08, 2012 4:08 pm
by Roger Rines
rabidric wrote:Because it doesn't actually tell you anything really?
Maybe it tells us more than is obvious at first glance, if we can assume that Ideas and an active imagination are the seeds to new methods.

When Jez published his module of the TrendPot idea, I found it interesting because the measurement of Trendiness is something that is hard to quantify ahead of when it is useful. However, it could be a useful idea to measure a market's trending characteristics, and thus current potential when considering a symbol's selection in a portfolio, or as a method for adjusting the ranking of a symbol in relationship to other markets. It might even be a good process to use as a consideration for adding size, or accepting more position risk.

At the heart of all this is the effort needed to test these ideas so that we can discover whether they would would fall within the boundaries our beliefs need for success.

In the state that Jez published his module, it is capable of producing the indicator profiles shown. His module uses a single line 1-color indicator. On my planet I like to use colors to tell what might not be obvious. In this case I was after the directional information inherent in Jez's module, and two colors show less clutter for this kind of presentation when viewed as a histogram. In the attached chart image, orange shows the potential is favoring a downward, and green is favoring an upward trend direction.

Other than changing the display, the output shown is all from what Jez packaged in his module.

With the output image shown, does seeing it in a different form give you any new ideas?

Thanks Jez. Nice work.

Posted: Thu Feb 09, 2012 4:12 am
by rabidric
looks like how most other oscillators behave- gives yiou a strong reading after the fact, which invites techniques like "after a really strong spike the mkt is overbought and you should be wary of new longs, and instead be looking to cover".

but a macd or stochastics/rsi would produce something very similar, so may as well stick with them if identifying overbought conditions is your cup of tea. :roll:
I bet a technician could find a few "reverse divergences" in that picture too if he wanted :lol:

Posted: Wed Mar 28, 2012 5:55 pm
by trackstar
I went ahead and programmed this as well. After I analyzed it for a bit I think there are some modifications that could be made to do more of what Roger described in his post.

"When Jez published his module of the TrendPot idea, I found it interesting because the measurement of Trendiness is something that is hard to quantify ahead of when it is useful. However, it could be a useful idea to measure a market's trending characteristics, and thus current potential when considering a symbol's selection in a portfolio, or as a method for adjusting the ranking of a symbol in relationship to other markets. It might even be a good process to use as a consideration for adding size, or accepting more position risk."

Knocks the socks off ADX at least.....