Diversified Trend Trading Approach

man, that's a hard task... I mean "listening to everyone".

I like simple life so I pay attention only to what gets my attention

I enjoy quietly reading your posts but.....I really have to disagree with those lines.
Those are the definition of curve fitting and curve fitting will sooner or later kill you in this business.You don't want to curve fit your strategy to the recent market conditions but what you want is a robust system that can endure changing market conditions(because markets stay the same as they are always changing) and will work for as long as free markets exist...with no curve fitting.

You also want systems that work relatively similarly over any market it is traded on, not just ones that perform best on certain markets or market segments.
The fact that it works best across some obscure crosses makes me feel like there is some element of curve fitting to your system.
Just my $0.02.

Yep...convergent is not my cuppa tea...and I understand what you are talking about...but I am finding that you can supplement divergence with convergence provided that divergence is the dominant part of the portfolio. You don't have to throw the baby out with the bathwater. :-)

For example, have a look at the chart below.



This represents a portfolio suite of of 10 EA's tested across GBPJPY that comprises core divergent (trend following) design logic for each EA (cuts losses short and let's profits run) but also includes an entry condition that is data mined and convergent in nature (within the degrees allowed by divergence) and responds to more recent market conditions.

The lower equity curve represents the return stream of a portfolio of EA's tested from 1/1/2000 to 31/12/2014 with the OOS period defined in red to current day. Note the trajectory of the red line and it's slope.

The upper equity curve represents the return stream of a portfolio of EA's tested from 1/1/2000 to 31/12/2015 with the OOS period defined in green to current day. Note the trajectory of the green line and it's slope.

Note how both curves have a very similar overall volatility profile. That is the divergent design represented in each of the underlying EA's displaying itself. When markets trend, the equity curve lifts. When markets don't trend, the equity curve stagnates.

Now closely compare the trajectory between the green and red lines across the same date range interval of OOS data and you will see how convergence (impact of 12 months of additional In Sample Data IIS ) is supplementing the overall divergent return distribution. This is what is causing the 2 equity curves to separate. Convergence is supplementing the return stream while divergence is still the dominant contributor.

The impact is small but is material in nature and can significantly add to total returns over the long term. The bottom line is that you can use convergence to supplement but not inundate the portfolio. Provided you re-data mine at say 6 monthly intervals you keep with recent market conditions and assist in those times when markets are stagnating.

PS......Nivad, of the 12 markets listed in your quote from me, they represent half of my research portfolio. Only the better performers over 20 years or so. The remaining markets were not as pretty...but I am agnostic in the ones I select to trade. But my selection of universe to trade is based on the available data I have to work with. I must have at least 20 years or so of D1 data set to work with. If not then they are not included in the universe. This is not a decision based on curve fit performance but a decision based on enough data to generate a sufficient sample size to substantiate divergence.

My bad. All I know is trend following. I know nothing of reversion to the mean systems. Apologies.

Shit N.....No need for apologies. Thanks for the chat :-)

I like to use the word "optimization" in regards to my approach... as my systems work with several parameters, on several insruments etc... but don't worry, I don't mind if you call it curve fitting lol

Hi H, it's not curve fitting if you use the same universal calibration for all pairs and/or on all timeframes.

working strong! Sunny times, different places

A good read :-)

AQR’s Problem With Machine Learning: Cats Morph Into Dogs

“If a researcher identifies a new signal that captures a particular form of asset mispricing useful for predicting prices, then as the signal becomes more widely known, more traders act on it, correcting prices more quickly,” wrote the authors of the report. “The market eventually absorbs the information, and the data generating process changes due to the very actions of agents in the market.”

Alpha is scarce and has remained so in the face of advancements over the past few decades.
AQR acknowledges this.
“Using new data and machine learning to build alpha (i.e., to find new unique sources of return predictability) heads straight into the most competitive aspect of financial markets,” according to the paper. “As more investors enter the market with similar tools, the mispricing corrects and that alpha compresses to zero.”

It only gets more difficult with wider participation.

Nick Radge nails it again. Great podcast.

The Blueprint: Create a simple trend following system w/ Nick Radge

Luv'd the inclusion of a market regime filter to improve drawdowns and lock in profits during the trail.

While the discussion centred on trend trading stocks long only with no margin.....so much is applicable to forex.

Enjoy

C

Roll up....roll up.....get it while it's hot. Haven't read it yet but have pre-ordered on kindle. If it is like his past 2 books it will be a cracker :-)

Trading Evolved: Anyone can Build Killer Trading Strategies in Python

Attached Image

I am waiting for the hard copy available from end of august.

The Yin Yang of Trend Following – Flipping Statistics on their Head in the Garden of Good and Evil

Attached Image (click to enlarge)

Readers of this thread would be quite aware now of the underlying philosophy of the systematic divergent trader.

1. They would be aware of the concern placed by divergent traders in using statistical techniques as a measure of predictive certainty;
2. They would be aware of many possible future paths called risk and not attracted to the myth of a single optimised future path called guaranteed profitability;
3. They would be aware that divergent traders seek abnormality and non linearity in market conditions for their survival as opposed to predictable normal market conditions;
4. They would be aware that we fully engage in systematic application as this technique predates on the rational mind and we need a technique that prevents second guessing ourselves and falling into the same trap;
5. They would be aware that trading the ‘normal daily churn’ of the market is a divergent traders worst nightmare resulting in progressive building drawdowns while market conditions remain ‘convergent’ and predictable in nature;
6. They would be aware that alpha in this divergent space relies on those unpredictable rare moments when markets become less efficient and crowd behaviour becomes more coordinated in nature;
7. They would be aware that diversification is not simply used to capture non-correlated return streams but more importantly used as a measure to shop far and wide across the liquid universe of the financial markets for trending conditions that ‘scream out at you’ and are more likely to be enduring events as opposed to random ephemeral chart patterns. As a result, you are less likely to ‘pull the trigger finger’ on everything that moves and hence restrict your trading activities to the meat of the larger trending condition;
8. They would be aware that we don’t pick tops and bottoms but rather focus our attention on the ‘middle meat’ of trends;
9. They would be aware that all our efforts are directed towards risk management where profitable outcomes are just a symptom of the process;
10. They would be aware of the ups and downs of a civilisation and have a two way approach of either long or short as they understand that while history may not repeat, it certainly does rhyme;
11. They would be aware that we focus on the Pareto principle and the outlier as a way to survive in this competitive landscape recognising that it is the anomaly where the long term success in wealth building lies, not the demand for immediate reward of regular and predictable cashflow; and
12. They would be aware that our philosophy is deeply couched in the principles of ‘cutting losses short and letting profits run’.

Now despite this awareness of the philosophies and practices that shape the systematic trend following mindset, it is tempting to conclude that we throw away all statistics that are representative of more ‘normal market conditions’ and only use exotic mathematics of chaos theory that are more appropriate for adaptive non-linear market conditions.

Well the problem is that no one understands how to apply these peculiar and complex adaptive models and we shy away from any purported mathematical process that seeks to ‘enshrine’ the future in certainty.

We don’t however throw standard statistics out with the bath water. ‘Au contraire’ my friends. Standard statistical treatment defines a population of market participants that like to play in the safe fictional world of certainty and immediacy…….and we like to eat them…..for they are tasty.

Rather we use standard statistical treatment to define what are regarded as ‘normal’ market conditions or standard Gaussian behaviour and use them to draw a line in the probability sand of participants where we define a safe predictable world as lying within and a hostile uncertain scary world as lying without. The majority of participants lie within but the real ugly weirdos lie outside that line in the sand.

We flip the statistics on their head using the Yin Yang of it all and simply use it to define that region where traders like to think they have a grip on a complex system and that region where ‘most traders’ fear to tread.

That region we inhabit is where the predictive models break down and any supposed grip of the notion of ‘certainty’ is lost. In that hostile environment the ‘certain trader’ is stripped of their life support systems where they become unstable, emotional and gripped in the primal nature of fear or greed. They just can’t think straight and while ‘predictably’ stunned in the headlights or running into brick walls, we simply cackle ‘You are on our turf now stranger, looks like we are going to be fed today folks’.

You have got to be real ugly and sinister to survive in this hostile world of uncertainty. No beautiful and elegant designs for pure speed and precision….but rather thick impenetrable skins, warts and tentacles and pointy bits…..and you only come out to eat at night.

But hey…..we have feelings too. Do you know what it feels like when all about you are feasting on certainty. I mean really know? We just sit on the sidelines with maybe a brother or two being lost to the ‘predictable cause’ here and there when it all get’s too much. Our brethren are ripped apart by you guys but fortunately the real stalwarts, the arcane magicians, the ogres and the orcs keep most of us on the one path.

I know, you white knights want us all gone and vanquished but that ain’t gonna happen anytime soon. Provided that this world of uncertainty remains hostile to you predictable folk we will always be out there just waiting……Wherever you go, whatever you do, we will be right here, waiting for you!

Sleep well and prosper. Nightie night.

C

Bwahahahahahahahahahahahahaha

Sometimes you just got to revel in the beauty of music. Be patient with it. Discordance sublimely morphs into beauty. So much like trading. You fight with it first in a discordant way....and over time as your experience builds you adapt to it's rhythm resulting in majesty :-)

C,

thanks a lot for recommending this book! Been looking for something like this quite a while!

Cheers
S

Cheers S. I have got through some of it and as usual....A great read packed with gems. I like the way he throws you in and avoids a long winded background to Python. Probably the best way to attack it for those who fear to tread into programming......like me :-)