Diversified Trend Trading Approach

That's great mate. Me too......I am getting so much out of those podcasts. They are a real goldmine of info. Have a great weekend mate.........have tax work to do tomorrow and unfortunately can't avoid it any longer. It's down tools and into the real grind. :-)

Take care C.. don't work to hard..

Hey C.. I see you have approx 12.5% of exposure and it looks like most of it is weighted towards the risk off side..

whats your thoughts on...say, if you were weighted the same amount of risk towards the risk on side and there is a black swan over the Weekend ? or would you not open so much exposure to the risk side.... reel it in a bit ?. I know your objective is to spread your exposure over as much as possible to non-correlated assets but the black swans I have witnessed the correlations kinda become one.. so if it happened over the Weekend ( which it will at some point ) whats your plan ?

Morning Ja

I am getting less concerned with managing correlation exposure via instrument (market) diversification as it is becoming a progressively more and more difficult task and as a result I am tending towards employing system diversification as my major method to manage portfolio risk. All instruments both within and across asset sectors are more correlated these days. However to be honest, correlation can either be good or bad for a trend trader. When the proverbial hits the fan as it did in early Jan, gross movements across correlated instruments was a major windfall for trend traders provided you were trading in the right direction. If you were selective in your choice of instruments using correlation as a measure, you would have missed a significant chunk of this opportunity. Trends however are getting fewer and farther in between, so you have to be in a position to make hay while the sun shines, so you need to look at ways to overcome this risk management obstacle.

What I therefore do to manage unfavourable correlation on the discretionary systems is to scale into my trades on a single instrument commencing at 0.25% trade risk and only add new trades when worst case breakeven has been achieved. This way, I am therefore managing realised equity drawdown (as opposed to unrealised equity drawdown) as I am prepared to sacrifice unrealised equity in the hunt for good trending opportunities. By scaling into trades this way, once a trade is >=breakeven you can exclude it from your calcs of portfolio heat which releases more equity for new trade opportunities. If I found that half (eg. 12 of 24) of the discretionary portfolio (being correlated) signaled a trade entry, then I would be in all of them risking a max total of say 3% equity, but in reality this has never occurred as the use of trailing stops that activate immediately on entry always reduces trade risk exposure of the portfolio. The most I have faced so far in testing over the past 3 years with the discretionary portfolio is a max exposure at any one time of approx 1.5%. The Swissy event last year lumped me with a 2% loss for the portfolio but that was with approx 100% slippage.

You may also note that the discretionary systems that undertake both primary trend directional trades and countertrend trades seem to be holding up better than we originally expected last year than those that simply focus on trading in the primary trend direction. The rationale is becoming clearer for this interesting phenomenon. For example, on the MN, W1 and D1 timeframes, I may have active trades in the primary trend direction whereas on he H4, H1 or M30 timeframes we may have countertrend trades running. The benefit of this is that it assists in reducing overall portfolio correlation. Only in times when the market is strongly trending do you find that all timeframes are aligned. To do this, you need to consider each timeframe in isolation and not be biased to only taking trades in the primary trend direction. Yes, we get a netting off in trade profit when market conditions are not straight forward, but this reduces overall portfolio heat and risk-on exposure against prevailing market direction. The system therefore is responding to prevailing market conditions as opposed to trying to force a result against a market that doesn't want to comply.

You will note that because I am using such low trade risk percentages, even if we had a black swan event tomorrow, the impact across the portfolio would not be material to overall performance. The benefit of diversification is that you are actually applying risk management without actually thinking about it. With the Swissy event, what I was thought was a severe negative result of 2% loss to the portfolio paled into insignificance against those traders or funds who were not diversified.

For the automated systems, I use system diversification as a major measure to manage correlation risk. While markets are becoming more correlated, a major weapon to employ in this regard to manage correlated portfolios is via system diversification. While you have limited control in directly managing instrument correlation, you have 100% control via system diversification as you can directly address correlation with system design. Merging uncorrelated systems into a robust blended portfolio is my major weapon to manage portfolio heat. You will note that the CTA's are also engaging this technique as the variability of returns in the trend following camp is progressively increasing attributed to the use by CTA's of additional systems to fight market correlation. Traditional trend following CTA's of 20-30 years ago may have employed 2-3 systems to supplement their core trend following strategy. Today most CTA's are using in excess of 8 or so systems to supplement returns and manage portfolio risk exposure.

I hope this makes sense Ja.

Have a great trading week mate.

C

PS A great case in point is February's result which I will be posting in the next few weeks. In the discretionary portfolio, the stuffing was knocked out of the EDTT1 approach which only trades in the direction of the primary trend. I suspect that we will be seeing some grief from the traditional trend following CTA's for Feb despite Jan's great result. This was because the majority of instruments in my portfolio were actually highly correlated and all undergoing trend reversal following the shaky volatile market conditions experienced in January. The EDTT approach however which includes countertrend trades, fared much better and this is attributed to the fact that we were engaging both trend following and counter-trend strategies in the same system. It was system diversification that saved the day with EDTT as opposed to instrument diversification.

To manage risk these days, we need to adopt new methods such as system diversification as traditional portfolio diversification methods are clearly not as successful as they once were.

C., these are optimized testing results (Somehow curve fitted) not very reliable, this system use 1:1 Leverage with no stops and was always in the Market with no additional filters, just raw SMA's.

Cheers Bass. The performance results had me intrigued to say the least. :-)

It makes complete sense, the more I delve into the Institutional systematic - algo driven side of trading the more I realise that there is so much more to learn past the bullshit written and believed by the majority of retail, the sort of information given by these CTA's is golden and gives the clues for longevity..
Fully automated - Semi automated or discretionary trading, I think this information is next level investing that needs to be absorbed and used ..

I hear ya with using trade models for diversification as asset diversification is limited.. it just makes so much sense..

As Bass said your BT results are Institutional standard and brings the Bench mark right up in this forum..well done to you mate..

I've been meaning to reply on this great thread but haven't had time. I wish I had some deep, new insights on the subject of diversification but the subject has already been covered in depth in this thread and in a very logical manner with lots of examples.

I agree with the conclusions that asset diversification has its limits. As Copernicus said, when it all hits the fan, the correlations go out the window (sometimes this is good and sometimes no).

At least when I diversify between systems I have some semblance of control over the systems' correlations. But correlation in general is a moving target. So in the end it appears that the greatest ally is the law of large numbers combined with systematic and statistically sound testing procedures and using LOTS of data. Automatically generating uncorrelated systems is the next logical step when one hits the upper limits on individual effort and time.

Last, it has helped me a great deal to test all assumptions and make copious notes and take and record lots of observations when doing head to head comparison tests aka simplified hypothesis testing. It's useful to know if tweaking the system building process in a certain way is a net plus, minus or neutral. This takes time to test properly and is manual labor, but is worth it to document so that you can see if what you are tweaking actually improves things or not.

Ja and FXEZ

Thanks guys. I really appreciate your comments. I know how hard you guys are at it in your own areas of specialty and your input is really valuable to me. Some of the great pearls of advice have come from you guys probably without even realizing it. It is when I am breaking down some of the things you say in the threads you participate in that give me some great light bulb moments. So I really appreciate the opportunity to bounce ideas off you and collaborate in thoughtful discussion. :-)

C

Something to chew on and consider for testing. The conclusions reached regarding the mean reversionary nature of the market since 2000 agree with my testing but it is not wise to rest on your laurels and assume mean reversion is here to stay despite their current very attractive return profiles. It is easy to fall into the mean reversion trap if you simply look at the short term trading horizon and demonstrates why adaptive strategies are worthy of consideration considering that particular market conditions can have particularly long life-cycles and why little nuggets such as these are very useful to ponder (thanks to Algo TraderJo for bringing this to my attention) :-)

As a general approach to systems development, the selective application of strategies that respond well to recent market conditions is not a bad idea. As with rotational trading strategies in general, if you have say 10 different strategies, each which respond to specific market conditions that span Van Tharpe's market condition matrix, then at say a monthly or quarterly interval, review the performance of each system, rank the risk adjusted returns and simply switch on the top 4 performers and switch off the balance of systems until the next review period. Alternatively you can continue to utilise each strategy in an adaptive blend and adjust weightings in accordance with recent performance.

After we have a few more diverse systems under our belt we will start investigating adaptive system strategies and compare their long term performance against static composite blends.

C

Rotational Trading Systems

Here is an interesting article relating to adaptive trading strategies which demonstrate sound techniques to ensure your trading system can respond to changing market conditions. This thread will shortly be assessing these techniques as a method to further accelerate system portfolio returns and I will therefore be posting articles of relevance as I come across them in the thread for future discussion.

There are two aspects of rotational trading that we need to look at:
1. Rotational selection of instruments (markets). In other words, evaluating the performance of the instruments in our universe over say the last 12 month rolling period and ranking their performance metrics...... then via a weighting ratio or defining a restricted range of the best performers, we bias our future live trades in favour of the top performers for the next month or quarter ....then rinsing and repeating the evaluation;
2. Rotational selection of systems. The same process being applied to evaluate a universe of systems and only trading those systems where recent performance indicates that market conditions are favorable for those systems.

....anyway, food for thought in a later discussion.

Cheers

C

A great interview with Nick Radge back in Jan 2015. Going back to basics can be a breathe of fresh air and a much needed medicine to bring your trading back on track. Let your head nod off it's shoulders as he deftly breaks down the world of trading using quantitative measures and experience as opposed to mere hearsay or myth.

Enjoy.

C

Worthwhile having a look at this strategy and it's application within a diversified portfolio.

Mark Ursell has some trade models developed purely in Excel for those who might wish to take the plunge and look at back-test alternatives to MT4. The good thing about modelling in Excel, if you are not familiar with a programming language, is that it forces you to understand the detailed mechanics of a trade strategy from the assumptions of the indicators used, the costs of trading and the mechanics of the trade process itself.

aveagreatweekend :-)

https://soundcloud.com/disco-tech-ed...io-guestmix-16

C

Another System to add to the Portfolio Suite

The following system specs are derived from the Bollinger Mean Reversion tests described in prior posts. One of the problems with mean reversion systems (convergence strategies) is that their lifecycle is typically of a short duration in the context of overall market conditions. For example, since 2000, mean reversion systems have enjoyed their season in the sun, however the nagging exposure that is faced with these systems relates to their positive skew, which can at times, lead to excessive portfolio heat. For example, to receive the full benefit of a mean reversion system, you need to give substantial breathing room to allow for mean reversion to take place. This may involve averaging down until the mean reversion occurrs, or alternatively utilising other exit measures as opposed to fixed stops which is what the BC3 and B3L strategies seek to achieve.

While a diversified portfolio can in general protect you from excessive exposure, it is not unheard of for professional fund managers that utilise excessive mean reversion to come undone. For example the rise and fall of Long Term Capital Management is a classic example of where mean reversion can suddenly break down leading to excessive over-exposure.

So....here is an alternative that seeks to utilise principles of mean reversion within trending markets that aims to enter trends at retracements and exit trends at optimal profit targets utilising a simple short term Bollinger Band with an initial stop on entry applied using ATR.

System specs

1. We use a BB with a standard deviation of 1.5 and SMA of 5
2. We define the long entry zone as the region that is 10% of the distance from the lower BB to the upper BB.
3. We define the short entry zone as the region that is 10% of the distance from the upper BB to the lower BB.
4. We use a SMA of 100 periods as a filter to determine if we are long or short.
5. We count the price bars that extend into our entry zones and wait for the 3rd bar of confirmation before entering.
6. We use ATR 14 period and 2 x to define our position sizing according to 0.25% trade risk and also use this setting as our initial stop.
7. We exit when either price reaches the upper/lower band or when the initial stop is triggered.

TrendRetrace3 Unscaled Gross result

Notes:
Across the 40 instruments of the portfolio, 30 were profitable over the test period with only 10 being unprofitable. The very low drawdown allows for significant scalability and the variability of returns are not significant. We have a nice Pwin of 65% which augers well for a trend following/swing trading system and have a significant trade frequency that lends confidence to the robustness of the strategy. The strategy does not exhibit significant variation between long trades and short trades reflecting a symmetrical strategy.


TrendRetrace3 Scaled Net results with Profits Reinvested


Notes:
The scaled results are impressive and despite the negative returns of 2015 and 2016 (YTD), this strategy has clearly outperformed the TF Index. Gross Returns have averaged 18% over the testing period and CAGR after fees has been an impressive 11.08%.

So how does it perform in a blend?

DAll MT/ FUSEX/ TrendRetrace3 Blend
The best of the blends to-date. We are now starting to get there with a blend that is right up there with the top performing CTA funds over the test period.



DAll MT/ FUSEX/ TrendRetrace3/ B3C Blend
Not to be outdone, a further blend that accelerates performance even further. Smokin'.....it must be Saturday Night Fever.





That certainly is worth having a little dance to.




Cheers

C

It don't mean a thing, if you ain't got that swing. It's Saturday night people's.....well Sunday morning actually in Terra Australis :-)

Trend Following Performance Report - February 2016

What a great month February was for the trend following CTA's. This supplements January's strong performance results and gives a great start for the 2016 calendar year. Volatility in the futures space continued on into February. Unfortunately my automated and discretionary systems did not enjoy the same windfalls for the period. The diversified automated systems produced a moderate positive result, however the discretionary systems that are focussed towards the forex and stock index universe got slapped around with significant counter-trend volatility during the period. This is why diversification across asset classes is strongly recommended.

The Trend Following Index (TF Index) which tracks the performance of 47 CTA's which are monitored in this thread produced a 3.85% return for the month which brings the YTD calendar result for 2016 to a very nice 7.22% return. This far exceeds the S&P TR YTD return of 2.25%, so the trend following CTA's are leaving the buy and hold mutual fund fraternity in it's wake at the moment.



....before we get deeper into the results, a message from your sponsor *you just know I had to do it*.



Below are detailed performance results of the CTA's being monitored and the best of breed automated systems from the camp of Copernicus, the majority of which have been explained in detail in this thread. Rather than wax lyrical regarding the performance results in this table, I will let you make your own verdict.



The Monthly Awards for February 2016








My next post will discuss the performance of the discretionary systems (DTT, EDTT1 and EDTT).

Cheers

C

Discretionary System Performance - February 2016

A negative month across the board for the discretionary systems. Unfavourable market conditions across the restricted investment universe saw the 'whipsaw song' in it's full glory. The joy from profitable trades in GBPUSD, USDJPY, XAUUSD, EURUSD, NZDUSD and CADJPY was not enough to drown out the lashes handed out by the balance of instruments in the portfolio. Them's the breaks........

Performance results are as follows.


Comparative results are as follows.


In a nutshell, DTT is holding it's own and has outperformed the EDTT and EDTT1 over the period.

....but wait, there's more. Have a listen to a recent podcast with Nick Radge on Chat with Traders. It is great to hear sanity prevail. The performance of his strategies are an eye opener....... Some great looking equity curves in there.

Have a great weekend

C

Lol..excellent... I didn't know Ed had such talents

Thanks for the chat with traders link... I've listened to Nick Radge a few times, the interviews are always a great listen imho...

Have a cool Weekend C..

You too mate. :-)

Nice Article from HedgeNordic Comparing and Contrasting Macro and CTA funds.








A possible Bright Future ahead.......from the Editor....







Have a great week ahead.

C