Diversified Trend Trading Approach

From Systematic Trading of Mr. Carver.

Interesting math-based discussion why day trading is very hard:



and also table of implied volatility target given holding period and risk per trade.



And recommended instruments because of costs:

Fortunately with EDTT holding period is about 8 days... I am not sure about cost of CFDs.... and commissions....

"Blending of Return Streams - How to Do it"

.....continued from post 2558.....Part 2

I'm back after a nasty virus....apologies for the interruption.

We left our portfolio blending at post 2558 where our aim was to generate monthly percentage returns for a defined position sizing for each of our return streams that we want to blend.

I have constructed an excel spreadsheet which we will use going forward to demonstrate how we blend two return streams generated from two different systems. Download the spreadsheet "Sample Portfolio Blending Matrix) (which is attached to this post) and I will describe how it works in detail.



Column A and Column B above represent the return streams in percentage terms for two different systems. System 2 is actually the EDTT and System 1 is the Bollinger Mean reverting system described in previous posts. Both systems have been tested using a fixed trade risk of 0.25% of trade capital.

Before we can blend these return streams we need to create additional columns to generate the trade statistics we need for blending, namely the cumulative equity position (assuming we compound results on a monthly basis), the max equity position which we use to calculate drawdowns, the max drawdown and also excess returns which we use to calculate sharpe ratios. The spreadsheet contains the formulas you need to calculate these columns.

Take particular note of the blending columns (Column D, E and F). Column D and Column E have a multiplier in cells F28 and G28 where we can scale our different return streams and column F represents the blended result of this scaling.

I have also included an equity curve and drawdown graph in chart form that represents the blended result.

So now onto the blending exercise.

Firstly look at column D which represents the return series of System 1. Currently the multiplier is set to 1x. You will note that cell G17 reflects a CAGR of 3.9% and Cell G18 reflects a drawdown of 5.49%.

The first thing we want to do is to identify our max drawdown threshold. As discussed in previous posts, 10% is a level I am hapy with and within my risk tolerance. So given that the current max draw for System 1 is 5.49, I can effectively double the position sizing to achieve this outcome on a single return stream. Therefore set the multiplier in cell F28 to 2.



Ok....so the actual drawdown after applying a multiplier of 2 (or a trade risk % of 0.50%) is 10.96%. This is pretty close so we will leave this return stream at 2x.

So now we go to System 2. You can see that under a multiplier of 1x the max drawdown in cell G18 is 2.71%. To get to our approx 10% max drawdown level we can multiply this return stream out by 4x or a trade risk % of 1%.



So far so good. Now look at the result of the portfolio blend in column F. The Max Draw is 10.57% and the CAGR is 38.0%. Now this is where I play around a bit with further adjustments to attempt to maximise the MAR ratio (CAGR/Max Draw)

Notice if I change the multiplier of System 2 to 5x that I significantly uplift the CAGR from 38.0% to 46.0% with only a minimal uplift in the drawdown from 10.57% to 10.74%. Well I am happy with this and am prepared to take on the slightly higher Max draw to achieve that superior return.



So guys that is all it takes to blend the return streams. All you are doing is blending your different return streams to take account of their inherent volatility of returns and create an optimal weighting to achieve superior risk weighted returns.

I hope this all made sense.

Cheers

C

Ouch mate. That hurt :-(

Mate FYI the tests that we run across CFD's are inclusive of all trading costs including commissions and holdings costs plus a conservative factor applied to slippage on every trade. Rob Carver is dead right however that testing without the inclusion of all trading costs is a huge no no......as the frictional costs of trading are a significant obstacle to overcome and requires that your edge must be able to absorb them. That's why I always ensure that my tests load up the costs in the most conservative manner possible to ensure that under live conditions you don't receive a rude shock out of sample. :-) C

Yep, but this is a part of the journey.

Everyone wants to have such curve:



But this means for example 3 year drawdown or 5 month drawdown...



Both pictures are from the same investing program.

Thank you for this!

When we crank up the system 2 little bit more we get cool CAGR over 50% with still 10% DD.



And do I read it right and wonder what is this sorcery? Both strategies are losing and creating profit for the month?

Spot on R

Nice R. That's what playing around with it gets you. I do tend to work in whole numbers for the multiplier but you can certainly go decimals if you want to.


I wish it would...but it simply adds the return stream from System 1 and System 2. If system 1 did 2.4 and system 2 did -3.5 then the result would be -1.1. Just watch out for the way excel displays negatives in the cells. It can appear visually misleading.

Nice end to the short term momentum burst on DJI.....and we are still alive on the longer term SDC. God Bless America. :-)

A flying sumo nailed me on the short term but saved me on the long term. I don't know whether to hate these big guys or luv them....who woulda thunk it :-)



This is when having wide pending orders significantly above the last prior swing high/low saves your arse....a classic sumo in action at about the same level as the prior swing highs.

Also note that there is no clear short term momentum uplift on the last move up of greater slope than the longer term swing...... signifying that the momentum charge may not extend into blue sky and we could see an imminent turn at around the prior swing high. A short term momentum burst adds vigor to the long term momentum indicating an extension.

I must admit I really like M's idea of arming or unarming pending orders. I am continually adjusting my pendings to ensure conditions are met for short term and long term momentum conditions just prior to making a final decision to enter. Once happy with all setup conditions I can then just set them as armed and walk away.

It is a shame we can't define the setup more definitively so we can fully automate...... but there is so much going in in decision making at this point.

It makes life so much easier having the trade fully automated once setup conditions have been defined.

For you Moof :-)

Using the Weekend to slow things down, review pending orders and plan to prevent overtrading

As followers of this thread probably recognise by now, the propensity to overtrade on the EDTT is a big issue when dealing with the heat of the market. It often pays on the weekend or after a nasty trade week to examine your pending order schemata using a multi-timeframe analysis as this method of review assists in:
1. Ensuring that the logic applied to breakouts on M30 is embedded in longer term market structure;
2. Analysing what possibly went wrong with failed breakout trades that appear obvious in hindsight when quietly reviewing things on the weekend. I was guilty of this error as you will see in my chart annotation and have slapped myself over the wrists already;
3. Allows you to slow down away from the call of market noise and recognise that it is not the frequency of trades that is important to the strategy but rather the efficacy of trade application in relation to overall market momentum; and
4. Allow you to get back into the rhythm of the market.

Here is how I view my pending orders and active trades on a MTF basis for my entire watchlist. Remember there is no right and wrong as each of us with our different experiences may interpret things differently. The importance of this exercise however is that it slows you down and forces you to apply logic to your trade decisions as opposed to possible biases that may be lurking in your mind during the course of the active trading week.

....and of course with the necessary music to make you get into the rhythm of the market.
























to be continued......

C

Using the Weekend to slow things down, review pending orders and plan to prevent overtrading continued........























to be continued......

C

Using the Weekend to slow things down, review pending orders and plan to prevent overtrading continued........











That's it for the weekend analysis. I have reset pending orders and now I am comfortable that I am back into the rhythm :-)

Cheers

C

Food for Thought

 

1. A beginner’s foray into (part-time) systematic trading w/ Kory Hoang
2. Everybody’s Doing It: Short Volatility Strategies and Shadow Financial Insurers
3. A Look at Diversification
4. Nate Silver: Don’t Fall for His Predictions
5. Benoit Mandelbrot: Markets Are Not Efficient
6. Daniel Kahneman: The Psychology Behind Trend Following
7. Historical Events When Trend Following Wins Big: 500 Years of Financial History
8. Tudor to Shutter Discretionary Macro Fund in Restructuring
9. Vineer Bhansali, “The Market Is Severely Underpricing The Probability Of A Sharp, Catastrophic Loss To The Downside”
10. Trading the Trend Podcast – EP 015 – Getting Attached to Unrealised Profits
11. Ep. 611: Michael Gervais Interview with Michael Covel on Trend Following Radio

Enjoy

C

Sometimes on this thread and other threads:



from

https://customer.io/blog/curse-of-knowledge.html

I had so many cases in real life too....

LOL :-)

PS I am probably a bit like you R. I have worked with numbers all my life yet I don't regard my skills as being mathematical. They are more visual in nature. Throw a statistic at me and it doesn't mean much....but throw a visual relationship at me at it means much more. That's why an equity curve means so much more to me that a set of performance statistics.

When I blend portfolio's I simply plug in any multiplier and see what the resulting curve tells me. I then choose lower or higher multipliers to get the optimal return series that I am after. A mathematician would shake his head and suggest a much more economical formula....and then I would just scratch my head and pay homage to them :-)

I think people are wired differently and as a result there are many ways to skin the dead cat.

PPS.... I hope I am not a "tapper". One of the things I attempt to do in this blog is to hopefully explain stuff in a way that all can understand. I might just well be however as good ol' self criticism is pretty hard to do....so if there is anything that needs further clarification just give me a hoy and I will try to explain it the best I can.