Portfolio Optimization

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Jason
Roundtable Fellow
Roundtable Fellow
Posts: 57
Joined: Sat Apr 26, 2003 10:31 pm
Location: Oakville, Canada
Contact:

Post by Jason »

Rabidric and Stan,

You both touch on either end of something I often reflect upon when developing trend trading systems. There is a continuum with "optimal trading system that makes buckets of dough from return" on one end and "optimal trading system for attracting business to make buckets of dough from raising assets" on the other end.

One would think that naturally all clients would want the best return they can get over their time horizon. However, in general, (there are always exceptions) high returns do not attract as many assets as low volatility.

In simple terms, most people are responsible with and work really hard for their money. These same people do not know much about the investment business let alone our unique niche. When they decide to give some responsibility to an outside manager, they tend to be cautious. I think this is pretty natural. Have you ever noticed that you are more nervous in the passenger seat with a driver you do not know well than you are when behind the wheel yourself?

The fact is that there are not as many people with the balls, guts and understanding of this niche to attract a lot of capital to the optimal return model. However, building a business can be done here.

There are lots of people who want a smooth ride and the comfort of a large organization. Return is not as important to these people. You can build a hell of a business but you might not be as proud of your returns over time.

Needless to say, understanding where you fit on this continuum is important. Both ends have their merits. Most of us find a place in between.
Old European
Roundtable Knight
Roundtable Knight
Posts: 144
Joined: Mon Apr 19, 2004 3:04 pm
Location: Old Europe

Post by Old European »

Great post, Jason! Very true!
BARLI
Roundtable Knight
Roundtable Knight
Posts: 650
Joined: Sat Jan 17, 2004 6:01 pm
Location: USA

Post by BARLI »

so back to optimization...Pork Bellies is nto something to trade with as a trend follower . the best results for bellies was 28/180 crossover exp averages and as you results are not promising (performance results on 1 contract)
Attachments
W.GIF
W.GIF (28.2 KiB) Viewed 10438 times
W1.GIF
W1.GIF (23.43 KiB) Viewed 10436 times
W2.GIF
W2.GIF (14.78 KiB) Viewed 10436 times
BARLI
Roundtable Knight
Roundtable Knight
Posts: 650
Joined: Sat Jan 17, 2004 6:01 pm
Location: USA

Post by BARLI »

By optimizing, its for the first that my dual system can beat buy and hold in Eurodollars. the best pair is 7/166 (I didn't go higher than 180 days mov avg in optimization). Results are quite good with profit Factor 7.
(performance results table on 1 contract). Something for an unsophisticated trend followe to go with :D
Attachments
Eurodollar
Eurodollar
WA.GIF (14.89 KiB) Viewed 10435 times
eurodollar
eurodollar
WA1.GIF (18.15 KiB) Viewed 10435 times
rabidric
Roundtable Knight
Roundtable Knight
Posts: 243
Joined: Mon Jan 09, 2006 7:45 am

Post by rabidric »

jason- true, but i think you have slightly misinterpreted my point, what you describe appears to be more a question of investors and their appetite for leverage.

I was trying to convey that by optimizing on historical smoothness, you increase the chance of departure from this in walkforward. accepting a "rougher" historical equity curve, is less likely to dissapoint in the future.

things that appear to have great Sharpe/MAR ratios that you can leverage the heck out of, usually aren't the full six-pack. i.e. you may have great favourable skew and kurtosis on your trade distribution, but your Long Term(Capital Management :oops: :) ) return/drawdown distribution probably has the exact opposite characteristics.

if your investors are risk averse, then just use low leverage.

maybe you won't be able to quote such great historical MARs with my "rough equity curve", but then you maybe won't have to explain why you aren't meeting expectations either.... :wink:

ok sorry BARLI, back to your curves...
BARLI
Roundtable Knight
Roundtable Knight
Posts: 650
Joined: Sat Jan 17, 2004 6:01 pm
Location: USA

Post by BARLI »

Forum Mgmnt, I've run exhaustive optimization up to 180 day(dual crossover exp mov average) on each single commodity , total 36, and then came with the average as:

36 values exp avg Fast Exp avg/36

and 36 values of exp Slow avg /36

so i got 2 averages one is for the slow and one is for the fast. I guess its how exhaustive portfolio optimization can be done
stancramer
Roundtable Fellow
Roundtable Fellow
Posts: 60
Joined: Sat Feb 07, 2004 10:34 am
Location: Washington DC

Post by stancramer »

For what it's worth, here is the procedure I would use if I were you:
  1. Choose a range pf parameter values for #days(fastEMA) and #days(slowEMA)
  2. Choose a stoploss
  3. Choose a portfolio
  4. Choose settings for slippage and commissions
  5. Run backtest simulations of the above
  6. Export the simulation results to Excel for plotting
  7. Use Excel to make a contour plot of Modified Sharpe Ratio, and MAR Ratio
  8. Choose a parameter set with Delightful performance, whose surrounding neighbors also have Delightful performance
  9. Run a backtest simulation of the chosen parameter set.
Shown below is one possible set of choices for this procedure, implemented in TBB software. The first picture shows the ranges (list item 1) chosen. Runtime is 42 minutes, 20 seconds (14:51 plus 27:29). The second picture shows the Excel contour plot of Modified Sharpe Ratio. I have circled the peak value. The third picture shows the Excel contour plot of MAR ratio. I have circled the peak value.
Attachments
Contour plot of MAR ratio
Contour plot of MAR ratio
contour_mar.png (60.92 KiB) Viewed 10167 times
Contour plot of Modified Sharpe ratio
Contour plot of Modified Sharpe ratio
contour_sharpe.png (62.28 KiB) Viewed 10167 times
Simulation setup showing parameter ranges
Simulation setup showing parameter ranges
EMA2_setup.png (43.94 KiB) Viewed 10166 times
stancramer
Roundtable Fellow
Roundtable Fellow
Posts: 60
Joined: Sat Feb 07, 2004 10:34 am
Location: Washington DC

Post by stancramer »

I decided that the Delightful parameter set was (40,12). Here are its results.
Attachments
Simulation results using Slow=40, Fast=12
Simulation results using Slow=40, Fast=12
eqcurve1.png (29.32 KiB) Viewed 10162 times
BARLI
Roundtable Knight
Roundtable Knight
Posts: 650
Joined: Sat Jan 17, 2004 6:01 pm
Location: USA

Post by BARLI »

in my backtest of 12/40 I got 15% annual return :P
nickmar
Roundtable Knight
Roundtable Knight
Posts: 192
Joined: Tue Oct 26, 2004 12:38 pm
Contact:

Post by nickmar »

Market selection is a very big piece of the puzzle.
Macro
Roundtable Fellow
Roundtable Fellow
Posts: 70
Joined: Sat Dec 03, 2011 11:22 am

Post by Macro »

Jason wrote:Sign at the local drycleaner:

Cheap, Fast or Good - pick any two.


Trend trader's version:

Return, Robustness, Low Risk - pick any two.

:D
By far one of the most thought provoking posts I've read.
Post Reply