May 16, 2010 by Andrew

< continued from part 3 >

Optimisation?

Yes.  Optimisation.

.

Another powerful (and therefore potentially very dangerous) feature of Wealth-Lab is the ability to optimise a model, by selecting a range of values for the functions/indicators/variables you use in your model.  For example, you could test which values produce the ‘best’ results for a Simple Moving Average Crossover against a specific stock, a range of stocks, or a market of stocks or other Financial Instruments.

Using BHP as a specific example: Let’s say you have $10,000 to invest, and are willing to risk 50% of your Account on any one order for BHP stock.  If you were running a simple model that triggers a Buy Order whenever the 20-day Simple Moving Average (SMA) of BHPs Closing daily share price crosses over the 60-day SMA, and Sells (Closes) the order when the 20-day SMA crosses back under the 60-day SMA, you would have generated a  return of 64.21% from January 1, 2001 to December 31, 2009, for an Annualised Gain of 5.67% this may not seem like a great return, but just imagine the impact of buying these shares on a Margin of 10x your investment (as you can with CFDs)!   The BHP return came from 18 trades, only 9 of which were Winning trades.  You should note, in the table below, that trading this model also generated a Maximum Drawdown of 15.65%.  OK in itself, but imagine the impact of a 10x Margin on this drawdown – You’re broke!

Results of BHP 20/60 Long Crossover

I took this simple model, and Optimised it for BHP, using the period of 01/01/01 – 31/12/06 as my base date range.  The optimal SMA Crossover results are to use a 20-day SMA as the trigger, whenever it crosses BHP’s 105-day SMA.  Re-running this optimised model for the entire 1/1/01 31/12/09 period produced the following result.

BHP Optimised SMA Crossover

Please note: the above Optimisation process on a single Stock is very simplistic and very dangerous, and I do not recommend its use.

To me, the advantages of using optimisation on a model across an entire market were twofold: Firstly, it allowed me to modify working models that others had submitted to the website, so that they “performed better” in my preferred markets (Wealth-Lab is US based, and the test results published on the website are based on tests against an array of US stocks).  Secondly, I didn’t want to run the same models using the same variables as others, and therefore trigger the same Buy and Sell orders as these others.

Understanding a little about correlation, and wishing to take advantage of non-correlated investment tools, I wanted to run a version of my Reversal Trigger Factor model with 2 or 3 other Model/Market combinations, and begin paper trading.

Unfortunately, the Reversal Trigger Factor and other models I was testing did not, based on historical backtesting, produce the returns I was seeking (please take note: I was seeking exorbitant, unrealistic returns). What to do?

The obvious solution was to boost the return, using a margin account or Contracts For Difference (CFD).  A Margin account would boost these returns by 100%, whereas a CFD account would boost my returns by 500%- 2000%.  Woo hoo!

I had first heard of CFDs when I lived in the UK.  Much of my spreadsheet based modelling I had created in the previous few years had been based on CFDs, so I decided to learn more

There was plenty of information available on CFDs.  Various websites were devoted to these little (very dangerous in the wrong hands) miracles, and some of these websites reviewed the various CFD providers (brokers) offerings.  I perused the CFD Providers websites, and downloaded and read their Financial Services Guides (yes, I am the person who has read many of these documents).  This process allowed me to both learn about the opportunities and risks associated with CFDs (although reading about “potential losses” is far less educational than experiencing them yourself!), and to select a CFD Provider for my initial foray into the market.

Having read the theory, and selected a prospective CFD Provider, I downloaded the associated trading software from the Providers website, and signed-up for a Trial account.

The trial account process is invaluable.