InvestorsHub Logo
Followers 18
Posts 877
Boards Moderated 2
Alias Born 02/16/2002

Re: irwin post# 409

Friday, 11/22/2002 8:00:45 PM

Friday, November 22, 2002 8:00:45 PM

Post# of 1453
Plans right now are to have a one signal file per Model limit.

Basically I don't want to make the software complicated to use and when you start defining external files in a Model, some people start to become uneasy. So I might introduce two AI editions, the standard edition that includes everything that's there now and a professional edition that includes the signal file filter functionality plus a few more things I'm working on. I'm still thinking about it.

If I do decide on the Pro edition, then I see no reason not to allow multiple signal files per Model (as anyone purchasing the Pro edition would presumably know what they were doing regarding signal files).

The other thing I have half-implemented at the present moment is an asset allocation function. Here's how it works...

1. You select the stocks/funds you want in your portfolio.

2. You select a Model for each stock/fund (just as you do now in the historical analyzer).

3. You select a date range and average commission as well as your TOTAL initial investment. Then click the appropriate button to generate your asset allocation chart.

4. The asset allocation module will historically analyze the performance of all the selected stocks/funds based on their associated Models. Once it's complete it will optimize your entire portfolio by generating a recommendation on how much of your initial investment to allocate to stock 1, how much to allocate to stock 2, etc. You'd then create your individual Automatic Investor portfolios based on the asset allocation module's recommendations. Or, if you already have existing portfolios, you'd rebalance them according to the recommendation.

If this sounds a little like Modern Portfolio Theory, it's because that's where I got the idea (I presented this idea at the AIM 2001 meeting and there's an article on the AI website). Right now I'm trying to decide on whether to use the MPT algorithm to generate the recommended portfolio components or use a heuristic algorithm.

MPT works great (because it will give you the optimum portfolios based on your inputs), however it makes quite a few assumptions on the input data (expected returns, standard deviation and correlation matrix). Some of these assumptions might not be valid. Also, a small difference in the inputs can result in large differences in the recommended portfolio component portions. So it's important to provide very accurate inputs.

The heuristic algorithm, on the other hand, seems to work better at predicting long-term future allocations because it is more forgiving with the inputs, although it doesn't always provide the most optimum allocation.

So although I like MPT, I'm currently thinking heuristics might work better. I'll post more when I've decided.

This feature will be available early next year as a free update.

Regards,
Mark

http://www.automaticinvestor.com

Join the InvestorsHub Community

Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.