I have found an excellent resource for computerized back-testing of trading systems based on the AAII weekly database, and that resource is Keelix.com. For those capable of understanding pseudo code and aping unfamiliar syntax, it should be relatively self-explanatory. If one is additionally familiar with the data available from AAII, then doing the backtesting should be pretty simple. Due to the data structure containing limited technicals and weekly updates, it’s more suited to FundaTechnical criteria and longer timeframes, so day traders that are happy day-trading can probably ignore this post.
The original idea of the Phantom Trades was to take some stocks meeting Value or GARP criteria and, over time, develop some technical criteria for speculating in them individually. I think the “phantom trade” idea is moot for me now, after gaining access to better data. Having more data available, to me it makes greater sense to forward-test the “basket holding” techniques that back-test well, rather than spend time developing stock-by-stock technical speculation methods. The Value and GARP ideas are on my list, my “mix tape” quant idea is, too, as well as one new idea based almost entirely on fundamentals, and two blatant rip-offs of others’ previously-developed ideas I liked best based on a review of their risk-adjusted returns – and no, I didn’t calculate their Jenson’s Alpha! LOL.
I strongly urge anyone who is serious about speculation in individual stocks to spend some time browsing all the material available at Keelix. Many people have taken their own stab at mechanized stock investing, and we can stand on their shoulders when trying to develop out own ideas. Bookmark it, play with some code, review the dozens of existing ideas others are testing, and test a few of your own. Even “value investors” should go there! Run twelve monthly-staggered (or four quarterly-staggered) tests with holding periods of a year, and viola!, you can build your own variant of the “Magic Formula” value trading system. Go to Keelix now.
13 Comments
Hi Bill,
Great blog here.
Is this back-testing free?
Thnx.
Yep, it’s free. Read the FAQ at the site, there is a registration and he logs IPs so if there is abuse he can take action. There’s also a forum, but it’s not very active.
Now we can all follow the same strategies that brought the demise of hedge funds during august. Thanks Bill!
i have not visited that blog (yet) but i used TradeStation to do my back testing and development of a mechanical trading system. i use it exclusively now for all my entries/exits in my Trading portfolio.
What kind of fundamental data are available at TS?
Tradestation 8.3, which is going through a phased roll-out and i do not have yet, is supposed to have historical fundamental data that you can use in your back testing and strategy development.
The current version has the run of the mill fundamentals, such as balance sheet info, cash flow, income, p/e, shareholder equity.
my algorithms are purely technical based on volume and price breakouts and trailing using a combination of a channel breakout and moving averages.
as such i have not used their fundamental data much.
You should browse the fundamental and FundaTechnical data available at the AAII & at Keelix. The weakness there is a lack of technicals. If you have the time, perhaps a set of “good” stocks from there could be traded on the day to day technicals. That was my plan to build on with the “phantom trades” but given my personal time constraints, I’d rather focus on holding baskets of stocks for weeks instead of trading handfuls intraday.
i am not day trading. I have been backtesting for months on thousands of stocks going back many years.
My goal is to get some parameters that work for longer term trend-following based on the turtle’s approach but adjusted for stocks and my comfort level.
i am still tinkering with it, especially for short setups, but i am finally comfortable with the long setups.
i just review my holdings once a night to make sure no exit signals got triggered. When i am not fully invested i run a scan in TradeStation on about 6000 stocks to pick new entries.
I tried playing with some fundamental data but found that in my system technical breakout usually precede fundamental triggers.
My system does not generate a re-entry signal till i am stopped out, so if i do not get in on the first breakout i may miss a big move.
Check out the Motley Fool discussion board Mechanical Investing. Extensive discussion on the Keelix backtester.
Also check out portfolio123.com. Their backtester is based on the same fundamentals database offered by AAII (Reuters/Mutex), plus it implements a number of TA functions. Very active community with many models and ideas available to start up your own studies.
Has anyone ever backtested backtesting? My bet is that here is no correlation between past trends and future trends. If I’m right then backtesting has zero value.
Good luck with that bet, Martin, it sounds very Niederhoffer-esque.
Bill,
Thnaks for this suggestion. It will be fun investigating it….dj