Subject: Re: NDX 100 Momentum Strategy - Code Repository
Robbie is a unique phenomenon whose gift of genius and determination elevated many of our investment capabilities. I certainly hope that he has just decided to change his focus on life and that he is still in good health.
I was always in awe of his clear thinking and excellent math skills which evolved into programming skills.

I just spent a while looking over some his notable milestones in the Mechanical Investing Archives

His earliest post in the archives is 7/19/2000 but he was at MICon 2000 which I believe was earlier in 2000.

Robbie: 2/2001 I`ve created an Excel 2000 spreadsheet that backtests screens of the DH-RS variety, that is, screens that first sort (ascending) by a recent return measure and then by a relative strength measure. With a little more effort, it will, with the same retrieved data, backtest both DH-RS and RS-DH screens (that is, those that first sort by relative strength, and then by recent return) at once. With yet a little more effort, I`ll have it capable of backtesting Mr. Toast`s Monster variety of screens and RS overlaps. Currently, though, the spreadsheet is 37 MB unzipped. Its size is due to the fact that it computes the selections and CAGRs for a given screen over all standard test runs at once.
Using a 500 mhz processor, changing the relative strength measure, recent return measure, or both requires about 30 minutes of data transfer from Access to Excel.

By 2004 Keelix was building his backtester with SIP data in a MySQL database running on his 1.8GHz Dell Laptop with 512MB RAM. His beta was up 4/30/2004 at kjeldahl.net/investing/ and data back to 1999 a month later.

By 9/2005 Robbie was calling for an on line backtester and uploaded GTR1-Linearized Sux@BeU Data.

On 11/24/2006 Robbie posted: C++ Sux Backtester Available: I started experimenting with how many backtesting tasks I could translate from Excel/Access VBA into VC++ 6.0 using the smattering of C++ I had picked up while working on VLWRP. In doing so I was mainly seeking better performance, but I also would like to one day know enough C++ to understand the GTR1 linearizer that TechCzech built for me. One thing lead to another and I ended up doing what I never thought I would have the time or patience to do: I moved all of my backtesting operations into C++. . . My C++ backtests are about 25 times faster than my Excel/Access VBA backtests, and about 150 times faster than the same backtests I conducted in 2001 using a Celeron 500MHz CPU. . . and the web-based GUI that others will hopefully build.

By 1/28/2007 Robbie prompted by VTAlumni “To do proper blends between VL and SIPro screens . .need to synchronize using daily data. . . Robbie: My full backtester (as opposed to the sample Sux backtester I`ve been posting) covers the entire US market, 1974-2005. I can add fields from any respectable data source to the stew, including SIPro, if anyone donates the data.

4/2007 Jamie and Michael each have on line versions of early robbie’s GTR1

2007 GTR1 came of age with detailed analysis of screens. YLDEARNYEAR Not Torpedoed, Value Line Analysts Torpedoed, High_Relative_Value Not Torpedoed. . .

7/2010 Keelix shut down his SIP backtester.

I tried contacting Robbie at the last email address I had for him but got no response.