Daily Speculations

The Web Site of Victor Niederhoffer & Laurel Kenner

Dedicated to the scientific method, free markets, deflating ballyhoo, creating value, and laughter;  a forum for us to use our meager abilities to make the world of specinvestments a better place.

 

Home

Write to us at: (address is not clickable)

Steve Wisdom

15-Jan-2005
Computing for Newbies in 60 Seconds

J. T. Holley asks: : I have done "long-hand" on everything. I feel like a mad scientist with my notebooks! Matlab, C++, SAS, Sql, R, seem to be the route over all of those Metastock, Wealthlab, TtradeStation paths. but what order do I tackle them in? I have plenty of time and am in no hurry.

The earnest & gracious Moby enthusiast again asks how to move forward with his computer skills. Unsurprisingly no one has replied, because any reply would be flamebait, just as my thoughts on "why I love greyhounds" would quickly be refuted by the editor et al., who'd point out the superior merits of labs, collies, Weimeraners or Bernese.

But I'll throw caution to the wind and give a quick "Computing for newbies in 60 seconds, as seen by Mr Wiz."

Lots of beginners move from E#cel to E#cel + VBA, because VBA is included with E#cel, and seems the easiest way out once the E#cel spreadsheet gets so Rube Goldberg'y as to be intractable. (I'll write an essay on this point later) I think this is a mistake, because then it's necessary to understand the "E#cel object model" (i.e., how exactly do I point to a certain cell or row?) as well a programming per se. Too much to digest all at once, though I'll allow that Tom Downing is very facile with it. Better is to make a clean break and move to Liberty Basic which is a successor "in spirit" to MSFT's DOS Basic's of the late '80s and early '90s. It's much simpler to use than MSFT's VB6/VBA or VB.NET, because the core programming ideas are in the forefront, rather than the GUI widget wrangling. LB is cheap, the author is easily contacted and friendly, there's an active email list, and it's the subject of the Dummies book.

For my own purposes I love R, and have beaten the drums for this before. But in fairness I already knew SPLUS when I took up R, so I was an order of magnitude "ahead" of a typical newbie. Anyway, to use R you may as well know simple programming (i.e., data types, data structures, program flow, looping & etc) for which LB is a good tutor. R is free, is actively & vigorously developed, and has an active email list. To my mind it's "better", all things considered, than SPLUS/Matlab/Gauss/Ox/SAS/Minitab and the other programmable stats packages

Mr Bollinger touts Python which is perhaps the best & easiest of (what used to be called) scripting languages. Indeed a good choice also for newbies. Alternatives are Ruby, which is arguably better but less widely known and supported, PHP, used primarily for web development, and Perl, which is widespread but "like punishment" (to use the deriv expert's phrase) to read (and therefore described by wags as a "write-only language")

Which brings us to the C "family", C++/Java/C#. I counsel newbies to stay away. C++ s a result of miscegenation between C, a clean fast "procedural" language, and Simula, an "object oriented" (you don't want to know) language. For the hobbyist programmer, C is essentially "harder to read Basic," but C++ is overkill. Java and C# were developed later as "C++ for dummies," which of course suggests C++ itself is unsuitable for dummies

As to SQL, there's a time and a place, but the time isn't Day-1 and the place isn't.here.

Jeff Sasmor adds: Newbies are better off with something like Python. Why? It's mainstream enough that there are magazine articles and books for newbies to read and find examples. It's essentially typeless, meaning that you don't have to declare whether a variable is an integer, float, double, etc. The object-orientation is simple and unburdened by artifice necessary to shoehorn objects into C. There's really nothing you can do in C++ that can't be done in Python. There's a lot of floobydust in C, C++, and even BASIC that you just do not need if you are a newbie. And I say this as someone who has programmed in machine language, asm for X86 and DSPs, BASIC, C, C++, and Python.

And from Wil Kenney: I rarely use Excel anymore; used Stata in school but never since. R seems the way to go. However Tradestation is more useful for pattern running. You can get a lot more done in TS when it comes to OHLC etc. I believe the same would be true even if my R skills were more advanced.

Steve Wisdom, a Pennsylvania-born Harvard graduate, has been Victor Niederhoffer's chief man off and on for more than twenty years.

More by Steve Wisdom>>>

More from Chief and Chair on Statistics