Comments on: Spreadsheets as functions

By: Giles

Giles — Tue, 30 Dec 2008 18:50:54 +0000

Harlan – glad you like Resolver One’s referencing syntax; here’s how you could use it to do something like SUMIF with conditions over two columns in a three-column range:

=SUM(c for a, b, c in D1:F3.Rows if a == True and b.endswith(“foo”))

You’re right that Black-Scholes is probably a smaller problem than you’d normally want to solve using RunWorkbook – we chose it as a small enough example that people could understand in a short screencast. Our aim isn’t really to replace code-based UDFs, but instead to help people who’d never dream of writing code to produce stuff that can be reused by other spreadsheet users.

Cheers,

Giles

By: Harlan Grove

Harlan Grove — Tue, 30 Dec 2008 00:58:21 +0000

I haven’t tried it, but the idea of using a spreadsheet as a udf seems suboptimal. Yes, you may be able to do interesting things with them, but in Excel terms, this isn’t much different than using Data Tables. Also, the Black-Scholes udf could be implemented with two single-expression udfs. The portfolio value udf could be replaced with a single, moderately complicated array function.

That said, I do like Resolver’s referencing syntax. Excel’s structured references into Tables is a baby step in the right direction. If only structured references included record selection syntax, e.g.,

SUM(A1:H20[Field1="XYZ"][Field3])

as a functional equivalent for SUMIF(A1:A20,”XYZ”,C1:C30). Better still if Field3 could be an expression, i.e., something equivalent to the n term in

SUMIF(A1:A20,”XYZ”,INDEX(A1:H30,0,n))

But that begs the question of how to implement OR criteria. Then again, simply making SQL.REQUEST a built-in function would solve LOTS of formula problems.

By: Mike Staunton

Mike Staunton — Mon, 29 Dec 2008 19:02:03 +0000

The MS article is just so horrible and impractical – I prefer to teach VBA as learning by doing – here is my call function, albeit with my own function vaCND for the cumulative normal distribution because the Excel one is not accurate enough

Function vaEuroCall(S#, K#, r#, q#, Tyr#, vol#) As Double
Dim d2#
d2 = (Log(S / K) + (r – q – 0.5 * vol * vol) * Tyr) / (vol * Sqr(Tyr))
vaEuroCall = S * Exp(-q * Tyr) * vaCND(d2 + vol * Sqr(Tyr)) – K * Exp(-r * Tyr) * vaCND(d2)
End Function

By: Giles

Giles — Mon, 29 Dec 2008 18:38:51 +0000

Simon – thanks for the mention!

dougaj4 – you’re right, how-to documentation is lacking for how to build UDFs using any spreadsheet. We’ve tried to do a better job ourselves with a load of screencasts (the one for the worksheets-as-functions stuff is here: http://www.resolversystems.com/screencasts/runworkbook/) and it will be interesting to see whether that helps with uptake among non-developers…

By: dougaj4

dougaj4 — Sat, 27 Dec 2008 07:42:57 +0000

That paper was an interesting read, but I’m not conviced that the proposed UDF on a worksheet model is actually any more accessible to the average user than the VBA UDF model.

From my perspective (engineering at the coal face) the reasons people don’t use UDFs as much as they might are:

1) Too busy to muck around with that stuff
2) Prefer to stick with what they know and what works
3) If they really need a UDF they’ll hand it to the “office office expert”
4) Lack of good “get you started” help packaged with the program, and surfeit of poorly-structured help on the Web.

I don’t think that lack of familiarity with programming is a big issue, at least in engineering. They do that at uni, and if they are into that stuff will continue to do it afterwards. If they aren’t, they aren’t going to do UDFs in a worksheet either.

By: Simon

Simon — Tue, 23 Dec 2008 14:34:37 +0000

I think they patented some part(s) of that too.

By: John Walkenbach

John Walkenbach — Tue, 23 Dec 2008 04:07:35 +0000

I’ve never used or done anything like that, but it reminds of of a 5-year old paper by Simon Peyton Jones: A User-Centred Approach to Functions in Excel.

http://research.microsoft.com/en-us/um/people/simonpj/Papers/excel/excel.pdf