You shouldn’t use a spreadsheet for important work

Because highschool math acquainted them with the notion that a function call looks like

  afunctioncall(argument)

thanks to sin(x) and cos(x) and log(x). Some people will even have vague notions of functions with multiple variable inputs like f(x,y).

Basically, highschool math introduced a lot of syntax that users will be familiar with - we're universally okay with the typical +-/* operators, for example, and ^ isn't a stretch, neither is f(). AND/OR/NOT are introductory Boolean algebra and while they're not universally taught in high-school, I find many scientists are acquainted with them (although languages using || && ! syntax are a good way to lose them). Excel's lack of boolean operators (booleans operators are functions in excel) is disappointing.

I've always found that SQL gives a good minimal arithmetic/boolean-logic toolkit for laymen. It's a good model to follow.

The beauty of Excel is its ability to do many things 'well enough' in a way that is accessible to a lot of people.

My dad put together his architectural plans for a new house in Excel because it is a tool with which he felt comfortable. Some people do calendars in Excel while others try and recreate a full General Ledger system. Because it works well enough, people don't see the need to invest the time in learning a new application and instead invest their time in pushing the limits of the tool.

As someone who builds financial models and who audits those built by others (and is a competent programmer by night), I think that the key source of risk in Excel models is that the tool has no knowledge of intentions. What I mean by this is that if I'm building a cash flow model or if I'm doing a pixel drawing, Excel doesn't care; as users, we are forced to create our own structure and build in our own checks and balances. If I make a balance sheet in Excel that doesn't balance, Clippy won't show up and let me know that things are broken.

I've often thought that it would be really amazing if a semantic layer could be built that uses Excel as the calculation backend. This sort of tool could understand the sorts of concepts of financial statements, projections, time-series and other concepts that often show up in financial models. It would have a built-in understanding of the domain-specific models that would let it leverage that understanding to reduce risk in the building of financial models. If I told it that I wanted to add a revenue stream that is tied to the output of production, the tool would connect the dots between the production schedule, any inflation and/or foreign exchange assumptions and would feed changes in working capital according to the associated collection terms, etc...

Before I get too carried away, the point is that this type of semantic layer would be much better at preventing and detecting anomalies and potential errors in the development of a high-risk financial model. Does anyone have experience with any such tools?

>"My problem is that Excel is terrible"

Do you know a better alternative? I have used LibreOffice and Numbers and I will take Excel any day. Saying that it is terrible is a very strong statement for a piece of software that arguably is what keep a lot of people in the Windows world.

I used to work for Deutsche Bank. All financially modeling ran through Excel, though admittedly a massive C++ library that is a big Excel macro plugin. These sheets get massive and there are a few clusters of blades that do nothing but run Excel macros day and night.

But why is this madness? It works for them. The iBanking world is the madness. You can put in place effective change controls and auditing with any system. However, if people just want it done and don't care how, regardless of whether these calculations were in Python, Mathematica or Fortran, garbage in will still be garbage out. So, let's not wantonly scapegoat Excel for a process problem here.

I stumbled across Pyspread the other day, a more recent attempt at a spreadsheet using Python at the cell-level. See http://manns.github.io/pyspread/.

  Pyspread expects Python expressions in its grid cells, which makes a spreadsheet
  specific language obsolete. Each cell returns a Python object that can be accessed
  from other cells. These objects can represent anything including lists or matrices.

I haven't used it yet, most of my personal spreadsheets are in Google Sheets and I'm not too keen to pull them out of Google Drive.