Hacker News new | ask | show | jobs
by viggity 4484 days ago
I (being a developer) really like the concept, but I'm curious what they've done to verify the market.

It's my impression that the kind of people that need to resort to using spreadsheets as a highly abstracted programming language are (mostly) mutually exclusive with the kinds of people that would want/need this deep grain analysis. But I'm just some dude. The more I think about it, I could imagine some non-quant in the financial services industry may be interested in something like this.

1 comments

We are former oil and gas analysts and found that interpreting and understanding spreadsheets is an incredibly time consuming task. We have spoken to dozens of financial analysts (whom we have discovered are our primary user) and have found out that the problem is amplified by an order of magnitude for them! They spend literally hours deciphering complexity in spreadsheets. FYI, we are mainly targeting those who are in Excel for more than 70% of the working day and developing models (normally financial or market sizing ones).
> normally financial or market sizing ones

A-ha. Here's the hidden assumption. You're looking at sheets that have a handful of parameters/assumptions and span out their consequences. (You want to stake out the market of people selling Monte Carlo add-ins.)

Many people use spreadsheets as ad hoc accretion calculators/graphers/explorers. This very week -- I estimated about a hundred separate generalized-extreme-value models in Mathematica (working basically by hand to import data from Excel; but this because there was a lot of careful culling), then exported a number of statistics from the distributions. Then I spent a couple of hours a day later looking at scatterplots in Excel -- trying transformations, whatever worked -- that helped me show how those models isolated a particular global feature across the hundred-or-so distributions.

I was looking at your demo and shaking my head -- "is this for cash flow models? But it doesn't even have features for stochastic parameters..." But I guess I see it.