Engineering principles for building financial systems

As someone not in accountancy but having worked in corporate and been forced to work on accounting software for managing budgets, approving transactions etc. - you have really gotten the short end of the stick from general usability perspective. There are so many things that don't make sense - even for the accountants that use the system on day-to-day basis - they have workarounds for (what I would consider) the most basic of things (or over-complicated solutions for something basic like manager approvals for logging expenses - why do I have to click through several screens that don't follow any sort of clear, logical flow). And it's so embedded in an organization - that you can rest assured once that client is signed up they're not switching that system. Same for HR software...

On the best practices, as noted by others, there are probably classes in your standard lib/ commons lib that cover this stuff better than storing in integers (e.g. BigDecimal in Java and Python Decimal have precision and rounding mode concepts built in).

Something I've found valuable is on is managing euler units/ratios (e.g. proportions 10^1, percentages 10^-2, basis points 10^-4) . Enforcing a standard that all interchange, data storage has either the same scale, _or is stored with its scale (and not in the name)_ will reduce errors hugely significantly.

> You forgot "consistent"

I once had an auditor tell me that the books are allowed to be wrong, but should be wrong consistently. I've even run into the timestamp issue, which generally turned out to be a non-issue unless the wrong date was used (like order date vs. shipping date for revenue recognition) or a considerable number of errors caused dates to end up in a different reporting period. If you're dates are consistently a few hours off, then they'll be consistently off across reporting period boundaries and that can easily be explained to an auditor.

> the necessity to batch reconciliation processes between these layers

I've never seen a reporting period closed without accountants making journal entries for manual adjustments. These can often be material changes.

> An accounting system does not convert currencies at will

This is correct, but for internal reporting you might need to. An accountant and cash flow will capture the important conversion rate when the funds are expatriated, but your finance dashboard showing sales in euros will have an executive that wants to see it in US dollars. It's impractical to use the conversion rate from the transfer that could happen weeks or months after the reporting period, so you come to a compromise of using a X-day FX moving average that's "close enough" and everyone's happy. What goes in your 10-K/Q will be a different story.

> Use integer representations of time

This sounds like a type system issue (and the linked Etsy post is also a type system issue), or more precisely languages / systems with a loose type system where you can just pretend a time is a number without having to explicitly say what that number means.

Conventions are good, but (real) type systems are better because they check everything every time.

Another interesting head scratcher (thank God I never had to deal with myself, but saw the mess it can make in the data easily) are government-enacted devaluations. And probably very similar to the concept, stock splits.