I hate to break the news but banks use floating point numbers all the time, for financial calculations. They shouldn't but they do and they use all sorts of tricks like alternate rounding schemes and some pretty deep knowledge of floats to try and keep the books roughly right. I have never seen a Money type nor anything like DEC64 in a real commercial system, its all floats or ints/longs for pennies.
> I have never seen a Money type nor anything like DEC64 in a real commercial system, its all floats or ints/longs for pennies.
I've been working in ERP land for a long time working with many different commercial systems of different sizes and market share and I've only seen NUMERIC/DECIMAL types used in DB for things like Price, Cost, Amount etc.
The only time I've ever seen floating point is for storage of non-money type values, like latitude and longitude of store locations.
That’s not too surprising to me, I imagine that many number types are used, and it depends entirely on the task at hand? If you have deeper knowledge of what gets done in practice, I’m still curious what criteria and types and how many decimal points might get used in the most restrictive cases. What do people use in practice for, say, interest compounding on a trillion dollars? I can calculate how many decimal places I need at a minimum for any given transaction to guarantee the correct pennies value, but I don’t have first-hand experience with banks and I don’t know what the rules of thumb, or formal standards might be for safe & careful calculation on large sums of money. I would imagine they avoid doing floating point analysis in every case, since that’s expensive engineering?
My electricity and gas bills both charge me $0.xxxxxx per unit price, or 10,000ths of a ¢, although the last digit is invariably a zero. I've also seen 4-5 digits on currency exchange places.
I'd have to break out a calculator to be certain, but my guess is that most of these transactions amount to sum( round_to_nearest_cent(multiply(a * b))), where a is a value in thousandths of a cent--which is to say, there isn't a single "this is the fixed-point unit to use for all your calculations."
For financial modeling, the answer is definitely "just use floats," because any rounding-induced error is going to be much smaller than the inherent uncertainty of your model values anyways. It's not like companies report their income to the nearest cent, after all.
Right, the number of digits you need for any given calculation depends on the magnitudes of the numbers involved, and the acceptable precision of the result. BTW round to nearest might be unsafe, I’m certain that there are many situations where people will specifically avoid rounding to nearest, I would not assume that companies use that scheme.
It seems like a decent assumption that financial companies are not spending engineering time (money) on a floating point analysis of every single computation. They must generally have a desired accuracy, some default numeric types, and workflows that use more bits than necessary for most calculations in exchange for not having to spend time thinking hard about every math op, right? That’s how it works everywhere else.
The accuracy used for reporting doesn’t seem relevant to the accuracy used for internal calculations. It’s fine for large companies to report incomes to rounded millions, while it’s absolutely unacceptable to round compound interest calculations in a bank account to millions, regardless of the balance.
Oh now that is somewhat surprising! Is this 64 bit or 128 bit floats? 32 bit floats aren’t accurate enough to represent $1B to the penny. Do error values get stored as deltas and passed around with the float values? Would love to hear more about this, do you know of any good summaries of the engineering workflow online?
Wonder if it depends on which side of banking: investments, accounting, clients side (web pages, and ui) vs the backend. Or if we're talking about central banks vs a small credit union.
I have done retail and investment banking as well as hedge funds for front and back office. Everything from giant banks everyone knows through to small hedge funds no one has.
Thanks. That’s fascinating. I have zero experience with any banking/finance stuff but through tropes and various anecdotal accounts was somehow sure floats were not used by banks. I guess it’s one of those persistent rumors that just refuses to die.
The noob-expert meme is really apropos here, as floating-point is really the option to go for if you know nothing about it or you know a lot about it.
The "problem" with floating-point is that you have to deal with rounding. But in financial contracts, you have to deal with rounding modes. Fixed-point (aka integers) gives you one rounding mode, and that rounding mode is wrong.
With ints/longs they usually use something that alternates the rounding so that it spreads, but that alternation can be done on a system wide basis, on a machine basis or in an individual account.
For floats what they tend to do is carry around the estimated error when they are doing a lot of calculations and then adjust for the error at the end to bring the value back.
Both of them are trying to deal with the inherent issues with the representation being biased and ill suited for money in practice. But oddly this is never pulled together into a Money type because it really depends what you are doing at the time, sometimes you just round and move on and sometimes the error is expected to impact the result because you are dealing with millions/billions with calculations on 10/100s so its going to matter.
But reality is the books are basically off by pennies to pounds every day because of these representations and its part of the reason no one worries about being off a little bit because the various systems do this differently.
