On Monday, Bank of America revealed that it may have hidden as much as $4 billion in losses from regulators over the past five years. The bank said it was an honest mistake. But this seemed like another “shame on you Wall Street” moment.

Making matters worse: Bank of America BAC asked and just received approval from the Federal Reserve to up its stock buyback plan by $4 billion, a figure that nicely matches up with the losses it was hiding. That is probably a coincidence, but it made the revelations even fishier. Regulators, unsurprisingly, have put Bank of America’s buyback on hold for now.

Outspoken CLSA bank analyst Mike Mayo seized on the moment to state once again that this is more evidence that Bank of America is too big to manage — the go-to response these days to any big bank loss — and says he plans to point that out at its annual meeting next week. “It’s garbage in, garbage out,” says Mayo, who has been tough, and often correct, on the banks. “They have a $2 trillion dollar balance sheet. What other losses are in there? It raises issues about control.”

MORE: Citigroup’s bad bank is doing better than its good one

The losses were on debt that BofA inherited from Merrill Lynch. Mark-to-market accounting rules require banks to take gains and losses when the value of their debt falls or rises, even if what they owe is essentially the same. They are unrealized gains or losses, at least until they try to sell their debt or settle it with whoever they owe it to. Nonetheless, such shifts have to be factored into the earnings that banks report to shareholders and the Securities and Exchange Commission. BofA did that, which is why the bank won’t have to restate any of its earnings, which is not usually what you hear when it comes to bookkeeping errors.

Bank CEOs, notably JPMorgan Chase’s Jamie Dimon, complain about mark-to-market rules, especially when it comes to debt. You could call it whining, but regulators don’t like the rules either. So even though account rules forced BofA to factor those unrealized losses on the Merrill debt into the bottom line it reported to investors, regulators allowed the bank to ignore them when it came to calculating its regulatory capital.

So here’s where we come to the part where BofA screwed up: At some point over the past five years, BofA started to sell that Merrill debt — it began with $60 billion of structured notes and now has $30 billion — which turned those unrealized losses into actual losses. At that point, BofA should have told regulators about the losses — losses that it had already told shareholders about — but it never did, until a week or two ago when it caught the error.

BofA’s excuse is that it told an employee to update a spreadsheet that kept tabs on those losses, but the file never showed up, or the wrong file was sent. That’s a really lame excuse, and might lead some to question whether BofA was trying to hide something. But lame things do happen, even at the nation’s largest banks. So you can’t totally discount BofA’s excuse, even if it is of the dog-ate-my-homework variety. Although it’s still not clear why it took BofA years to figure out that that person wasn’t doing their job, or why that was found out now.

MORE: Banks need another $68 billion in order to be considered safe

One question worth asking is why BofA’s shares fell by just a little over 6% on Monday, wiping out $10 billion in market value, far more than the actual losses. I get why regulators were so upset. They were given the wrong numbers. But why are shareholders so peeved? They were never lied to. BofA reported the correct numbers to them all along. And BofA can clearly handle the $4 billion loss. It has. The money is already gone. And probably has been for years. Maybe investors should feel better about the strength of BofA. It can lose $4 billion, and it won’t even affect its operations. It won’t even notice it’s gone. I mean, it hadn’t noticed, and neither did the regulators.

Perhaps the better question to ask here is, Why are banks forced to keep two sets of books, one for regulators and one for shareholders? (Actually, there is a third set of books for the tax man, but that’s another story.) That’s not the sort of thing that regulators usually encourage. Bernie Madoff kept two sets of books. So did the London Whale’s assistant. But at the big banks, it is now required.

And that doesn’t make a lot of sense. Are the books the banks keep for regulators any more accurate than the ones they use for shareholders? BofA did end up actually realizing a loss on the Merrill debt. And it probably will have additional losses on the remaining $30 billion in notes it still owes. Wouldn’t regulators be better off factoring that into their calculations now? And if the regulators’ books are really a better way of measuring things, shouldn’t the SEC adopt that method for investors?

“Regulators seem to want to do their own thing,” says Jack Ciesielski, publisher of The Analyst’s Accounting Observer. “As a result, we have a system that is so complicated that banks can’t get it right. And this is another example of that.”

Banks are odd creatures — especially megabanks like BofA, JPMorgan JPM , and Citigroup C — made up of investments and loans that generate revenue, but the true value of which we won’t know for years. So there is never going to be a perfect way to measure how much they are worth. We would be best served to pick one method and stick to it.

But that’s not what we have done. Instead, we have a system that is prone to errors and surprises, especially when regulators are looking more closely, which is what they have been doing for the past few years, though apparently not close enough to notice BofA’s not-so-hidden $4 billion loss. If that’s not enough to cause bank investors to run for the hills, I don’t know what is.