Berkshire Hathaway’s Stock Price vs. 32-Bit Integers

They probably just use BRK.B for most HFT involving BRK. The one-way transfer from Class A to B shares (and the different voting rights) reduces the supply of Class A shares due to a myriad of reasons. This makes the daily volume for Class A shares lower than Class B (when normalized by the trade ceiling imposed by Berkshire).

Oh boy, the possibilities of an automated trading disaster triggered by some algorithm coded without consideration for unsigned wraparound.

https://www.bbc.com/future/article/20150505-the-numbers-that...

Here is a screenshot in case they fix it:

https://i.imgur.com/8z3s0qA.png

It is a rare day that holding BRK.B is preferable to BRK.A. Perhaps today is one of them.

If NASDAQ has this problem, there is a good chance that a number of brokers and back-end systems do too.

Edit: The more I think about this, the more I expect that Berkshire will be forced to split. We'll know within a week or so, I guess.

Berkshire-Hathaway: chaos-monkey-elect of May 2021.

Google Finance has the worst corporate actions processing of any public source of market data. Their splits and mergers are wrong in a lot of cases, not just this exceptional one.

Found the person who added all the special cases to the legacy code I’m working on :)

YTs counter had was even signed. So the first dirty fix was very easy, because there was an unused bit. This is different, because all bits are used in the stock price, so no easy dirty fix.

The hallmark of great software engineering: tell product to adapt to technical lazyness

Processors have had BCD (binary coded decimal) instructions since the earliest 8-bit micros. They usually deal with 0-99 and then you have to add digit by digit but it can handle carry propagation.

When I worked on a semi-similar problem (a realtime-ish feed with lots of numbers), we found that it was better to just eat the extra RAM costs.

Once you have variable width values, it gets much more expensive to answer questions like "Here's a file with every stock price, we know that Berkshire is the 1,000th one in the list, what is their current price?" Instead of being able to jump straight to the relevant bytes, you end up needing to parse (potentially) the entire file. Even with better CPU support for variable integer sizes, you're still essentially turning an O(1) operation into one that is O(N)

You don’t really need dedicated instructions for them. Programmers can do that today, and many decades in the past too, with software. For example, Java has BigInteger in the standard library for decades now.

The reason they did not is less about memory (it would save memory on average, many numbers in that system need less than 4 bytes) more about CPU time. That system is likely very old, older than wide adoption of 64-bit processors.

The reason we don’t often use variable-length integers now, memory and storage are cheap. For new software people simply use 64 or 128 bit integers.

Still, variable-length integers are widely used when storage or bandwidth matters. They are all over binary network protocols, multimedia containers, and media codecs.

NASDAQ programmer, somewhere around 1982: "What kind of company would let their stock get over $400,000?"

Exchanges compete on speed. If exchange A delivers data in half the packet size and requires smaller word sizes to operate on and exchange B decides it's okay to double the packet size and requires twice the word size... exchange A gets the trade and exchange B is left holding the bag.

Back in the day the major exchanges, NASDAQ and NYSE in the U.S. and the TSX in Canada used to operate as a mafia with a monopoly on the stock market. Entire exchanges came into being strictly on the basis that they had lower bandwidth requirements and operated faster, such as BATS and DirectEdge (which ended up merging), CHI-X, Omega, and a host of competition emerged to actually focus on technology. They ended up taking a large chunk of market share away from the major exchanges.

Now NASDAQ and NYSE have since caught up, and in some ways they did so by buying out these competitors. Modulo some embarrassing glitches over the past few years, the major exchanges need to continue supporting their legacy systems for major institutions who are very slow and hesitant to upgrade any aspect of their technology.

NASDAQ has a more advanced and modern data feed and that feed doesn't have this issue, BRK.A is being reported properly at its price as we speak, but that feed isn't widely adopted except by firms that generally build their own technology and have a team that can keep their tech up to date. The feed that has this issue is like 20 years old now and rarely gets updates, and many firms like to keep it that way.

>Perhaps a CPU built around a base numeric type more like UTF-8 would be useful? Short values would take only a smaller amount of RAM but longer values would still work?

https://en.wikipedia.org/wiki/Carry_flag

One reason is that they now have the highest stock price and that's a cool selling point for BK. Not their only one, clearly, but a neat point to make.

BRK.A are "voting" shares. My understanding is that they want to keep the "voting pool" to be richer / upper class people.

Rumor is that the shareholders meeting is a giant party, a who's-who of the elite world. You can only have that culture if you have a very high barrier to entry. In any case, the high share price ensures a certain class of people in the voting pool: an old-school sense of community and elitism but... hey, you can't argue that it doesn't work.

Not much other reason needed.

I read it's because he only wants the most absolutely committed investors owning shares, not superficial speculators. You would overthink if you had to spend that much on a single share.

Although modern platforms with their fractional shares might negate the advantage a bit.

Buffet has said before that it's intentionally high to keep out short-term traders and inexperienced traders. He wants investors who will hold for long terms. If the price is so high, there's only a particular clientele that can afford to buy it.

They do have a different code of class b stock that is more affordable.

My guess is they didn't consider the possibility when they designed the system, and had forgotten the limitation existed by the time it became obvious that it would happen.

No they knew it but this particular feed having the issue is an older NASDAQ feed.

NASDAQ's newer feed doesn't have this issue but a lot of major institutions don't like updating their infrastructure so they stick to the older feed.

To be honest, it's not nearly as big of a deal as it's being made out to be. It's not a situation where NASDAQ was ignorant of this issue or neglectful, it's a situation where NASDAQ has to balance support for legacy systems by not doing a binary incompatible update to a protocol that has worked well for many users for 20 years... versus pushing older institutions to update to their newer technology which doesn't have these issues.

I'd imagine they didn't think that a company would ever get above even 5 figures, much less 6. Buffet's refusal to split is also not normal.

I think they absolutely could, and the SEC would be OK with it, but BRK.A is listed on the NYSE, not the Nasdaq, so it doesn't have a contract with them. Like other US stocks, you can still trade it on other exchanges than the listing one.

Longest pun setup in history

This should be the clickbait article title.

Isn't the real solution just to use a bigger int?

You cannot live with it in any system where agency can be taken because of how exploitable it is. If you set up a genetic algorithm in a life simulation and foolishly set it to use floating points "nature" will select for utilizing free energy as a matter of course.

HFT + floating point = economy crashed by duping inflation without real item duping

Floats are used in finance for many applications. Most of the front office applications I've worked on use floats for money, because it's like science: the model is already off by more than floating point error.

The whole "NEVER use floats for money" thing is cargo culting. In this situation though I don't see why floats would be any better a solution than just using 64-bit integers.

In scientific computations, it is often sufficient that the final answer is correct to a few significant figures. There are several reasons for this. One is that the model uncertainty is typically quite large (unless you are dealing with particle physics). Another is measurement uncertainty. Except for some unique scenarios measurements aren't going to have more than a few significant figures as well. So there isn't a big difference between 3.14 and 3.14159 and pi to 100 decimals.

I'm sure one can think of plenty of counter examples, but I'll give an example where floating point numbers are used heavily and you don't typically don't care about minor rounding issues.

Computational chemistry can be used to predict the geometry of molecules. The model used to make this prediction has to describe the interactions between the atoms, which is computationally expensive (solving Schrodinger's equation). So often times the molecule is only modeled in a vacuum and not as a liquid (where there would be many other molecules included). All of these add up that even if there was no IEEE-754 floating point error, the answer would still maybe only be right to +/- 10%.

So science is quite different from finance, where the main goal is to converse a certain quantity. Money shouldn't be created or lost though rounding errors.

Rounding error in finances is a major plot point in the introduction SuperMan III. https://www.youtube.com/watch?v=N7JBXGkBoFc

Sure, it's humorous and a little bit silly, but the issue is a real one, and this is a funny way to familiarize yourself with the tricks that can be played.