Comment by w4llstr33t | Hacker News Reader

Comment by w4llstr33t | Hacker News Reader

Parent Full thread

At what cost are they providing this service though? They make a ridiculous amount of money doing so. Shouldn't the exchanges have offered the best price, and provided liquidity, in the first place?

Flash Boys: A Wall Street Revolt by Michael Lewis painted HFT in a pretty bad way. I have read criticisms of the book, but it's hard to separate out bias from the criticism.

There's also all the heat on Robinhood about selling order flow, which I'm surprised was even news to regular investors. It's great they eliminated fees for normal trades, and I also understand most major brokerages sell order flow as well (and still charged for trades for a long time). I read that Fidelity is the only major player that doesn't sell order flow.

Do you have some sources that someone could learn more about this, ones that don't have a vested interest in painting it in a positive way?

Also, I'm still wondering, considering dark pools [1], and the inside information that would come along with that, since those trades wouldn't hit public markets, how the stock markets can be considered a fair place to trade?

[1] https://www.investopedia.com/articles/markets/050614/introdu...

> They make a ridiculous amount of money doing so.

I suppose it depends on how you define ridiculous, but HFTs actually make a surprisingly small amount of money these days. Probably single digit billions across the entire industry (https://quant.stackexchange.com/questions/34856/how-much-pro...), though 2020 was an exceptional year for many firms.

> Shouldn't the exchanges have offered the best price, and provided liquidity, in the first place?

Either your wording is a little funny, or this question indicates great ignorance about how trading works. Exchanges do not provide liquidity, market makers do. In 2021, "HFT" ~= "market maker".

I don't know that much about this, so you could chalk it up to "great ignorance", I suppose.

I'm a software engineer, interested in crypto, and not that involved in traditional markets (except for holding an S&P 500 index fund).

I do think the exchanges in traditional finance shouldn't have required HFTs in the first place (i.e. it's an antiquated technology). I also think hedge funds and the ultra rich have privileged info, that retail investors don't have.

Anyway, I appreciate the clarification. I like learning about all this.

Today's crypto has the same issues: the fastest decision-making computer with the shortest time to the network makes the block.

Just like cutting inches off mainframe cables, manufacturers of ASICs purpose-build computer chips for crypto. GPU operators tune their cards and even download new code to get every last bit of processing capability from their devices. Crypto in these ways is just like HFT.

I still like the distribution of power to those who wouldn't have had a chance otherwise.

Alameda and Cumberland?

I meant that the flat system of crypto where (at least before industrial mining), people could spin up mining rigs (which is still possible, I think at least with some altcoins), but also, that the barrier to entry for folks trading crypto is lower than traditional markets. Some may chalk this up to the need for KYC/AML, but I think there are intentional barriers to entry for individuals that people in power don't want to have access to say, equity markets, that don't exist as much in crypto markets. I don't think crypto markets should be used to supplement terrorism, money laundering, tax avoidance, or anything else like that, but if you take a look at say, Robinhood, who tried to lower some of these barriers to entry to traditional markets, I think crypto as a whole is trying to do that. I also have said in other posts, and will continue to, that blockchain analytics will only continue to get more advanced, so the narrative that crypto is only used for illegimate activity is bullshit. I do think that it is a power to the people that can't be stopped, due to its decentralized nature, and that it is the only possibility I've seen so far as an alternative to everything else going on that requires sovereign government policy and currencies (that are backed by war, control, etc).

Isn't that the opposite of how it works?

Transferring data and calculating the core of a new block takes a fraction of a second, and then it takes an average of several minutes to find the right random numbers to finish the block.

A latency advantage in HFT lets you take most of the profits. A latency advantage in cryptocurrency mining gives you a fraction of a percent better profits.

A calculations-per-second advantage helps in mining, but it's strictly proportional.

> Isn't that the opposite of how it works?

No.

> Transferring data and calculating the core of a new block takes a fraction of a second, and then it takes an average of several minutes to find the right random numbers to finish the block.

A PoW, say Bitcoin, is configured in the consensus algorithm to avoid duplicate spending. The fastest, correct miner for a block that communicates the quickest to the network will get the block.

I think you're very confused about how mining blocks works. It only takes microseconds to test a specific hash. Every miner is testing a different random bunch of numbers to see if they get a lucky result where the hash starts with enough 0 bits.

There's near-zero overlap between the numbers tested by different miners. So a latency advantage does not make you win. Everyone is picking lottery numbers in parallel. If three miners have the exact same hash rate, and one of them has a 2 second head start, then the one with the head start is only going to win 33.4% of the time.

You're making some incorrect assumptions about what I wrote, based upon somehow reading things I didn't write. I am intimately familiar with the bitcoin network. I never said that many miners simultaneously work on the same block. Don't worry, you aren't alone in the crypto world with making assumptions.

"The fastest, correct miner for a block that communicates the quickest to the network will get the block." is misleading at best and completely false at worst.

If I'm misinterpreting you, feel free to clarify, but to me it looks like you're saying that latency matters (more than a fraction of a percent) with cryptocurrency, and it does not matter.

And moreso, "the fastest decision-making computer with the shortest time to the network makes the block." is not true. Even if you mean "lucky" by "fastest", since the computer that finds a block doesn't have to be fast, 99+% of the time blocks are found far enough apart that time to network is irrelevant.

> I'm a software engineer, interested in crypto, and not that involved in traditional markets

As you learn more about crypto and traditional finance, it'll be fun to compare the two. Your confusion about the role of an exchange in traditional finance might be because you see the crypto world, where a single entity often performs the roles that many entities perform in traditional finance (exchange, clearing firm, broker, etc.).

> I do think the exchanges in traditional finance shouldn't have required HFTs in the first place (i.e. it's an antiquated technology)

I'm not quite sure what this means, but it's important to understand that HFTs exist in the crypto space as well. Capital markets don't function particularly well without marker makers, and absent some rule explicitly preventing high speed trading, marker makers will tend towards being the fastest traders in any market.

> As you learn more about crypto and traditional finance, it'll be fun to compare the two. Your confusion about the role of an exchange in traditional finance might be because you see the crypto world, where a single entity often performs the roles that many entities perform in traditional finance (exchange, clearing firm, broker, etc.).

Yes, I agree!

In terms of what I called antiquated technology, I think there are a lot of layers on traditional finance, and a lot has changed since its beginnings. I think crypto will go through a similar evolution, in terms of tech, regulation, etc. I think we're in the very early stages for crypto and it has a chance to be an even better system.

I do know that HFTs exist in crypto, and it still is the wild west in some ways, but in the end I like that innovation is happening and that there are alternatives to existing systems.

That said, I appreciate all the responses and I'll take some time to learn more about traditional markets.

Crypto volume is pretty hft/market maker weighted as well.

There’s just not

a. Enough random interested parties willing to buy/sell various coins so that you have low-spread and liquid markets

b. Non-hft players who keep crypto markets in line with each other

This isn’t super surprising. Managing posted liquidity is a difficult task that sort of requires being halfway to a market maker, and naturally most non-market makers just want to buy and sell right away instead of posting orders and waiting/hoping.

The result of this is that most liquidity is provided by HFTs, since they’re the only party that can and even wants to have a bunch of bids/offers out for you to trade against.

I'd like to recommend the book "trading at the speed of light". It's an in depth look at the world of high frequency trading based on first hand interviews with traders, court documents, etc. It would appear based on what little information is actually available that HFT firms net only modest profits after taxes and fees on trades. Historically, it appears most firms fail to turn a profit and shut down. A paper of Laughlin (2014) suggests virtu, a hft firm that now executes Robinhood orders, earns an estimate profit of 0.24 cents per trade, with 0.05 to 0.1 cents per share traded being a respectable profit.

I highly recommend this book though if your into this topic

Thanks. I'm going to check this book out.

Reply on news.ycombinator.com