[mananaysiempre]

You might even say forcing the prices of things to be accurate and exposing departures from reality is the purpose of financial markets and the thing an investor gets paid for.

(How well that works is questionable; my completely amateur impression is that this view would force us to legalize insider trading while declaring HFT completely useless, for example.)

[IMTDb]

> declaring HFT completely useless, for example.

Please do ! The amount of resources invested in making HFT possible and profitable is no where close to being justified by the benefits it brings to society as a whole.

Making financial information move in days instead of week ? Sure !. In hours instead of days ? Yep !. In minutes instead of hours ? Why Not !. In seconds instead of minutes ? Mmmh yeah, I guess. In milliseconds instead of seconds ? Stop it. Opening a road to put high quality cables and hiring some of the most clever folks on the planet to cut half a milisec on a buy/sell algo ? Fuck you.

[bidirectional]

Are they? Spreads have tightened massively since the advent of HFT which has knock-on effects for the small investor both in terms of cheaper personal trading and lower costs for those managing their retirement/mutual funds (see Vanguard's opinion on HFT[1]).

In terms of resources, Virtu is publicly traded and has a market cap of 6.3bn, Citadel was recently valued at 22bn, the total sector is maybe 100bn. Meanwhile our society ascribes about 260bn of value to Coca Cola alone.

[1] https://www.ft.com/content/ff8c6486-cb37-11e3-ba95-00144feab...

[gregw2]

Spreads may have tightened, but payments for order flow suggest to me there's a new flavor of front running involved via HFT and dark pools. But I'm no expert.

[bidirectional]

How? No offence but it just sounds like conspiratorial nonsense based on some scary sounding terms.

Payment for order flow is an example of the incentives of retail traders, retail brokers and HFT market makers aligning. The market makers are able (and obligated) to improve prices because they have paid to receive flow they are almost certain is not toxic. Meanwhile they charge the likes of hedge funds higher prices as they can't guarantee the reasoning behind their trades (i.e. they may be about to have their faces ripped off).

The HFT firm takes their (tightened) fee for making a market, kicks some back to the retail broker, and because the broker earns money that way, they offer commission-free trading to the retail trader.

[noodles_nomore]

The book Flash Boys deals with this. It's a long time since I read it, but IIRC brokers need to split large orders and route them to different exchanges to get the best price (which is required under 'best-effort execution'). These orders will hit the exchanges at different times giving enough time for HFT to observe them at one exchange and front-run them at the other.

[unmole]

> The book Flash Boys

Is a hopelessly bad book. It's a thinly disguised advertisement for IEX.

> front-run

I do not think it means what you think it means.

[sonotathrowaway]

Most large orders won’t be retail traders though.

[unmole]

> No offence but it just sounds like conspiratorial nonsense based on some scary sounding terms.

It's exactly that.

[quickthrowman]

Are you worried about getting front run for fractions of a cent? I’m not. I’d rather have penny wide spreads instead of 1/8 dollar (or larger) like the old minimum tick size.

Dark pools reduce volatility, it’s just liquidity outside of the normal order book. Regular people don’t need dark pools because they aren’t slinging $500M block trades. Furthermore, trades can happen off-exchange without dark pools.

It sounds like you aren’t even an amateur, let alone an expert, no offense.

[unmole]

> but payments for order flow suggest to me there's a new flavor of front running involved via HFT and dark pools.

I know what those words mean individually. But put together like that they make no sense.

[bradknowles]

Naw, cable is slow. Even if it's optical. Still slower than direct microwave connection via line-of-sight point-to-point systems.

Anyone doing fiber is doing the wrong technology, if they want to minimize total end-to-end latency.

[simonh]

It forces prices to be what people believe to be accurate, where that belief is strong enough for them to put down money on it. I think that may be the closest we're ever going to get.

The bit about belief above is why insider trading is illegal. If people can't trust the information they have, and at least importantly that they have fair access to accurate information, then they won't have faith in markets and won't put down their money. Belief implies trust.

[pis0m0jad0]

I have bad news for you, insider trading isn't about fairness. People trade, legally, on inside information all the time. Adverse selection abounds in markets, so you have to be careful. Insider trading laws are rather about theft: did you illegally appropriate information owned by someone else to benefit yourself and/or a third party? That is insider trading. In the US, it can be even more relaxed than that, with the standard being you had intent.

[lupire]

> People trade, legally, on inside information all the time.

Example?

[simonh]

It may be a contrived definition of ‘insider information’ in an attempt to concoct a controversy. Like a company buying it’s own shares based on its internal knowledge of its condition, or employees buying share options in a scheme for their company. If it was actually an issue I expect it would have been stated instead of coyly implied. Its true not all market participants have access to the same information all the time. The point I made is that this can’t be so skewed or biased that it’s an obstacle to investment.

[daniel-cussen]

This is true.

Speculators play a vital role in stabilizing the price of a commodity over time. All other forms of arbitrage are meaningless besides the biggest one of them all: temporal arbitrage. Buy low sell high. And with it, sell high buy low.

Government employees, for instance for Codelco, Chile's state-owned mining corporation, value speculators because those speculations are the best predictions of the future. Suddenly you can make guarantees to your stakeholders, and sell futures, and plan, feed the prediction into your other math.

Also I'm not sure HFT is totally useless. If you set a minimum latency in the market and forbade anybody go around that you would get cheaters, and less integrity, and roll back trades. Instead, with HFT, the only challenge of the fastest is to really be the fastest. They gotta win. Further, at a societal level, I divine there are harmful macro artifacts that happen at large multiples of HFT's granularity. So when the latency was milliseconds, artifacts happened in years, in microseconds, they happened in days.

[hakfoo]

It feels like a theory/practice problem.

If there were zero external costs to having a minimum-latency market, no big deal.

But we're still short of that. So there's a lot of money and brainpower being thrown into marginal decreases in market latency. Do we want so many of our best engineers chasing HFT performance?

It also creates a new ecosystem of arbitrage-- the guy with 20ms ping can take advantage of price differences the guy with 40ms can't-- which isn't necessarily producing real value.

This also assumes that the goal of absolute price discovery is itself meritorious. Maybe we're better off not being able to correctly value assets with infinite precision, because it tends to encourage the mindset of chopping up businesses for parts.

I'm a fan of the old-line conglomerate model. By bundling so many diverse assets, they inherently block price discovery, which creates openings where risk hedges and long term plays can survive and pay out. But investors hate being unable to pull those components out of the stock price.

[daniel-cussen]

I don't know that they actually are that competitive about marginal decreases in market latency. I think they're competitive for some stuff, but for others, it's easier to split the atom than split their eyelids. Literally.

Like I came up with a sorting algorithm I thought HFT--just some single monopolistic HFT company--would love, like die for. And I kept it so secret, did so much so nobody could hack it under any circumstances, never put it in an email, nothing, and? and? I sent it to over 40 FAANG and HFT equivalents and never heard back.

They really don't care about time, like really in real life. After the fact, yeah. Like when you're explaining to them why they lost money I guess. Not beforehand, when they could make money.

You're projecting about how competitive it actually is. Maybe if I actually split the eyelids first, then it would be competitive, but without that it is it is about as competitive as auctions for the best uranium mineral rights in the 1920's. Nobody gives a shit!