Why Do High-Frequency Traders Cancel So Many Orders?
bloombergview.com
bloombergview.com
Most work on a maker taker model. Which means the trader who initiates the trade pays a small fee and the trader who is the passive side, the one who had their order in the market already, gets paid a small fee. as a side note there are inverted markets but lets leave those aside for now.
This means to get paid you want to be at the top of the book, which means you are the first order to get filled when someone crosses the spread to get their order filled. the way priority is determined is first by price and second by time. So you have a very vested interest in being the first to cancel and move your order to the newest price level.
Exchanges have tried introducing some order types to alleviate this constant send/cancel routine such as the order type "Hide not Slide" but people tend to get upset at these order types.
Once you understand this, you start to realize that almost all HFT firms aren't quote stuffing, they are just jockeying for position at the top of the order book.
I've never really understood quote stuffing, the same firm that quote stuffs still has to deal with those quotes coming back, its not like the market data has a flag saying ignore this quote change as its caused by your own quote stuffing.
The way most markets are setup is that quotes come from gateways and multiple symbols all share a single gateway, usually assigned alphabetically, so A-F tickers all share the same gateway. This means that if someone is actually slowing down market data for say AAPL then they are also slowing down quotes for AMZN as well but again, the same firm that is quote stuffing also has to deal with their own mess so I can't see the benefit.
Another comment complains that HFT firms don't like the fragmented market. That is true to a point, but keep in mind most HFT strategies only work due to the fragmented markets and RegNMS. So while they may not like 11 venues, they certainly want atleast 3 or 4, and many of hte top HFT firms run their own dark pools, adding to the problem:)
As far as Hillary Clinton introducing legislation to curb HFT trading, she was the senator for New York. I'm dubious of her coming down hard on Wall Street.
Quote feeds don't work that simply. you still have the following issues:
1) Your quotes are slowed down as the gateway is spammed so legitimate quotes are delayed to you as well as everyone else. So you are still blind as to where the market is just like everyone else.
2) The quote feed doesn't just say, "hey a new order was added and guess what, its yours!!"
You still need to parse the entire message to determine if the order matched one you sent, and even then it won't say its yours, it could be someone else putting in the same 100 share order at the NBBO. I'll admit this can be helped if you have your own number at the exchange to identify the order sender, but then this only applies to exchanges that release that information and even then you still need to parse the order to find out if its yours, so you're basically back to square one.
Now you might say, well then I'll just submit orders far away from the market so that I can more easily identify them as my own, but by doing so you've already outed your self as a quote stuffer and its game over.
It's one thing to rapidly CFO your orders to keep up with the NBBO and changes on other markets, its another to rapidly CFO orders out in the weeds. The latter will get your direct access yanked if abused.
So ignoring your own quote spamming is almost impossible as you still need to assume every order might not be your own.
FPGA's don't help at all here, except to make the parsing faster but they do that regardless of if someone is quote stuffing or not.
Time from message sent to seeing back on marketdata feed can be a few ms, or it can be faster. Depends on exchange and traffic. Few ms didn't use to be totally uncommon on CME, but it may be closer to 1ms in the 95th percentile these days, for sure.
Politicians say they're protecting the innocent, but really they're just tipping the scales for whoever is backing them. This is how they do.
GS can already bend the FBI and NY State to their will: https://en.wikipedia.org/wiki/Sergey_Aleynikov
Might be futile effort though since the DNC isn't even inviting him to the debates despite his having raised 1M in a single month.
ISTM this would only work in one direction? That is, if you're moving closer to the other side, you can just put in a new order. Cancelling the old one isn't such a priority because no one would pass up the more enticing new price to get to the old price. If they did, free money for you!
If this phenomenon isn't clear from the data, that would argue against the hypothesis that maker-taker accounts for the cancellations.
IANAHFTrader.
Also most of the cost is fixed. If you've already built a super-fast cancelling system that can cancel an order in 8 microseconds, there's no value in running a slower cancelling system in parallel for your less important trades, you'd just run all your cancels through the same system.
Why should one symbol be tied to a completely unrelated symbol? This doesn't engender confidence in the architecture of these markets (which markets are you referring to, specifically)?
>you start to realize that almost all HFT firms aren't quote stuffing, they are just jockeying for position at the top of the order book.
I'm not sure this is an important distinction. What is quote stuffing but position jockeying?
The article uses the term "front-running" incorrectly. Front-running is where a firm places their own trades ahead of trades they're placing for a client, to capitalize on the price movement that client order might generate. This is illegal.
What the market makers in the article are doing isn't front-running. It's just being smart with their orders.
And that's generally why HFTs cancel orders- they're reacting to market conditions that exist on the span of microseconds and will want to change their market positions very quickly- including canceling orders that they no longer think are suitable.
He's technically "not lying," but many people won't notice the footnote nor appreciate the difference. All they will hear is "front-running == should be illegal"
Probably still worth pointing out, since one of the activities is illegal and harmful (uses non-public information) and the other is just reacting quickly to the public market information.
Liberately quoted:
'[...] man, remember when "front-running" meant something? [...] But then came "Flash Boys," [...] And now, basically any time anyone trades on public information before someone else, it's "front-running," [...]'
Market conditions of real money players don't change in micro-seconds. For some reason this flapping of HFT strategies is a Nash equilibrium among the HFTs.
If a value trader goes and does a lot of good research on a company and then executes smart trades based upon that research they will make a profit. All well and good.
If a value trader goes and does a lot of good research on a company and executes some smart trades based upon that and their broker front runs them, a substantial portion of their profits are handed over to the broker. Not good. Value trader may not bother doing all that research in future (market for lemons; equity market becomes ever more disconnected from the real world). Broker parasitically extracted the value of value trader's real world research from them.
If a value trader goes and does a lot of good research on a company and executes some smart trades based upon that and an HFT detects the trade they're putting through and trades ahead of them using their superior speed, a substantial portion of their profits are, likewise, handed over to the HFT. Not good. Value trade probably won't bother doing all that research in future (market for lemons).
Front running is both a principal/agent problem and a market for lemons problem.
I describe the process in a bit more detail here: https://www.chrisstucchio.com/blog/2014/fervent_defense_of_f...
This enables market makers to price discriminate between Joe 401k (who pays less) and David Einhorn (who pays more).
tl;dr; Joe 401k should pay more for his retirement savings because by indexing, he's freeloading on the valuable labor provided by prop traders. HFT is bad because it makes this freeloading cheaper.
(Weird to hear argument lionizing hedge fund managers and calling out workers as freeloaders coming from left wing types.)
Others often (rightfully) feel that HFTs who engage in latency arbitrage where they take advantage that everyone else is using the NBBO (because they have to), and the NBBO is lagged, are front-running assholes who extract value without creating anything.
And we refer to that thieving, predatory, value-stealing activity as 'front-running', even though technically you're engaging in a slightly different activity.
Now, possibly rightly, market makers in general, through the ages, have had a bad rap. They are indeed trying to get more information than the average joe, with very clever, and risky, techniques (see below) and skimming him after having done so. Nevertheless, you must remember, that without these people/machines taking these risks, you would not have a continuous market in which to trade. You'd have a much more stepwise price action and much more risk. They're providing s service.
Perhaps most controversially, being a good market maker means having some capital, so that you can wear a loss which is entirely possible during your price discovery. Thus, market makers who make money, inevitably already have money. This doesn't help their cause.
But the idea that HFT per se is the problem is wrong. If you don't like HFT, you don't like finance, period. That may be a legitimate view, or not, but the two are inextricable. They are not different one from the other - HFT is simply Amazon doing what Barnes and Noble does, more efficiently (without the monopoly aspects - HFT is fiercely competitive).
Without HFT, bid offers would be wider. Fact.
For the incredibly small minority of people who can engage in it, and who enjoy special rules, maybe. For the majority of the people who's money is actually extracted by this system, it's an exclusive club.
> Without HFT, bid offers would be wider. Fact.
The majority of people who just want to save for retirement would prefer wider bid offers instead of having such a large chunk of money extracted from their future bank accounts. Fact.
The other reason why HFT is non-competitive is that you cannot go and start a market with your own rules without being deeply in bed with the government and the financial status quo. HFT is forced down our throats by a system that calls itself capitalistic but thrives on enjoying custom-made loopholes in heavily-regulated statism.
Making money making markets takes risk capital. i.e. Money. The money makes more money. Fact. I agree. But don't blame HFT. Blame finance. That is how finance works. I see a completely legitimate case for being anti-finance. I don't see a legitimate case for being anti-HFT only. Indeed, the opposite, if HFT reduces the bid/offer paid by the average retail investor. Which it does. Fact.
Do you know who hates HFT even more than the general public? Human market makers. I think that says more than any of my arguments.
The point is that it is not the end users who are getting hurt. It's the old monopoly - the human market makers.
DISCLAIMER: I (was) a HUMAN market maker.
The difference, from what I can tell, between the HFT "elite" and the human market-maker "elite" is that the human elite actively colluded to retain their status. Compare the largest HFT firms to the largest investment bank, and the number of entrances and exits in the market for electronic trading firs.
HFT have no such obligation, and so, when there are rapid shifts in prices, may just wait out the chaos.
HFT gets the benefit of taking the spread, without having to pay the cost of ensuring orderly markets.
http://nysearcarules.nyse.com/pcx/pcxe/pcxe-rules/chp_1_1/ch...
From the document that you listed, it looks like NYSE Arca has a requirement for 100% continuous quoting, but the quotes can be 8% away from the current market price (section 7.23.a.1). This is basically a free pass; nobody wants to trade against a quote that wide. For reference, take the most liquid ETF: SPY trades above $150 and regularly has a $0.02 spread, one THOUSAND times tighter than the 8% requirement.
On other exchanges, there are requirements for tight quotes, but they usually come with relaxed requirements on quoted time. For instance, maybe a market maker could be required to quote "90% of the time within a 0.5% spread". In such a scenario, allowing market makers to pull quotes for 10% of the day is basically giving them a free pass on the most volatile points of a day.
I have seen very few situations where registered/designated Market Makers are obligated to suicide themselves to provide liquidity; there's usually an "out". In practice, the top-tier HFT market makers (de-facto) are already exceeding the obligations required of Market Makers (registered).
http://libertystreeteconomics.newyorkfed.org/2015/10/the-liq...
It doesn't help the conversation to start using the name of a crime to describe something that's legal but you don't like (even if you think it should be illegal, but agree it doesn't fit the legal definition of the named crime).
I don't like that back when rape and pillage on the high seas was a large threat, some copyright holders were able to convince people to start calling copyright infringement "piracy". I don't like our current tendency to over-label things as terrorism or exaggerate the role of narcotics smuggling in financing Islamic terrorism. I also don't like labeling reacting quickly to public information and reacting to expected orders as front-running. I think it's a cheap trick that harms the quality of the dialogue.
Ultimately investing runs on trust. HFT is consuming public trust in the financial system at a prodigious rate. Is it a trillion dollars a year? A billion? Hard to be sure. But it certainly isn't clear the tiny market-making improvements are worth it.
It's not clear.
Yes it has, but why would you consider that a good thing? CAFTA and NAFTA have been pretty horrendous for the finances of Americans and TTIP and TPP are likely to be even worse.
They can't be killed unless enough people understand them, either.
Whereas, with HFT...
Edit: As Harryh said, large mutual fund managers believe that HFT allows them to get better prices for their customers: http://www.cnbc.com/2014/04/25/vanguard-chief-defends-high-f...
Ballpark esitmate: drop $20k/year into retirement (1 lot of SPY/year) x 1 penny/share being robbed from you x 50 year working career, you've lost $50 to the evil HFTs.
Ballpark figure, of course.
https://finance.yahoo.com/q/is?s=AAPL+Income+Statement&annua...
Which is probably more relevant than AAPL's net profit.
Anyhow, HFT affects execution, 'electronic trading' affects how you put in orders, both make the process of buying/selling securities cheaper.
Such markets are created specifically for the purpose of avoiding HFT: https://en.wikipedia.org/wiki/Dark_liquidity
HFT hasn't been around that long, either. Before 2005 it basically didn't exist.
Dark pools are beneficial to institutions in some circumstances, not to individuals.
And yes HFT hasn't always existed, once upon a time you'd have a pit of screaming traders and brokers, and for an individual to buy/sell stocks you'd have to call your broker on the phone, who'd charge you an obscene amount for the privilege.
Anyhow, HFT is more or less just a euphemism for high-speed arbitrage/market making.
I look at the execution I get on my trades today, I couldn't imagine not having that service available.
Yes, to hide them from HFT.
Read on a little.
I'll tell you why they use dark pools. On the open market, if you sell lots of shares, buyers will see that, and drop their bids. Likewise if you put in a large bid, sellers will raise their asks. In a dark pool, institutions can move large blocks at a given price without market forces interfering, and reduce trading costs.
It's like the difference between searching for a house on a public database, or a broker's private database...
What point are you trying to make about them? That if we didn't have HFT, we wouldn't need them? That's an argument for HFT, not against.
>That if we didn't have HFT, we wouldn't need them? That's an argument for HFT, not against.
Dark pools are an evolved defense against parasitic market players that is not cheap. These costs are then passed on to you via your pension fund, index fund or if you buy insurance.
I don't see how that's an argument for them unless you or your friends were personally profiting from it.
You and I buy stocks in small amounts. Intuitively, after each of our trades, we understand that the market moves to take the impact of those trades into account. That's the point of markets!
So why on earth should it be that a hedge fund should be able to buy or sell huge amounts of stock without having the market move? You don't have that power. Why do they? The fact that a giant entity is trying to move huge amounts of stock is information. The point of the market is to capture that information and build it into prices. That's exactly what HFTs are doing in this scenario.
The expectation that hedge fund managers have that they should be able to capture the spot price of a security and then buy or sell arbitrary amounts at that price, taking their information advantage out of the hides of every other market participant, seems totally unfair. Again: you don't have that privilege on the market. Why do they?
That great innovation where you no longer have to shout in a pit to get your trade executed because, you know, computers.
Virtually all trading these days is electronic.
>And HFT is
Algorithmic trading by computers that's "fast" (usually regarded as sub-second trading, although definitions vary a little).
The more you know...
For instance: Google [odd eighths scandal].
And that's a modern example of humans rigging the markets, exploiting lack of competition and automation. Things get much worse the further back you go in time.
Humans are of course perfectly capable of rigging the market with or without computers, a fact so mind-blowingly obvious, I'm not sure why you'd accuse someone of believing that it's false.
This particular thread was not about market rigging, it was about whether HFT can really be credited with decreasing spreads in the early 00s.
You have not described a difference between two things that you are trying to distinguish between.
That's not true, not like HFT anyway.
First, you can quickly and succinctly describe the traits and benefits of a smartphone. You can be high-level at first, you don't have to explain how every detail works. There's no question, for example, that a touch screen doesn't actually work as advertised. You can just say that a smartphone is a cellular/wireless computer device with a usefully large touch-screen display.
In contrast, the unknown question with High-Frequency Trading is: is it actually providing value to anyone but the traders themselves? They might say they are "market-making" but are they really market-making (a trading role with proven market value) or are they merely exploiting structural inefficiencies in the trading infrastructure that mostly hurt everybody else? Maybe they are, but they should be able to describe it in a high level terms first (market-making, liquidity, etc.) where their role provides an obvious benefit to the market, and if you don't understand the high-level terms (eg liquidity) you can look them up.
So far as I can tell (and I am willing to be proved wrong here) there is no consensus answer to the question of whether HFT is actually valuable to markets.
Replacing slow expensive humans with fast and cheap computers has dramatically reduced the cost of trading. You can see this because buy/sell spreads have shrunk by at least 10x.
High-Frequency Trading, while also an umbrella term, virtually always refers to a subset of algorithmic trading involving arbitrage over extremely short timeframes. HFT is not about being faster than "slow expensive humans" it's about being microseconds faster than other HFTs.
http://blogmaverick.com/2014/04/03/the-idiots-guide-to-high-...
In my experience though, electronic market making, which I regard as a very good thing, is a direct subset of HFT. To do it properly you must be fully automated, fast, across venue and trade alot. By nearly every definition I've seen that makes you HFT.
Now, HFTs are competing with other HFTs, but initially, they were competing with "slow expensive humans". The current situation only tells you how far we've progressed.
The thing to remember about the markets is that each individual trade is always zero sum, but the value of the markets comes from the aggregate total.
The behaviors that we want in our markets, price discovery, liquidity, easy risk management are all outcomes that are enabled by speculative market participants like market makers engaging in lots of zero sum activity.
So instead of decrying the zero sum activity what we want to do is drive down the price of it to the non zero sum participants. And HFT market making has been prodigiously good at that.
HFT trading is zero sum because the traders don't actually want or keep stock.
Are they zero sum?
Aside from that, how does your statement relate to inventory risk? Market makers and grocery stores take on inventory risk. An absence of market makers in BRK.A* would only show that nobody wants to take on that risk. It does not change the role of the exchange. Exchanges facilitate matches, they are not a counterparty.
* which is not actually true, see http://batstrading.com/bzx/book/BRK.A/ during market hours and you'll find active quotes at a 2% spread.
The difference is all about timing. I may want something else more than you do but am willing to sell now. If at the time you close out your trade (that is sell the shares from me) the price may have risen or fallen. If it rose you won and I lost by not holding longer.
This time mitigation is precisely what market makers have always done and what HFT market makers have driven the profits (and thus the costs to outside participants) out of.
Even stock to stock transitions can be meaningful as Bill Gates had a lot of MS stock and wanted a hedge so he sold stock. What he got was probably worth 'less' the diversification was valuable to him making the transaction a net positive.
That the markets provide those behaviors is what makes them valuable but the actual trades that make up those aggregates, your selling of shares when you need a car to someone else is zero sum. Either you would make more by holding or you wouldn't.
That something other than that is more important to you indicates that you are in the market for a middle man to bridge that time gap and buy some of the risk from you. Your time horizon is from share purchase to "need money for car", not from share purchase to "optimal selling point". The service you take advantage of when you bridge that gap is provided by the aggregate work of many zero sum interactions between speculative participants like market makers and "investors" like your self.
If I put a sell order on the market at 12:00 the only impact is the sales price. If it executes at 1PM or 2PM it makes zero difference to me as I can only access money at the end of the day. So, I only gain liquidity if I would have been otherwise unable to sell by the end of the day. Therefore, I don't gain liquidity from HFT.
https://www.chrisstucchio.com/blog/2014/how_to_not_get_rippe...
IE: People want actual stock. The stock has innate value due to the potential for dividends or to influence the future of a company.
That is a significant amount of value.
None of which we get from HFT.
HFT cannibalizes the research done by value investors (oh, but it's not legally front running if they're not your customer!), rendering that a market for lemons, so there goes price discovery.
HFT floods the market with more liquidity than it needs during normal periods (there is no benefit to you being able to trade at split second intervals. nada. none) and then extracts it all during periods of market distress when liquidity would acually be useful.
And risk management? Please. They're only managing their own risks.
No but we do get it from market makers. HFT has led to a dramatic decrease in the price of market making. Unless your claim is that market making is not something that should be allowed?
> And risk management? Please. They're only managing their own risks.
My point about risk management was about the aggregate benefits of the markets, not about HFT providing someone risk management. If you have some risk that you want to sell, there will be a buyer for it because of the aggregate sum total of all the zero sum transactions available in the markets.
No, my claim is that market making was made significantly cheaper by electronic trading in the 90s-00s but HFT had very little effect on that.
HFTs do a lot of market making, make almost no profit from it and mainly use it as cover for their more nefarious activities.
>My point about risk management was about the aggregate benefits of the markets, not about HFT providing someone risk management
Great. So even you agree that if we banned HFT we'd be no worse off.
Levine is explaining how you can get a 95+% cancellation rate simply by running the most brain-dead simple possible market maker strategy: because you're required to post orders at multiple exchanges, and because every price change involves order cancellations (potentially lots of order cancellations, even on a single exchange, because of pairs trading and price ladders), and because adjusting prices on exchanges in near-real-time is the basic job of a market maker, virtually anyone running an electronic market maker is going to have a huge cancellation rate.
Levine brings this up to illustrate the silliness of proposals to regulate HFT by targeting entities with huge cancellation rates.
Comes now 'cdroconnor. You're playing a semantic game. You're defining "HFT" as "bad HFT", and everything else as simple "electronic trading". FINE. Nobody disagrees with you, except on the very boring point of what labels to attach to things.
But your argument here doesn't make any sense for the thread, because the good simple electronic trading you're condoning is also targeted by the cancellation regulation Clinton proposed. Which is the whole point of the article.
The discussion went off course way before I dived in.
>Comes now 'cdroconnor. You're playing a semantic game. You're defining "HFT" as "bad HFT", and everything else as simple "electronic trading". FINE. Nobody disagrees with you, except on the very boring point of what labels to attach to things.
There is a very substantial non-semantic difference between robot-executed sub-millisecond trades (HFT) and trades which are are just executed electronically.
In every discussion about HFT the probability of someone falsely attributing the decreased transaction costs of "not shouting in a pit" to algorithms that execute sub-millisecond transactions approaches 1.
>But your argument here doesn't make any sense for the thread, because the good simple electronic trading you're condoning is also targeted by the cancellation regulation Clinton proposed
I'm no particular fan of that either. I'd prefer Italian style micro-transaction tax. That wipes out nearly all of the sub-millisecond trading and leaves the rest intact, including market making.
What you haven't made clear is why you believe you're actually arguing with anyone here. I am 100% certain, because I've had the conversation with him multiple times, that 'kasey_junk agrees with you that there is such a thing as malignant electronic trading.
Exactly what is the controversy here? The people who are talking about HFT reducing spreads are talking about benign electronic trading, and none of them appear to be denying that there are other kinds of electronic trading.
I was arguing that sub-second algorithmic trading cannot be credited with substantially reducing spreads in the early 00s.
kasey_junk linked to an article that claimed that.
It's a defense of HFT that's rolled out so often that it's practically become a cliche.
How? It seems like this should not be that hard to explain.
http://www.bloombergview.com/articles/2014-03-31/michael-lew....
Quote from the posted article:
" it should be said that market makers have existed in the stock market for a long time and that electronic market makers do the job waaaaaaaaaay cheaper than their human predecessors."
Microsoft Word is phenomenally heaper than hiring a typewriterist, because it is electronic.
That said, when I was in the industry the common usage of the terms would be that electronic market makers were a subset of HFT.
That is there are HFT strategies that are not market makers, but there are no electronic market makers who are not HFT.
Decreased spreads; trading is much cheaper now than it was before HFT.
By how much? And is HFT the cause or is it merely correlated?
In any case, compete they have, pushing their profits pretty close to zero. Look at the graphs in http://www.zerohedge.com/news/2013-02-13/how-getco-went-hft-... (article is terrible, it was just the easiest place to find the graphs).
That's simply not true; your premise is flawed.
It does not matter if trading is/isn't zero sum. Or perhaps you can explain why it matters?
Almost every attempt to declare something zero sum will miss out relations, factors, dependencies and links because the model they pick is too simple. Hence the never-ending arguments about trying to define something ZS or not.
But I'd add, HFT is like kids playing soccer on the freeway. They don't create value for anybody but themselves, and they screw up things for everybody else. I wish they would go away.
In fact, the fragmented market results in huge costs (4 datacenters worth of infrastructure, low latency connectivity, etc) In reality, most market makers would vastly prefer to simplify this away. This is one of the reasons many traders have moved to alternative markets that have fewer trading venues (for example, many futures and options trade primarily on a single venue)
[0] http://www.bloombergview.com/articles/2014-03-31/michael-lew...
In this current, though, we have three major operators with 2-3 exchanges each - many with single digit percentages of market share. I suspect that the savings in execution cost due to competition are vastly overwhelmed by the increased cost of infrastructure for most market participants (excluding the largest firms).
Arbitration between physical locations is something that can't be helped. The other things ways that high frequency traders make money can be helped by better systems, but there are no incentives to make those systems when exchanges make so much of their money from high frequency traders themselves.
This is a very strong statement with little support. I agree that some competition among exchange operators is important, but how do you justify exchanges like CHX, with approximately 1% market share?
I am not sure how to respond to your comment regarding front running without more detail. Many sources have already debunked the Michael-Lewis-style argument regarding front running. Where do you see front running? (Using the proper definition of trading ahead of a customer order based on knowledge of that order)
What needs to be justified? It is competition and decentralization.
The fact that there are multiple issues with high frequency trading is one of the reasons it ends up being so polarizing. It ends up being an ambiguous term that sometimes means something that is a natural result of decentralization, and sometime means techniques that would be grating to most people's common sense of what should be legal.
"Americans love commuting. If not, they wouldn't be spending millions of hours in their cars each year to get to work!
Front-running is profitable against traditional orders entered by humans. But with spoofers in the mix, the picture looks quite different: When the front-running HFT algorithm jumps ahead of a spoof order, the front-runner gets fooled and loses money. The HFT’s front-running algorithm can't easily distinguish between legitimate orders and spoofs. Suddenly the front-runner faces real market risk and makes the rational choice to do less front-running. In short, spoofing poses the risk of making front-running unprofitable. Because spoofing is only profitable if front-running exists, allowing both would ensure that neither is widespread.
http://www.bloombergview.com/articles/2015-01-23/high-freque...
I am not saying payment-for-order-flow doesn't exist, but the buyers are firms like Citadel and other "internalizers", not exchanges.
If by "paying" you are referring to the maker/taker rebate model, that is paid to any market participant, not just retail brokers.
When you do the wrong over several times without getting caught it becomes a standard (read: make-or-take rebates)
This abuse of the term completely confuses the entire debate.
As far as the tax: personally I'm very in favor of slowing down trading... unfortunately, what's being proposed introduces as much structural game theory as it eliminates. What's the dominant strategy for a trader faced with a rule like "any trades resting for less than 300 micros get taxed"? To be as fast as possible without triggering the tax; staying just above the threshold.
All that economic waste on low-latency tech will just be redirected to low-jitter tech. Those who can reliably land an order within +1 microsecond of the tax line will outperform those who only have an accuracy of +10.
I'd like to see more exchanges experimenting with things like random variance in speed bump size, frequent batch auctions, etc.
http://qed.econ.queensu.ca/pub/faculty/milne/322/IIROC_FeeCh...
As a retail trader, no.
As an example, yesterday I placed a sell order on $90,000HK worth of a stock. Once I hit 'send', my order was fulfilled before my browser could load the confirmation page, and at the market price I was quoted seconds before.
This is, in large part, thanks to market makers who use HFT. Before this, the broker/market maker might take a spread worth half a percent or more (and being a retail investor, you probably wouldn't get the 'market' price), now it's pennies or less, and at the market price.
Thanks to HFT, spreads are smaller, execution quicker, and it definitely 'levels' the playing field.
>it definitely 'levels' the playing field.
Have you read Michael Lewis's Flash Boys? Because HFT demonstrably doesn't level the playing field.
If your entire strategy is arbitrage, then of course you're going to lose out to someone who is quicker and has better technology. Of course some people who lost out on 'low hanging fruit' are going to be mad someone else bought a bigger ladder.
But if your strategy is anything else (investing or any type of speculative trading) then HFT benefits you.
I've traded on markets with low liquidity and no HFT, and high liquidity with HFT. I don't miss the low-liquidity markets, waiting half a day to see your order executed only to see the price move against you because no one could match your order, meanwhile institutional traders are trading the same stock outside the exchange is the worst kind of infuriating.
The argument against HFT is like saying that bank tellers who exchange currency for a 2% spread are cheating out back-alley money changers who take a 20% spread... Yes someone is losing, but it's not necessarily a bad thing.
Also, slowing down trading doesn't mean eliminating HFT, it's a matter of what constitutes "high" frequency... I'd argue we're well past the point where incremental increases in speed result in equal gains in liquidity. And those increments now cost more than ever before.
And you're right, we don't need market makers trading as fast or as frequently as they do, but they see an opportunity, so they go for it, and we benefit anyway.
I'd personally be happy with 10 second execution, but if I can get 1/2 second execution, why would I complain?
Is that true? Isn't there a high barrier of entry? I was under the impression that large trading firms were building high-speed connections, which is obviously not something a small firm could ever do.
For "kind of fast" - 10-100 microseconds - there are a variety of brokerages that can get you started with costs of approximately $1000-10,000 a month. This is rapidly changing though - the exchanges have been continuously increasing the prices of their market data to the point where consuming the entire US equity market proprietary feeds costs around $50,000/mo in licensing.
https://www.chrisstucchio.com/blog/2012/hft_apology2.html
This can be partially fixed with a very technocratic market microstructure change (eliminating the subpenny rule). But politically that's very much a "huh?" point - imagine Bernie Sanders saying "I believe we should let traders quote in increments of 1/100 of a cent, not 1 cent".
https://www.chrisstucchio.com/blog/2012/hft_whats_broken.htm...
(This would fix things on the placing orders side, but not on the cancelling orders side.)
Decimalization would help even with equities where the natural size > 1c. HFTs who want to get to the top of the book could compete by offering $10.0073 instead of racing to be the fastest at $10.0100. HFTs would compete on price rather than speed.
Do you have support for that claim?
(I don't have an opinion; Trying to form one based on data)
Since then, we have additional data points from the decimalized US equity markets. See http://www.sec.gov/rules/other/2014/34-72460.pdf for a bibliography. The weight of the evidence points towards thinner books. Incidentally, the SEC is looking to increase the tick size for illiquid small-cap stocks for this very reason.
I run algo strategies myself - not HFT - and my monthly trading costs are less than my monthly beer costs. Capital invested is perhaps 1-2 years savings for any software engineer (e.g., my best recent trade was dropping $40k on SPY when it was at $190, it's now at $201.66).
Although of course making such a distinction by law is problematic, because updating an order to be "out-of-the-money" (have an absurdly low or high price) is almost equivalent to canceling, and objectively determining if an order is "out-of-the-money" is problematic.
For the same reason that people don't generally cheer when a gang war happens. Just because the gangsters are mainly shooting other gangsters doesn't mean the violence isn't costly to society. The same with Wall Street. Just because people don't trust big banks and suspect them of rooking the common man doesn't mean that they want to legitimize said rooking, or that it isn't costly to society when banks rip each other off.
There is no violence here. Only competition.
In most industries we applaud this disruption.
Also, the people who applaud 'disruption' are not this sort of populist, they don't cheer because of the damage done to incumbents but because the disruption will hopefully mean better products/services/prices for customers. HFT just does the job of the old investment bankers slightly more efficiently. If you think investment bankers are bad, you're not going to think that more efficient investment are an improvement.
This happens a lot more than you might think. There is always a temptation to stuff the order book to keep it going in a direction profitable for you (ie, fake volatility).
This is the #1 reason why canceled ordered from HFT are suspect. It's too easy to stuff the order book and cheat a little. You really need highly credible market makers who have proven not to do this sort of thing. Even then, such folks are very rare and hard to appreciate. Your algorithms get so complex and obfuscated it's hard to tell what's stuffing the order book and what's simple market making.
The fact is, everyone is stuffing the order book. Sarao is more a patsy than anything else. His biggest crime wasn't belonging to a Goldman Sachs paying bribes to political entities.
Is this something you have personal experience in? I ask because a lot of people relate this concern because they read about it in Zero Hedge, which is regarded by people in the industry as (as someone here once put it) "a conspiracy theory site without the theories".
WRT suffing the order book - at least as far back as 2003, in Eurex, there was a swiss trader who would do that in bond futures. There were a lot of comlpaints, and even death threats IIRC, but a Eurex investigation at the time found he did nothing wrong - their main finding was that he would occasionally get those "stuffing" orders executed, which means that (as far as they are concerned) they are not false or misleading in any way.
That's probably the gist of it: as long as you are willing to take the hit if your bluff goes against you, there's nothing "fake" about it. It's been years since, and I know not of any (officially investigated) cases back then, or of any recent cases - but I'd be surprised if it's not very prevalent. (Also, I've been out of the HFT world for a while).
Here's a good article for you to read: http://www.bloombergview.com/articles/2014-10-02/prosecutors...
There's a reason why they made spoofing illegal. The problem is, that you can't tell the difference between spoofing and just market making, especially as the algos get particularly complex. And when you start looking at in aggregate you see the real problem.
Wouldn't the market be better served by a window structure with a scale on a more human timespan? Say a 5 min process in which:
* for 4 min orders are taken in confidential secret.
* there is a 1 min blackout window in which no orders are taken, and in which any results of settling the outcome of orders is not published.
* At the end of that period the new results are published. (It doesn't matter who reacts first, it's who reacts best.)
The profit between what sellers are asking for and what the buyers are willing to pay still needs to be accounted for. This could be the market's operating fee (the cost of the sale's commission), it could be attributed to the government as a form of sales tax, it could be returned in some way to the parties involved (buyers pay less, sellers earn more), other, or some combination of the above.
You could slow down the process, but that adds risk (that one party is getting the wrong price). The market makers would say, I'll sell you one 1 share for $1, 10 shares for $1.20, 100 shares for $2, 1000 shares for $10. So then someone who wants to buy 1000 shares for $1 a share will buy only 1 share at $1. Then in the quiet period, the market makers will adjust their prices. 1 share for $1.01, etc. Then you'll sell another share for $1.01. Then you'll wait 5 minutes, and the market makers will adjust their price.
The market is going to work the same way at 5 minute delay as it works at nanosecond delay, but just be slower. The market makers won't be getting you a better price, and your large transaction won't not affect the share price.
High Frequency Traders (HFT) do a lot of positive things for a market - mainly providing liquidity where they might otherwise not be. They also act as a sort-of balancing force for undervalued or overvalued stocks (in the volume the HFT's trade with over time), "normalizing" the VWAP and TWAP (among other health indicators) for a given ticker.
In other words, markets want (and benefit from) HFT's.
If not, why isn't there a bigger push for market infrastructure that doesn't advantage high frequency trades?
I don't know that most high frequency microtrades are a real problem either way, since it seems mostly zero sum among the people playing that game. But whenever there is a macro change in a price that takes place over a tiny time scale, it seems that we shouldn't be advantaging speed in getting there first.
He does a great job presenting a fair and deep view of a lot of finance issues, like HFT or Unicorn valuations.