Guy trading at home caused the flash crash of 2010 | Hacker News Reader

Guy trading at home caused the flash crash of 2010

bloombergview.com

63 pointsbreitling84 comments

My (limited) understanding is that almost all high frequency hedge funds employ this tactic of placing and cancelling limit orders. Not sure why this guy is being prosecuted.

> We investigate the trading of one hundred Nasdaq-listed stocks on INET, a limit order book. In contrast to the usual view, we find that over one-third of nonmarketable limit orders are cancelled within two seconds [0]

When so many participants and breaking the law and the law seems to be applied somewhat subjectively, then it is ripe for corruption and political targetting. Kind of like how political dissidents get arrested for 'tax evasion' in Russia

To be fair, this study was performed in 2007 and spoofing became explicitly illegal under Dodd Frank passed in 2010.

I mentioned this in a comment in an earlier reply, but I think some skepticism around market participants is healthy. If market participants assume that there is no fraud or that fraud will be prosecuted by central authorities, they will be more easily defrauded.

Edit: Looks like the number of unfilled orders is over 90%

> New data from the US Securities and Exchange Commission (SEC) show that only 3.2% of the orders placed in the stock market in the second quarter of 2013 actually went through. [1]

[0] http://papers.ssrn.com/sol3/papers.cfm?abstract_id=994369 [1] http://qz.com/133695/96-8-of-trades-placed-in-the-us-stock-m...

You are right that most place and cancel orders rapidly, but that isn't necessarily spoofing, which requires some intentionality. For example, Nasdaq even has an order type that auto-adjusts itself such that it will never execute, but it was mainly created to help market-makers dodge quoting requirements, not to enable spoofing.

In symbols where they are a registered market-maker, market-makers are supposed to be obligated to make a market in the symbol. In exchange they get several regulatory benefits, like being able to naked short a stock (sell it without even locating/borrowing the shares).

In practice most automated market-makers used to just put out bogus quotes they never expected to fill at $0.1/$1000000.0 bid/ask. The SEC tried to regulate this and say you had to really be making a market and put your quotes within X% of the current bid/ask. Market-makers started automatically cancelling and replacing at exactly X% away, but that was too much for them. So they had Nasdaq create an order type that automatically cancels itself and replaces it X% away from the bid/ask anytime the NBBO changes.

What they are doing in that example is completely wrong, but not analogous to what the guy in this article was allegedly doing (this guy put his quotes near the money to make fake interest, market-makers put theirs as far away from the money as possible to be fake market-makers and get regulatory benefits like naked shorting in exchange, their orders don't appear like real buy/sell interest in the same way this guy's allegedly did).

---

The author of the article tries to distance algorithmic-HFT from this guy operated, to cleanse its name, but come one:

    Sarao may have been a spoofer, but he doesn't seem to
    have been doing the sort of high-speed algorithmic
    trading that usually qualifies as "HFT." He himself 
    claimed to be "'an old school point and click prop 
    trader' who had 'always been good with reflexes and
    doing things quick,'" and the "Layering Algorithm"
    that he used was a customized version of "a program
    that allowed non-programmers to engage in automated
    trading using spreadsheet commands and functions." He
    was also the sole owner and employee of his trading
    firm, which he "operated from his residence." In style
    and substance he is not all that different from other
    spoofers we have known and loved, who did their high
    speed trading by just punching keys really fast.

In style and point-and-click background maybe he wasn't a goliath HFT operation, but in substance, spreadsheet automated trading at faster than human speeds, he definitely was.

If the charges hold up, they will have been able to convict this guy because of evidence of intentionality (the emails to his FCM). But with trading strategies now starting to delve into modern deep-learning techniques, we're probably going to start seeing algorithms learn to spoof from a blank slate. The only way to tell if one of these algorithms is really intending to buy or sell or is spoofing would be to reverse engineer something rather inscrutable... And is it the author's intentionality or the algorithm's we're really concerned with? Likely we'll end up holding things to some gamable metrics that will do about as good a job as the market-maker quoting requirements do in assuring market-maker algorithms are providing liquidity.

Wouldn't that be illegal under Dodd-Frank then?

Do you know the name of that type of order?

He's probably referring to a Market Maker Pegged Order.

Thanks! Love HN.

It seems like the real fix would be to have a cancellation fee. If an order is canceled within 60 seconds, penalize the trader .1% of the total amount of the order. That would make all this gamesmanship unprofitable, and the order book would actually be a collection of people wanting to buy and sell shares.

It seems very wrong to me to make an activity feasible, and allow it to be profitable, but then arbitrarily drop the hammer of criminal penalties when a trader crosses some murky line.

There already is. If you add liquidity you get a bit of a bonus, if you remove it (cancel your limit order), you are charged a small amount.

Why are people so fixated on the order book? The real information is the spot price and how fresh is that information (and the volume), the order book is a fantasy anyhow. There are people willing to buy stuff at pennies always, and sell at thousands. How does that help you? What if they cancel? It shouldn't matter if someone does it closer to the spot price, or closer to the best bid-offer.

If you want to buy, look at the price, decide how much you think it's worth, put in a limit order at that price. Go to bed. If you want to sell, basically do the same. If you want to live on trading, then smarten up.

> Nasdaq even has an order type that auto-adjusts itself such that it will never execute

That's not true, the order could execute if someone slams the market with a massive buy / sell order. It auto adjusts to stay away from the active market. The idea behind the order type is that you don't want to be at the current 'market price' but rather just outside to handle big fish. You might argue this sucks, but the the big fish don't mind as it gives them more liquidity that wouldn't otherwise exist. The trade also lightens load as you don't have to constantly send cancels.

You're being really pedantic, yes a single big sell order that blew through the entire book in one go could hit it. After the regulation was added, but before Nasdaq added the auto-adjusting type, market-makers, in stocks they didn't really feel like market making in just constantly cancel-replaced their order each time the quotes changed, a single whale order could have hit those too.

Maybe you aren't being pedantic: are arguing the intent of the regulation was to have market-makers make markets only as a bulwark against single whale-orders that will probably be thrown out as clearly erroneous?

That is a real role-redefinition. A quoting obligation that falls apart if the big-order trader on the other side randomly decided to break his order two (or N) near-simultaneous orders?

I'm not being pedantic at all. It's the whole point of the order type and Market Making. Being a provider of liquidity as last resort. This order type is only available for MMs, btw.

Ideally, the order will never execute as the regular market will handle the volume. Things are in bad shape when you have to dig into the Market Maker's pockets.

order type description: https://www.nasdaqtrader.com/content/productsservices/tradin...

Ok, you are saying market makers are there to handle single whale orders, and not floods of one-sided orders during rumor panics etc. And you are saying that market makers should be obligated to take on a single whale order, but if the same whale-order-placer instead splits his whale it into chunks and simultaneously places them, maybe to hit multiple exchanges at once, the market maker obligation goes away as long as the first chunk is small enough not to blow away the book.

Yep, the order type is only for market-makers. That's pedantic too; I only ever talked about it in the context of market-makers, as an example of place-cancel not being the same as spoofing. I'm not sure if you are pointing this out to say I am conflating them. Maybe I should have picked a more noble non-spoofing example?

Just a reminder that this is an instance where we are saying someone is engaging in "algorithmic HFT" from a house 4000 miles away from the trading venue, using what appears to be Excel and Visual Basic.

How is flashing $2.5 billion in orders at super-human speed not "algorithmic HFT"?

You maybe have equated HFT and ultra-low-latency trading in your mind? The latter is almost always high-frequency because it is hard, on a few pennies per trade, to make any money without the F in HFT, frequency. But all areas of HFT aren't necessarily so latency sensitive as areas like index-underlyings arbitrage where you have to be colocated at the exchange to be competitive and 4,000 miles away sounds like a ridiculous punchline.

At a heavily automated market-maker I worked at, we, not so long ago, had all our servers almost a thousand miles from new york, even farther from Chicago, and still did about 8 or 9% of all US equities volume. It sure as fuck didn't run on Visual Basic, but you would be surprised how much algorithmic trading does out there.

If you want to call one guy's Excel plugin-in talking to Globex from London HFT, that's fine; I'm just saying: that's what were talking about here. If HFT includes things you could run on a WinXP-era laptop with WinXP-era software --- again, fine --- then we're in a weird place w/r/t/ the idea that HFT is something further segmenting the market into haves/insiders and have-nots/outsiders.

The "2.5 billion in orders" is just a number; in fact, the unreality of that number is the problem.

This seems really naive. A lot of HFT is taking something a trader has figured out and can do in real time for one symbol, automating it, and applying it to 2000 symbols at once. A WinXP era-laptop can be fine, if the trader's strategy worked at human-scale latency and limited human input bandwidth on one symbol, a WinXP-era laptop might indeed be able to do the same for some thousands of symbols.

I think we might be a bit biased on this site because a lot of the HFT recruiting targeted at techies in other industries is going to be in cutting-edge areas of ultra-low-latency or generically-dealing-with-massive-data where experience in a different industry can directly cross over.

I think you think I'm making an argument I'm not making.

For the most part, I think HFT is innocuous.

For the most part, HN does not think HFT is innocuous.

All I'm suggesting here is, if an Excel plug-in talking to Globex from London is "HFT", that shoots a hole in one of the primary arguments used against HFT here.

Really, what we're running up against here is the silliness of the term "HFT". By the definition we're contemplating here, virtually every ATS in the world is an HFT.

> that shoots a hole in one of the primary arguments used against HFT here.

Which one specifically? That the big players have an unfair advantage over the little guy?

> Really, what we're running up against here is the silliness of the term "HFT". By the definition we're contemplating here, virtually every ATS in the world is an HFT.

Not really, if it doesn't deal with market micro-structure, it probably isn't HFT. There are lots of ATSes that do routine things like rebalancing portfolios, or more sophisticated things like automated hedging of human traders, that is still happening at a human pace.

I haven't seen a detailed article yet that covers how many order modifications he made, and how far outside the money his spoof orders were (really far and it is something he could have managed by hand and I wouldn't call it HFT, pennies-close and his London Excel setup couldn't have handled it, but I think there is definitely a big HFT Goldilocks zone in there).

Look, spoofing is already illegal, and is already a problem in manual trading. Whether the guy did HFT spoofing or not isn't really damning to HFT at all. It really just shifts attention away from how fast all the illusionary HFT liquidity (tightening spreads and providing valuable liquidity is how pro-HFT arguments normally counter anti-HFT arguments about it a zero-sum game) evaporated during the flash crash.

My original comment can shed some light on that. Part of the regulators' approach, post flash-crash, to getting more liquidity in events like that was to add more stringent quoting obligations on market makers. The regulation was already ineffective and poorly thought out, but to get an idea of how much the industry really values liquidity don't look at the bullshit they sell in congressional hearings (we pay the market in liquidity!), but instead look at the product they demand as customers of Nasdaq (a new, explicitly anti-liquidity order type).

(edit: ok a bit overblown on that last part, I don't think the flash-crash was that big of a deal. The HFT industry just really oversells liquidity and it is kind of a farce, the market literally shuts down for two days each week and takes ten days of holiday each year.. literally a near universal zero liquidity for 30% of the year or more if you consider market hours)

It is legal to change your mind quickly. It is illegal to place an order with the intention to cancel it before it executes.

Which makes the law entirely unenforceable unless the perpetrator is dumb / arrogant / ignorant enough to brag about it...

But these are high-frequency trades, which mean that they're executed by a computer programmed to do that. If you can examine exactly what it was programmed to do then that could probably help prove intent.

"You're honor, it wasn't me, it just deep-learned it that way."

If a program uses AI to come up with strategies, how do you make sure it doesn't learn to spoof? Coming up with the goal-formulation of "don't spoof" seems like it would require the program to introspect about its own intent. Dunno, I would like to hear what kind of metrics one would use to make the learning algorithm reject spoofing strategies. I guess it would be something similar to whatever heuristics the SEC uses to flag potential spoofing, but they get a lot of false positives and almost never convict someone on metrics.

But what does "intention to cancel" mean when you're offering to sell stock at an above-market price? What if I'm perfectly happy letting someone buy that stock from me because I can replace it at a lower price once the market settles? I'm still trying to manipulate the market prices, but am I "spoofing"?

If you're actually willing to trade at that above-market price if the market moves suddenly, then that's a legitimate strategy and not manipulation. You're playing the role of a shock-absorber. If your plan is to cancel as quickly as possible if you see the market moving toward you, then that's illegal.

But I can do both of those at the same time. Let's say I'm willing to trade at that price if the market moves suddenly, but I want to change my offer if the market moves slowly. Am I intending to cancel now?

he's referring to the unconditional intention to cancel an order. if any of the future states of the world result in you wanting to trade the order that you placed, it is a valid order.

Great, then the smart spoofer isn't spoofing! Because they accept that they might actually trade the order.

Even if they admit to spoofing as a general activity, they're just talking about the common case. They're not placing orders with unconditional intention to cancel.

Well that doesn't lead to a very helpful law...

You could say the same thing about almost any law. The difference between an accidental fire and arson is criminal intent. While in theory it's not possible to observe someone's inner mental state, in practice their outward actions usually give a pretty good clue, and edge cases where the same actions could be interpreted either way are rare. The same is true in spoofing: while it's theoretically possible that exactly the same strategy might be used by a legitimate trader and by a spoofer, differing only by intent, in practice some strategies are overwhelmingly more likely to be spoofing. The edge cases are tricky, but they don't invalidate the entire concept.

In this particular case, the trader in question would have been financially ruined if his spoofed orders had been executed, so it's pretty clear he didn't intend for them to be.

Caveat: In my opinion, this isn't a distinction well suited to the criminal justice system. I'd rather that exchanges identify probable spoofers and ban them from trading (after some warnings), with the burden of proof (in arbitration) on the accused to show that the suspicious orders were part of a legitimate strategy.

>In this particular case, the trader in question would have been financially ruined if his spoofed orders had been executed, so it's pretty clear he didn't intend for them to be.

Are you sure? If he bought the stocks at the heightened price he could just resell them right away. I'd have expected it would take a much more permanent shift to ruin him.

Or if I put it another way: It seems to me that there is a solid method for making money that's 95% the same as spoofing and 5% actually being happy when your "fake" orders execute. So the intent is still to manipulate the market but only through truthful orders. Then you can no longer look at this pattern of behavior and say that it's overwhelmingly likely to be spoofing. Now what do you do? Do any laws stop this?

>> Not sure why this guy is being prosecuted.

It's traditional to praise the big finance houses when they pull this stuff, but for individuals it's a crime, because they found a way to 'cheat' the big boys.

Here's another example - http://www.computerworlduk.com/news/security/3244186/norwegi...

Could you find the most specific example you can of a large investment bank (or equivalent firm) being praised for manipulating the market? Specific enough that we can actually dig into what specifically they're being praised for doing?

Any central bank since 2008.

Wow, that case in contrast to the current one actually seems absurd. Upon further research though they were cleared of any wrongdoing on appeal in 2012 [1]

[1] http://www.ft.com/intl/cms/s/0/e2f6d1cc-9447-11e1-bb47-00144...

HFTs place and cancel orders with the goal of getting fills. They rapidly place and cancel orders with the goal of getting fills at the best price. They also engage in all sorts of other practices (e.g., buying order flow) designed to get more fills.

I.e., if the nbbo for SYMB is $10.00/10.10, an HFT might place an order at $10.01, then a bunch more pile on. Then the market moves down and the HFT cancels the $10.01 order and puts in another one at $9.95, in the hope that SYMB will move down to meet him. The goal is to get a fill from an uninformed (i.e. small) trader.

In contrast, this guy was deliberately avoiding fills, and more importantly, sent an email proving this was his goal. (That latter part is very important when the illegality of a specific action varies depending on what your goal is.)

My understanding is that this guy did exactly the same thing you've described a HFT would do. The intention of $10.01 is to move a price, not to get a fill.

No, the intention of 10.01 is to capture the fill when the market goes there. The intention of a spoofer is to induce the market to go somewhere it shouldn't, jump the hell out of the way so as not to fill at that bad price, and then profit from the correction back to the right price.

> Not sure why this guy is being prosecuted.

You have to start somewhere / make an example of someone. Lone wolves trading from their houses have fewest friends and are easiest targets. Hopefully they go after bigger fish as well.

Compete fantasy. Eric Holder is on record stating he did not pursue criminal charges against large financials because of systemic risk.

One law for some, another for everyone else.

The job of a market maker is to ensure that there is always a counterparty for buyers and sellers; an MM takes a chaotic marketplace and, for those who need it, presents the illusion that (for instance) you can buy a security the same way you buy a book on Amazon: here's a price, click buy, done.

If you try to think about it, it's hard to get your head around how MMs can provide this service. Ensure a counterparty to buy from, or sell to? The market is constantly moving. From where do they get the superpower of always having stuff to sell, or always being able to buy? How are they not taking a bath doing this?

One reason MMs can exist is because they charge a premium and pay a discounted rate for the securities they charge. The premium you pay an MM when you buy a security offsets the risk that they're getting the fuzzy end of the lollipop, which, if it happened over and over again, would force the MM out of business.

But: it's tricky to charge a premium (or pay a discount) against an estimated fair market value that is constantly changing. Also: the size of the premium/discount (the spread) depends on factors that also constantly change.

For me, knowing these things makes it intuitively obvious why MMs need to constantly submit and cancel new orders. It's like the edits in a simple diff algorithm; cancellations are the "-" lines, order entries are "+" lines, and a "change" in an order requires both.

"the fuzzy end of the lollipop"? I can guess the context, but I have absolutely no idea what that analogy means.

EDIT: http://www.urbandictionary.com/define.php?term=fuzzy+end+of+...

Could someone point out where what this guy did was illegal? The guy simply outsmarted the algo traders by acting like he was going to push his chips in, than yanked them back. How exactly is that illegal? And if it is illegal why is the FCM allowing it? Even more so, how is it legal for people to see what he MIGHT do?

This feels like a whole lot of sour grapes by the big boys. They're pissed because someone figured out their weakness.

The complaint http://www.justice.gov/sites/default/files/opa/press-release...

lists which exact laws he's accused of breaking on the first page.

Spoofing was made explicitly illegal under Dodd-Frank and is placing orders with the intent of cancelling them before execution.

Generally its hard to prove intent. You can place an order and cancel it after new information becomes made available or your needs change.

Placing orders that can't execute has apparently been verboten long before Dodd-Frank.

There's a technical answer, but more simply, you answered your own question: you can't push your chips in and then yank them back.

Try that at a poker table - yes, you'll "outsmart" the other players by being able to see their reaction. You'll also instantly break the integrity of the game, because you're not outsmarting them, you're breaking the rules that allow the game/market to actually function.

For a market to function properly that integrity that an order on the books is in good faith is vital. Of course, it's up to the SEC to enforce that.

I think he was making the point that if something is permitted by the "rules of the game", then by definition, it cannot be a violation of the rules. And in the case of financial transactions APIs, the rules should exist as application logic. So if you're not supposed to place and immediately cancel an order, it should simply not be possible.

I could certainly believe that if it was one of the large HFTs doing this, they would have had the necessary wheel-grease to not get in trouble...

Placing and immediately canceling an order is fine and allowed (though doing that too much is penalized for other reasons).

Spoofing is pulling them out in a coordinated fashion before they can be put at risk with no intention of them ever trading.

The problem is that an api that prevented that would also prevent legitimate cancels that would have side effects that could be bad (ie making it riskier to make markets and therefore increase the bid/ask spread).

Spoofing is about intention. Intention cannot be determined by algorithm (yet).

Also to your point about a big HFT not being subject to this. Allston trading is a large HFT market maker that is currently in arbitration over spoofing.

You can totally push your chips in and yank them back. What you can't do is pretend to push your chips in, such that it is almost impossible to ever lose them, to bluff other players without taking any meaningful risk.

And these were real orders with real risk, so...

So we only have a pattern of behavior and any communications about said behavior to determine intent.

But but but, how come it is even possible to yank your chips back? It would be so easy to force those orders to stay for a minimum amount of time.

That wouldn't prevent spoofing. Lots of spoofers put orders in over the course of a long time. Its when/why they pull them out in coordinated fashion that the intention becomes clear.

Still: if someone tries to buy an order, the transaction is automatically accepted, and the seller can do nothing about it.

Right? Right?

Don't tell me one can withdraw an order after it has been established that someone else is trying to buy it? That would be way too easy to abuse. "Oh, someone actually wants my stuff? Sorry, I just happened to change my mind."

No you can't do that.

Spoofing is really nuanced and analogies make it difficult. Usually spoofers will put orders in the back of the order book where they feel they can cancel them before anyone can realistically get to them. But they are in the order book, so they are "technically" at risk of being fulfilled.

Thats why intent is so important.

Thanks for pointing out. I feel the exact same way about it.

What is wrong in tricking stupid bots in the market? Lets not forget, it was a flash crash, meaning that if only emotional, slow humans would have been trading this thing, it would probably have never happened (at least not in these proportions).

Pretty far fetched to make one small guy responsible for a conceptual problem in the system.

I think it was that this guy traded to "cause" instability, by placing bogus futures trades that were basically the value of the entire order book for those futures.

If you have a Level II stock data feed, you can view the order book for particular stocks, and similar data is available for futures from CME group.

As an example, a friend of mine day trades Tesla stock (& options) and basically a swing trade, has figured out if the stock cycles down by a few cents, and they put a relatively large order at a certain price. The market sees the trade with a higher price and the stock increases. This has resulted in some large profits for swing trades, somewhat due to believing in Tesla's fundamentals and in Elon Musk. Some institutional investors don't share my friends views, and trade Tesla stock down. It is legal because the trade was made in good faith, and not trying to "deceive" the market using bogus trades.

In traditional HFT, the legal justification for fast trades that in some cases were never intended to be filled, is murky at best. As long as the market stays reasonably stable and the big boys profit, the complaints are somewhat muted.

A friend in the same group, who also trades Tesla, was on the console during the flash crash. It was "obvious" that these automated trade bots were unloading, and there was not major news (eg. GM bankruptcy) to justify the sudden price drop in specific stocks. Her console showed "black swan" type data, and she switched to an option trade as stocks caught up in the flash event wouldn't stay down for long. The options that were "out of the money" (almost worthless) became valuable when the stocks rebounded.

What caused the actual flash crash (triggered by futures trades or not) was a phenomenon called "exponential backoff" with a large automated trading bot closing out its trades, triggered by stocks below a certain price, causing downward pressure on the stocks, and the market. Another automated trade bot sees this, and also closes out its position. This automated close out then occurs exponentially, and the market goes into a "death spiral" crash.

Note: I do not own stock in Tesla. I'm not a stock broker and this is not investment advice.

Well I sure am glad we found a scapegoat that doesn't require us to rethink HFT in any way. /s

As the article points out - this guy stopped his "spoofing" strategy and the market crash accelerated. This bit of info - coupled with the fact that it took this long to find him makes me think that this guy is only a small part of what happened that day.

Perhaps covering up more major players or corporations or even potential underlying fundamental flaws?

The E-mini S&P futures market is incredibly illiquid, which is why one guy can game it this way. Look at the current volume on the June 15th contract: http://finance.yahoo.com/echarts?s=ESM15.CME+Interactive#sym...

It's 343! That's zero to a good approximation.

So this guy found a low-volume market that he could game with large sell orders above the current price that he would cancel (automatically, eventually) so he had a low (but crucially: non-zero) risk of actually having to sell at that price. Others would see the large sell volume and sell themselves, driving the price down. He would buy at the lower price, stop placing sell orders and watch the price bounce back up, absent his manipulation.

This is likely illegal in the letter of the law as written in 2010, and fair enough. Illiquid markets are easily subject to this kind of thing and if you want people to invest in them maybe you need to police them against it. Whatever.

But to call this "the cause" of the flash crash is like saying my neighbour's love of fireworks is "the cause" of my house burning down after I deliberately poured gasoline all over it. Sure, maybe my neighbour should have been more careful, but they had a reasonable expectation that my house would be ordinarily fire-proof, and behaved accordingly.

And of course this guy dominated the market when it was falling rapidly: his algorithm would have a field day in such an event, placing and cancelling multiple sell orders as fast as it could all the way down. So to cite that as some kind of evidence is to put effect before cause.

The only way this could have "caused" the flash crash is if a whole lot of large trading algorithms were using E-mini S&P futures as a a major input, and doing that--using such an illiquid, easily-manipulated market to drive large orders--that is criminal. Or should be.

Dude the front month S&P e-mini is one of the most liquid products on the planet. You are looking at a crappy chart. Look here: http://www.cmegroup.com/trading/equity-index/us-index/e-mini...

This is a joke, they found the poor guy who was not big enough to have friends at the cftc. There is nothing wrong with flash crash and rally, that is just volatility, you just use the right leverage.

The great depression was 'just' volatility if you look at a long enough timeline.

Flash crashes are scary for lots of reasons.

But they're also opportunities to own parts of a company much more cheaply than normal.

There are two possibilities here: this guy has been set up as the patsy to take the fall for this or, more terrifying, our markets are so brittle and levered that a guy at home with no juice can wipe out a trillion in notional value from his home.

I pray that the former is the more likely scenario as the latter is simply too awful to contemplate.

The latter is somewhat true, and probably unavoidable at this point. "The cat is out of the bag", so to speak--markets can't just go back. The problem is that the incentives are so wildly disproportionate. If a really smart guy works for the SEC or some surveillance body, he might $200k a year. That same really smart guy might make $200 MILLION trading. It has to be awfully damned unlikely to not choose the latter. And in reality, you're probably pulling $100k somehow or another while you wait.

With respect to the idea that this guy is the culprit, that's literally laugh-out-loud funny. It's possible he was spoofing, etc., even with decent size. But the clearing firm (Hi, MF!) controls the throttles on those pipes. But there is what is called "sponsored direct access" in these markets, and that basically means the clearer wants your business enough you can just hook up directly so you can go really fast, and they (the clearer) will just pretend that they're looking at your stuff.

200$ millions trading by themselves, without external investors? Very, very unlikely.

Old gripe of mine. I hate this phrase:

investors saw nearly $1 trillion of value erased from U.S. stocks in just minutes.

No. They did not. A stock's value is not "what some other dude is payed for it yesterday"; its value is whatever money you get when you sell it (plus, God forbid, whatever dividends it pays -- quaint idea, I know).

It's worth pointing out that this kind of behavior gets people in trouble on a fairly regular basis, what is new is that the DOJ is starting to bring criminal complaints.

It's long been the case that the SEC, or in this case the CFTC, will fine and temporarily/permanently ban you from trading for this kind of behavior, look at http://www.cftc.gov/LawRegulation/Enforcement/EnforcementAct...

Oh No! We should BAN guys trading from home! Only licensed professionals, employed by the big banks, should ever be given such responsibility!

It seems most likely that another algorithm was looking at the full book and reacted to his spoofing order with it's own orders (that were actually executed) and which accelerated the price downward. This would have tripped algos looking at momentum and other stop loss orders in rapid succession.

This is why it's dangerous to have algos run on the full book, especially without some logic to remove outliers.

Thank goodness for Matt Levine. I read three other articles on this earlier today that didn't even raise the questions that he answered.

He is like the John Stuart of Fin reporting.

I have an solution for you - instead of in real time trade in rounds every minute or ten or even only once an hour. Now everyone has some time to think about his decisions. And the real world doesn't change much within a couple of minutes so no need to worry that you can not react to events in a timely manner.

problem with this soln can be seen in exchanges' auction sessions: sessions of several minutes/hours during which orders only net at the end. with visible orderbook this does not avoid the problem of spoofing or eliminate the value of speed. with invisible orderbook there is still benefit in placing trades right at the last moment before the netting because you continue to accrue micromkt information from continuously trading markets and economic info from the real world throughout the netting session. and if you want to make all exchanges in all locations of the world net at the same exact moment, this is a Hard Problem.

But with an invisible order book you face the trade-off between submitting your orders early an get priority over later orders and submitting late to incorporate the latest information.

in netting phases the queue is almost entirely price based and not time based.

I can't help but think, "if it's really that easy to crash the market, how come terrorists aren't doing it on a regular basis?" Seems way easier, safer, and more effective than shooting people or blowing them up.

You can't die in a blaze of glory (possibly) on the world stage from behind a computer screen.

One guy in his pajamas caused a market crash? How many of these events will it take to cause a fundamental rethink of the financialization of our economy? Where are the supposed benefits to justify finance gobbling up more than a third of corporate profits [0]? When will we finally see a financial trading tax [1] that ends High frequency trading (HFT) for good?

[0] http://angrybearblog.com/2012/01/where-has-all-money-gone-pa... [1] http://en.wikipedia.org/wiki/Financial_transaction_tax#Unite...

I'm sure it was a collection of Sarao's that did this, rather than Sarao alone. He probably just stood out and was easiest to prosecute. Hopefully more prosecutions are coming.

HN Mods, why this discussion was dropped two pages down?

Not a mod but I believe that there is an "HFT penalty". And probably rightly so. The discussions around HFT are notorious for not being very productive.

so the system was the culprit

Reply on news.ycombinator.com