Understanding ETF “Flash Crashes”
factset.com
factset.com
I used to work for a forex broker (forex is 24h a day) and we had offices in different timezones (Toronto, New York, London, Singapore) so we would always have trader coverage without requiring night shifts.
Not very good reasoning. It's not exactly on point, but I'm reminded of a quotation from G. K. Chesterton:
> In the matter of reforming things, as distinct from deforming them, there is one plain and simple principle; a principle which will probably be called a paradox. There exists in such a case a certain institution or law; let us say, for the sake of simplicity, a fence or gate erected across a road. The more modern type of reformer goes gaily up to it and says, “I don’t see the use of this; let us clear it away.” To which the more intelligent type of reformer will do well to answer: “If you don’t see the use of it, I certainly won’t let you clear it away. Go away and think. Then, when you can come back and tell me that you do see the use of it, I may allow you to destroy it.”
I mean, this is Hacker News. We discuss all day about what should or shouldn't be done. If you want action, go out and do it. Otherwise, we're just armchairing all of this.
Sure it does. Why should Chinese investors only be able to trade US equities from 9:30pm to 4:00am?
Discrete markets are, under certain circumstances, more efficient at connecting buyers and sellers than continuous markets. It's similar to how an eBay auction is a more efficient way to get the best price for your rare item rather than having a continuously open market where you have to keep deciding whether or not to accept people's lowball bids.
Why do we want efficient transactions? If it's for the benefit of the buyers and sellers, why shouldn't they be given the choice? It's as if we were saying that the stock market is for less sophisticated users than eBay!
EBay doesn't have the same kind of clearing houses and broker-dealer regulations that would make sold items be anything like stock shares, even if they were all brand new and retail packaged.
There was a ton of volume in both the futures and the pre-market, but it was still dwarfed by all the activity at and just after the open. Most people just don't want to trade until other people are trading, so there's this sort of feedback loop that drives volume to the open.
It is perhaps a calming influence. Companies can release results after the close, and everyone gets to think on it overnight. In a logarithmic sense, the market's 33% uptime is quite close to 100%, and it allows for a rhythm, maintenance, and reflection that would be lost to continuous trading.
Most animals sleep, and we've even evolved the seven day week. Perhaps it's a good thing.
If you take the opposite view, someone will always trade with you after hours at the right price.
And those after/pre market hours are only 4-6 hours.
The problem is already somewhat apparent: During the trading day the volume changes and a lower volume increases price impact. It seems logical that longer days spread out the volume even more, thus increasing price impact of orders.
And then you obviously close it for 63000000 ms each day.
More algorithms are trading around the clock though, since markets are more interconnected than ever. To trade effectively, you need to be more aware of what has happened since market close as well. More explicitly, instruments have global presences.
Running and maintaining systems, and trading 24/7 would be a major shift for a lot of institutions though. So, my guess is that it once an ATS can get approval to start trading continuously, outside human hours, the rest (including exchanges) will need to adapt.
http://www.marketwatch.com/story/a-brief-history-of-trading-...
Technically, there's no reason that the markets can't participate 24/7, but the Nasdaq operates from 4a to 8p, and the crash that occurs at 9:30 still occurred despite markets being available for participation between 4a to 9:30a.
A another solution would be a single auction a day, and everyone would get more fair but less timely trades. And a lot of the goofus mechanics around high speed continuous auction would disappear.
It is a trade off.
There's not actually so much new information coming out throughout the day that the market needs to update prices on a sub-second basis for hours and hours every workday.
But regardless, when you have a derivatively priced security like RSP (which holds actual stocks -- it has some inherent worth at all times that's knowable) any selling at a discount is supposed to be arbitraged away. In these cases, it wasn't,because the trading windows were too short.
(Hi, I'm the author of that blog post).
I assume you are referring to naked short selling.
To be clear, anyone may borrow shares and then sell them. Although legal ownership does pass to the borrower, this is not what most people think of if told they must "own shares prior to selling them".
In fact under SEC regulations, the borrow doesn't actually need to take place before the sale, so long as the broker has reasonable grounds to believe that the security can be borrowed to satisfy delivery for the sale.
From the retail traders perspective, this is typically transparent. They simply enter a sell order in a security that they don't own, and their broker worries about locating it for delivery (or doesn't allow the trade if it believes it can't locate the shares).
Until 10AM that day, the problem was zero buy-pressure. No bids at all to speak of.
http://www.etf.com/sites/default/files/images/2_rspsurveyor....
Also most high-frequency traders are market-makers or unofficially acting in that role. ISO orders are used by all executing brokers trying to sweep the best price, it's equally likely that the low sells were by funds executing stop loss orders through a brokerage or bank or wholesale market makers offsetting inventory at a loss after providing liquidity to retail stops.
So, knowing where AAPL and GOOG are trading, the ETF market-maker sets his bid and offer in MGT such that he can try to trade AAPL and GOOG at a better price to offset his risk and capture a small profit. Since this position is fully hedged, he gets special capital treatment and can accumulate large inventory in the ETF if there's a large imbalance between natural buyers and sellers.
Now, imagine AAPL is halted. The ETF market-maker can't hedge his risk or have confidence in what the ETF is worth. At first, he updates his pricing to assume the worst (remember, this is a dumb risk-averse computer, maybe AAPL is halted because it's the next Enron) and bids a lower price. He can't get the same capital treatment on MGT hedged with GOOG or some other ETF, and this is a much risker trade, more of a statistical arbitrage than a pure one, so as people keep selling MGT to him, he hits a risk or capital limit and pulls his bids completely.
Real ETFs can have 100s of components. Since this is a mechanical process, high-speed market-makers can charge a very low spread and be the best bid or offer in the market almost all the time. They just trade 100s of ETFs at once to make it up in volume. This drives humans out of the market-making business, since human traders who did the same thing are undercut by machines. Because of this, few well-capitalized human traders are sitting around waiting to pounce on ETFs trading at a massive discount since it rarely happens. They're all doing something else with their lives.
A good solution would be not to allow market orders in ETFs when their component stocks are halted, or to halt trading in ETFs when the components are halted. Market stop orders should probably be banned in general. I can't think of any other business dealing where a person would walk into a store and say "I'll buy/sell this at any price", aside from buying lobster on a date.
In a fictional world, an envelope costs $1, a piece of paper costs $0.50, and a stamp costs $0.50. In total, a full letter costs $2. You know that anyone on the street would be willing to buy each of those items at the component price. Would you not agree that seeing all 3 sold as a bundle for $1 would be irrational?
Or that people are actually, as has been demonstrated in any number of ways, irrational. The idea that perceiving irrationality means there is something wrong with your world view rather than that there is actual irrationality in the world is somewhat odd.
Stock opens at X, you buy at 0.5X, you sell at 0.9X, it goes back to X. Then the exchange breaks the 0.5x trade. Now you still have to deliver the order you sold at 0.9x, but you're buying it at X, so you've earned a neat loss.
The rational thing to do is to get a cup of coffee. Unless you enjoy losing tons of money, in which case your idea is great as well.
If you believe the stock is worth a lot more than 0.5X and you're generally a buy-and-hold type investor, buy it at 0.5X and wait. If the exchange chooses to break the 0.5X trade it costs you only the opportunity cost of being able to buy something else at a good price during the next few days. And if they allow the trade, you got a great security for half price -- so you can be patient and sell at a later date, or hang on to the asset for whatever dividends it might pay out.
Yeah, this can't be overemphasized. When some widow or orphan (or more likely GS) is on the other side of the trade, then clearly it should be broken. Wouldn't want them to lose any money. /sarcasm
As you say, the rational action is not to play.
Other people have opined on this and said that, if the "clearly erroneous" rules are properly defined, then a trade should never never never be broken. It should simply not be allowed to go thru. That makes more sense to me.
Edit: And for those who think this is merely hypothetical, or just whinging, you should read this: http://www.ft.com/cms/s/0/37fff9c6-0b36-11e3-bffc-00144feabd... GS trades were broken. Hopefully the FT paywall won't block that link, it worked for me when I searched for the story. So who knows.
The rational action is not to attempt a short-term play. Having one side of your two-sided trade broken is a very bad outcome.
But if you're making a long-term play, that consideration is irrelevant. There's no "other side" of the trade for you to match up with. If you get an asset at 0.5X and then don't sell it and then the exchange breaks the trade, you get your money back and you're no worse off than when you started. But if you happen to grab an asset at 0.5X and the exchange doesn't break the trade, then you gained an asset for half of what it's worth. Now you've got a great asset in your portfolio, only you have twice as much as you should have been able to afford. That's a great move!
I especially liked this quote: "The equity market is set up to dampen small movements (market makers tend to buy when others are selling), to exacerbate mid-size movements (as market makers capitulate and stops get triggered), and to dampen large movements (as circuit breakers cut off trading and let algorithms catch their breath)."
From my layman's point of view, it seems like this sort of "system behavior" is a peculiar emergent property that derives from the rules of the system and the individual actions of the actors working under those rules.
How can we take the fundamentally abstract idea of a company, its products, its business model, and its future potential, abstract that one level to a claim for some "portion of the company" (one share of equity, or a stock), abstract that into an index that tracks a portfolio of companies based on some abstracted notion of how to weigh different companies to model some desired strategy that tracks a market, sector, or industry. Then we're going to build software in order to abstract the task of trading whatever residual is leftover from this process amongst ourselves using formulas and algorithms that we think will outperform or trick other formulas and algorithms.
It's fascinatingly complex, which is why I suspect so many people on HN are trying to figure out how things like an ETF flash crash could happen. But for the average retail investor, it's damn-near impossible to comprehend.
The average consumer doesn't need to know anything about the drugs she takes. She just needs to know that the drugs are FDA approved and that she can trust the FDA.
I guess it comes down to how much you trust the SEC, FINRA, NYSE, etc. to stabilize and support markets. I have no idea if they're doing a good job or not. Any thoughts?
Of course, opinions can be influenced by whether the management of the markets is to your financial benefit, and perhaps that's why there seem to be a more people saying nothing: they're heads down, quietly making a fortune ...
It's worth noting (again) that the LULD circuit breakers were introduced after the 2010 flash crash, as an attempt by the regulators to "stabilize and support" markets. In some previous events, I think they've worked better than they did on Monday.
If they really feel that a circuit breaker serves does something useful here (and I'm skeptical this is the case) they should stop trading for longer periods of time and allow a larger window between haltings in order to allow all trade participants put in an order and not just a few of the fastest HFT engines.
I don't see what you think extending the circuit breaker time would accomplish? "Allowing a larger window between haltings" would defeat the whole point - if the price moves by more than 10% immediately after the re-opening auction, the circuit breaker goes off, as it's supposed to, for all the reasons it goes off whenever there's a sudden >10% price move.
It happens both up AND down. This last monday was really the first time they'd been tested en mass.
These people increase market liquidity, they too serve a purpose. If only long term investors were in the market, the exchanges would end up looking like ebay.
A 'stop loss' order will sell once the share price falls below a threshold. It will sell using a market order, meaning whatever the current best price might be.
A 'stop limit' order will sell once the share prices falls below a threshold too, but will sell using a limit order, setting a minimum price that you're prepared to be paid.
In a crashing market, stop loss orders can result in you selling stocks for pennies, regardless of their rational worth. Stop limit orders protect you from excessive price drops.
It might slightly favor HFTs if things do change across those 5 minute downtimes. Otherwise, your orders should be just as valid before the freeze as after, no?
They aren't fractal in that they are strongly scale sensitive, and at a small scale over a short enough time period on an unusual day you're basically looking at noise.
As per innumerable statistical analysis given enough noise samples all possible values (of price) are eventually, however temporarily, touched.
"Lastly, if we’re concerned about protecting individual investors, we need better education and controls around the most dangerous tool in the ETF investor’s toolbox: the market order."
Seriously, those things should just be banned from retail investment and retail investment education.
The problem is we aren't just talking about ETFs...there were similar effects happening in individual stocks where they were down much, much greater than this ETF. We should just ban stocks for retail investors?
What's wrong with people learning about things before they throw money into something they don't understand? Banning this and banning that blindly just makes people more dependent on someone else doing their homework for them. People need to take responsibility for their own actions and read up on stuff before dumping money into something.
Sure, it would be best if every retail investor learnt the market before participating - but that doesn't make it bad to take steps to protect those who don't (assuming those steps aren't particularly costly). If you can't remember the last time you placed a market order, surely it would be no big loss to you if they were banned entirely?
The best protection is to be well-informed of what is going on. Certainly, we can't know everything (such as how our food was prepared, quality of it, etc) but every investor putting money into stocks should know something as basic as a market order vs a limit order....that is NOT too much to ask.
I think you're greatly overestimating this effect. The government already regulates the stock market a lot. One more regulation isn't going to make retail investors suddenly start taking much bigger risks.
Why would this be the case?
A regulation prohibiting market orders doesn't even in theory serve to protect against poor-quality securities that will decline in value, it protects against poor quality market price information and/or order execution that results in an order being executed at a substantially different price than expected.
It might reduce the degree to which retail investors research information on order execution between different entities through which they might trade securities, but the jump from there to it reducing the degree to which they research the underlying securities seems unwarranted (if anything, it would seem more likely to focus their research more on the securities themselves.)
But, the thing is, its not like retail investors are bothering to research and understand the issues it would address now much at all, so its not like it would actually change anything except the degree to which they get bitten by them. (And, conversely, the degree to which counterparties benefit from them getting bitten by them.)
What? That's absurd, you want to ban "buy/sell now" at the market price? There is no justification for wanting to ban the most basic and most ordinary transaction type.
I'd argue fill my order "now" is pretty much aways everyone's intent.
So I wouldn't argue for banning market orders, but I'd probably stop using them as the defaults for people. Making people set limit order prices seems like it would encourage more people to understand the market dynamics.
Were I trading, I might think differently, but traders should be expected to know about limit orders.
If you encode that value into your order you are protected from flashes and guaranteed to get your order for up to that price (assuming the liquidity exists to support your order).
It buys at the current market price, which during the filling of the oder may slip as liquidity slips; that's the entire point, fill this order now no matter what. To not have such an order means having your entries and exits fail or partially fill which means you'll just have to reinvent the market order yourself to continue pushing through your order. To ban a market order is to tell me I'm not allowed to enter/exit a position in a single order, that's non-sense. If you don't want slippage, don't use a market order but that comes with the side effect of failed/partially filled orders.
You could carry on trading as you currently are. If a stock is trading $100/$100.1 and you want to buy it ('now' as you say), then just set your limit to $1000.
Of course, you will have no one to blame but yourself if, during the ~50 milliseconds it takes for the order to reach the exchange, the market makers have switched off and you end up crossing a massive spread and buying the only available offer at $600 or something. An incredibly unlikely scenario, but it can happen.
So I imagine you'll set your limit to something sensible, like $101. It'll still get filled 'now' 99.9999% of the time. But you won't be taking the risk of doing something epically stupid. There, that wasn't so hard now was it?
Your scenario doesn't handle the case a market order does, i.e. close my position now. In a fast moving market, using limit orders for stops or exits is just going to result in partial fills or failed orders. If I want out, I want out, I don't only want out if I lose between X and Y, I want out period; that's what market orders are for, for that convenience, you accept slippage as a reality.
> There, that wasn't so hard now was it?
No, it's easy to do things that don't work, I can pump my order into /dev/null, that's easy too, and it's just as useful as a limit order for stops.
I covered that. If you really, truly, want to sell at any price(and I can't imagine why, that is terrible trading), then you set your limit to zero.
That's all a market order is - a special case of a marketable limit order (immediate or cancel/fill and kill/etc) with the limit set to zero (in the case of a sell) or infinity (in the case of buy).
Banning vanilla market orders just forces people to acknowledge that fact, which is a good thing.
Perhaps to avoid a margin call; your imagination is limited, you seem to forget you also have to buy/sell to exit a position when holding on is not an option. Terrible trading is holding on until you get margin called; when your risk management says get out, you get out. A stop loss isn't a stop loss if it can fail to fill or only get partially filled, a market order is exactly what one desires in a stop loss.
> That's all a market order is - a special case of a marketable limit order (immediate or cancel/fill and kill/etc) with the limit set to zero (in the case of a sell) or infinity (in the case of buy).
Yes, and thus no need to ban it.
> Banning vanilla market orders just forces people to acknowledge that fact, which is a good thing.
You can't force people to acknowledge details they don't care to know about. Ban this, ban that; it's all non-sense from people who think they can force the ignorant to learn, you can't. Banning market orders achieves nothing.
I don't care how small a timescale you're looking at, the price of GOOG doesn't flit down to 0.01 or up to 1,000,000.00.
Is there another mechanism that serves the same purpose as a market order, but that isn't quite so vulnerable to these flash panics? Maybe something like "sell if it stays below $X for more than an hour"? Though I guess that wouldn't save you if the crash really was because of bad news.
In short there is no perfect "stop loss" order. There is either price risk or execution risk - pick your poison.
That said, unlimited price risk and zero execution risk is probably NOT the optimal point to be at.
Certainly, I would never put in a market order when the market is operating under circuit-breakers and things are unstable. When things are stable, it can be a reasonable choice for someone transacting in stock who just wants it done so they can get on with other things they have to do.
It makes people think about how the market actually works instead of how they hope it works.
I would be in favor of restricting the use of market orders by non-professionals during any time a market special rule is in place (circuit breakers, Rule 48, etc.) However, I think allowing them during normal market operation makes transacting easier for investors, which is also a good thing.
So I've seen things like having a trigger in an order manager that at a certain price will place an aggressive IOC limit order to unwind something. In normal conditions that will behave exactly like an at-market stop loss order, except that in the case of really extreme disruption you won't get filled before your limit. So the hope is in those cases the market will bounce. You don't want to get unwound at any cost on your stop if the market is going to rapidly recover afterwards. It also means there's no information leakage if you're worried about your broker taking advantage of the information in your stop (paranoid, but yes...).
As you've said, there's no panacea.
The issue there is that in a fast moving market, you might not get filled. When the stop triggers, the order "becomes" another limit order in the limit order book, and if the market's already moved bellow your limit, you dont execute.
Also, Market Orders are ALWAYS executed before limit orders, even if the limit makes it executable. That's just the priority system of the market.
In the continuous market time priority rules. If a marketable limit order arrives before a market order, the limit order executes first.
Stops are overused as a hedging strategy but there aren't really many alternative order types (stop limits are generally less vulnerable but not by much). If you are using stops it should because you've evaluated all your other hedging choices (options, futures, etc) and know they are right even in the face of flashes.
As for market orders the generally better choice is a fill or kill order that crosses the book.
Really you should buy the thing you actually want - a down-and-in put option struck at $70. But that would cost money, whereas you can do a stop-loss order "for free", so people prefer to do that.
IMO these stop-loss orders are bad for the market, people trying to get something for nothing, and if the increasing prevalence of flash crashes gets people to stop using them that will be do bad thing.
Stop losses are meant to exit you from a position, not just protect you from a crash, so it's not at all equivalent to getting a put. People use stops for profit taken when the trade is over or for cutting losses when the trade didn't go their way. It has nothing to do with getting something for free.
But I would lean towards the halts being an unintentional (as in: not well thought through) consequences of all the layers of regulation and counter-regulation.