In the wake of yesterday's market bloodshed, the NASDAQ announced that they'd be canceling trades more than 60% off of prevailing last sale before things went haywire. Now, I addressed this in last night's post, and others are writing about it this morning.
First of all, back to a solid explanation of what happened, courtesy of, believe it or not, a Georgetown Finance professor, James Angel:
“The LRP model doesn’t work,” Angel said. “The idea that when the market is going crazy you can slow down trading in one market and not others means that sell orders were churning through the books at every other market. NYSE dropped out of the running.”
So when the NYSE tries to slow things down, Reg NMS requires orders to be rerouted to other market centers. A commenter on my previous post explained it thusly:
" I believe the core problem here was a real order imbalance with lots of volume (which took us down the original 350), then NYSE halts these stocks, the market orders get rereouted (REG NMS and all) to ECN's where there is much less liquidity, and what liquidity is normally there is mostly provided by the nefarious HF strats who were rightly scaling back risk. So the 10K shares that might have been a downtick on NYSE blows through the BATS book completely."
Market orders can do a lot of damage to thin order books. Folks - if you're using market orders, you had better be aware of the risks. Interestingly, the NYSE, who has been losing market share to other market centers as a result of the ongoing market fragmentation, is positioning their positive spin on the crash, basically blaming the other market centers for failing to stem the cascade.
Paul Kedrosky asks aloud: "why are we wiping out all the errant trades by runaway algorithms and market battle bots?"
David Merkel points out, emphasis mine: "NASDAQ should not have canceled the trades. It ruins the incentives of market actors during a panic. Set your programs so that they don’t so stupid things. Don’t give them the idea that if they do something really stupid, there will be a do-over."
And the Law of Unintended Consequences rears its ugly head again. Merkel's point is simple and accurate: if buyers who step in later see their trades canceled, it removes all incentive for them to step in - and then you don't get the bounce back that we saw! Think about how much havoc it causes a trader who astutely bought cheap stock, then sold it out at a profit. He's now short! Or, he spent the entire day wondering if his order would be canceled, in a state of limbo. What's the alternative - that traders should just assume that the orders will get canceled, and NOT buy stock? Guess what - if no one buys, the stock stays cheap! SOMEONE has to buy, and that someone shouldn't be penalized in favor of remedying the ignorance of the seller who screwed up.
I commented on both Kedrosky's and Merkel's blogs that the goal wasn't to protect the battle bots, it was to protect the retail investors who screwed up because they used the wrong order type (although that doesn't make it right - as I mentioned in yesterday's post too). Again, we see the "bailout" theme - bailing out those who screwed up and preventing them from bearing the consequences of their bad decisions. In case it's not clear, of course I don't think that algos gone wild should be bailed out either via canceled trades.
If anyone wants to defend the decision to cancel the trades, I'm all ears - but your argument needs to be better than "HEY ITS NOT FAIR THE COMPUTERS RIPPED ME OFF AND MY STOP ORDER GOT EXECUTED AT A PENNY WTF OMG *$XYS !^^!@&!*#" If you don't understand that this can happen with a stop order, don't use stop orders.
-KD
38 comments:
Cancelling trades is akin to redoing mortgages. Someone has to be a victim to a greedy brokerage firm. We must protect them!
Such bullshit. As you point out, if you do not understand the trade you entered, you should not be entering the trade.
I cringe at the thought of some congressman- why am I thinking Schumer- coming forward today or this weekend demanding that Congress investigate what happened.
I agree with your point, but I do have a technical quibble:
Guess what - if no one buys, the stock stays cheap!
Strictly speaking, this is not necessarily true... Because "if no one sells, the stock will get expensive". The bid/ask can move on literally zero volume.
Doesn't change your point, though. They are cancelling the trades because the wrong people lost money, plain and simple. It's borderline criminal.
Do hf for a living - had a few thoughts:
First, as per "bailing out the algos" - thank you for being a rational voice. From everyone I have talked to yesterday (several shops), they did ok and are not the ones who need cancelled trades. If there are some shops/algos who are just blindly sending market orders and lost tons of money, I'd agree that they don't deserve to get them cancelled (you're own stupidity cost you money - who would have guessed that could happen).
Two, at least for me, busted trades are a massive pain in the ass and I'd really prefer they didn't exist. Imagine trading around a position you think you have (you're long and you sell to close out), only to later have one leg cancelled. Especially if you're running anything cross asset (or for example hedging an option). I'd prefer they didn't exist for that reason - and put things in place to actively filter against trades that have a greater probability of being busted.
Now those being said, I can see only one good reason (and its not even that good) for cancelling - retail traders. If you're a pro and you have a stop market order or just blindly send size, you deserve what you get. But, I think there's a good amount of retail that use stop market orders or market orders in general not knowing the implications or risks. It is appealing to say "well, that's you're own damn fault and you deserve what you get," but I think you have to protect them somewhat. I know this isn't a great argument, but there seems to be a general view that the market is rigged against retail already, and allowing someone to lose close to 100% on a stock because of a market structure issue would only further that view.
Also, it feels like screwing people on a technicality (I don't honestly believe anyone meant to trade $0.01 in accenture) and I'd prefer that transactions are made in good faith and if the possibility of busting a trade didn't exist, I could easily imagine games being played to suss out people stupid enough to use market orders - the possibility of busted trades prevent this, and I tend to think of that as a good thing (despite what zerohedge says, not all algos are not blindly predatory just looking for ways to game the system).
So - to sum up - I wish they wouldn't cancel trades, but I understand it in the case of wanting to protect retail market orders. Also, as a larger point - they should have some fixed rules for when trades will be busted - the fact that its somewhat discretionary just makes things worse.
nice comment. here's something simpler: we have index circuit breakers. how hard would it be to have individual stock circuit breakers? why not PAUSE trading if a stock falls 20%? i guess it's CONCEIVABLE to make the argument that such a pause would penalize the first responders - but i don't think that argument holds much water, and that such 20% trading pauses couldn't do much harm while they could do much good...
That might help things. Additionally, it might help if they had a similar rule that just disallowed non-limit price taking orders once the price on any stock has moved > some percentage. No one could claim they got screwed then because the only way you could trade some ridiculous price is if you (or your algo) entered the price - in which case it's your own fault and you deserve what you get.
Hi Kid,
isn't there a no-bust range for error trade?
if you execute an error trade that falls outside the no-bust range, your trade gets busted.
but in this case, they are saying that there are no error trades.
so then why are they busting the trades? it just doesn't make sense and seems capricious too me.
I have to side with David Merkel on this.
fundamentalist - i agree with merkel!
we need another tea party.
how can they changed the rules mid stream?
mad as hell and ain't going to take it no more
They shouldn't bust the trades completely, but set a lower bound on the price. Any trades that were executed 50% below the market price at 2:40 are remarked at that 50% price. It protects retail from selling at 1 penny, and those who stepped in and executed at those low levels are still sitting on nice profits.
KD,
Got it. But even if they are retail orders, I would still say don't cancel the trades. I almost never use market orders, and advise my readers not to as well.
But I see your point. It could have been jumpy retail.
David
david - to clarify, i don't think they should be canceled even if they are retail. but i'm guessing that the Authorities may have a different view on that. I also think Adam's comment above is perfectly reasonable.
If T.S. Eliot wrote today, he might have added:
Between the bid
And the ask
Falls the Shadow
Fundamentally, doesn't every buy/sell transaction run the risk that there will be no interest on the other side in a given time & place? We focus so much on the "How much?" that we forget that, even prior to that, is a "Whether?"
IOW, there is PROFOUND RISK inherent in investing, has to be. And I'm sorry but I just can't shake this nagging sense that, as the NYSE Euronext CEO admitted, this was a product of thin volume. If you don't have 22 guys to field a football game, for chrissakes please don't sell me a Monday Night Football game of 9 v 8 and just have do-overs to pretend what things WOULD HAVE BEEN if the defense had 2 corners instead of one.
/rant over
> if buyers who step in later see their trades canceled, it removes all incentive for them to step in
Can't agree more with most of the comments here, which thankfully are of excellent quality.
I write HF strategies and trade my own capital, and will be putting in trade filters to attempt to not profit from trades that could be busted. This is a bit sad, since the system would work much better if I could instead focus on writing strategies to profit from a lack of bids in quality names, and write a stategy that would add liquidity during such periods in the future.
Alas, instead of the system working as it should and me spending my time trying to profit from similar episodes (which would also contribute to limiting the speed or size of a fall), I will now not make what look like good stat-arb trades to prevent trade busts (leaving me with unwanted positions overnight), further exacerbating this situation, and making a repeat more likely.
I personally have no issue with NYSE wanting to run auctions and pause trading on their listed issues, but just call a halt on those issues so that reg NMS doesn't force minimal market orders on the sell side to crash the market.
NYSE may have a point about their model, but in the context of reg NMS, they can't stop trading and not halt trading at other market centers. This falls squarely at the feet of NYSE in my opinion. They screwed up.
-PeterPeter
A question that may be relevant:
I'm a retail trader who's never used stop-loss. But I just went to check whether my platform has stop-limit orders. (It does.)
Are you sure that all retail trading platforms have stop-limit orders as an option?
I agree with Merkel's point about rewarding those who step in during a panic, but there's a difference between buying at depressed levels and buying a $44 stock for a penny and then watching it bounce back to $44 an hour later.
I was broadly short yesterday, but I have sympathy for those who have stop-loss orders that were triggered. Most folks who are semi-active traders can't watch the market 24-7. While I may know that a stop-loss order might be triggered well beyond the place I entered it, if you're a semi-active trader who cannot attend to one's account at all times, it's terrifying to know that your only options are 1) risking precipitous losses because you have no stop in place 2) risking precipitous losses because your position could get liquidated at a price that bears no relation to the value of the security.
What happened yesterday strains faith in the market. And, if you're someone whose stop-loss was triggered at 39 in PG or at a penny in smaller names, I think you'd properly feel as if you were robbed.
Yesterday was truly extraordinary. I think it's a great idea to have breakers on individual names. But until such time as there are, I think the policy consideration for busting trades that bear no relation to the market's current valuation - remember we're talking trades that were more than 60% off in a matter of minutes - is perfectly appropriate.
Anon about stop limit orders: i'm NOT sure that everyone has that as an option, but that's no excuse for using a stop order and complaining about fills afterward! my advice: DON"T USE STOP ORDERS then! DON"T USE MARKET ORDERS!
I find it incredible that PeterPeter is the second person to comment on my blog today that says they are already re-writing their algos to try to NOT buy on prints that might eventually get canceled. that is AMAZING to me - the speed of the adaptation.
i'm guessing more and more that most HFT guys did NOT get carried out yesterday, and even if some did - why are people complaining? (other than the trades getting busted! of course!) isn't it a great thing if you have stupid crazy computers doing stupid crazy things in your market? it makes it so much easier for us intelligent, reasonable, humans! I put in a bunch of discount bids today in a variety of stocks. i'm PRAYING they screw it up again so I can scoop some cheap stock - but I doubt it will happen.
but Kieran, the options you list are a gross mischaracterization. If you're a semi active trader, then you need to know wtf you are doing. You have ZERO excuses if you get plugged on a stop order. NONE. If you're going to play the game, you're going to get burned if you don't play it right.
the option you didn't mention is that traders didn't have to do anything yesterday - no one is ever forced to sell at the wrong price! if you were long ACN, you could have gone to buy your mom a mother's day card, come back, and seen that the stock made a round trip from $40 to near zero and back - that doesn't hurt you! It doesn't cost you anything!
> I find it incredible that
> PeterPeter is the second person
> to comment on my blog today that
> says they are already re-writing
> their algos to try to NOT buy on
> prints that might eventually get
> canceled. that is AMAZING to me -
> the speed of the adaptation.
Well, the risk which was illuminating yesterday is too great to just let it ride. I imagine that there are a few funds that died yesterday through broken trades...
If you run a strategy that is largely market neutral or holding positions for short periods of time, the realization that what seem like the best trading opportunities in the world could land you with instead massive losses due to broken trades on one side of what was a hedged or closed out position, is just too important to ignore.
I did ok yesterday - with my software shutting down at 2:41PM (when in doubt, I have my software automatically shutdown to play it safe, and there were lots of red flags going off by 2:41PM). But, had I continued trading and tried to buy at the bottom, I would have been wiped out this morning from broken trades, since I would have closed out my positions well before they peaked, and therefore would have had massive massive (put me out of business style) losses, from waking up with short positions that were really closing out of long positions established at the bottom.
I think *everyone* who does any kind of stat-arb, momentum or order anticipation strategy with even the slightest hint of risk controls in place will be writing this kind of software if it isn't already in place. The risk is far too great to ignore.
But, if everyone is writing the same style risk controls to avoid having positions which could get partially broken, then the most volatile markets will in the future become even more volatile, as everyone stays out of the pond after sharp moves down.
Yesterday's move by Nasdaq I think will meaningfully change the future of market movements during a crash. I have sympathy for the dumb money investors and speculators who used market and stop orders and sold out to non-existent bids, but I think the reaction from both liquidity providers and drainers with automated strategies to yesterday's trade busts will make future episodes much much worse, as there is great financial risk in trying to bottom fish if your strategy is one that hedges positions.
And the problem is made harder by the fact that noone really knows when trades will get busted.
So, going forward, the long term effect of yesterday's busts will be that once spreads widden beyond a certain point, some participants will start pulling out of the markets, further widening spreads, making others pull out of the markets etc. etc. etc. The reaction from Nasdaq while understandable is going to make this happen again, but with less selling setting it off, and less of a chance of a quick snap back rebound. In short, the markets on a risk adjusted basis are worth much much less today in my opinion, since there is now a new material downside risk.
-PeterPeter
@Kieran:
First, I get your point re: $44 and .01 and valuation. IMO that's the best argument for canceling: orderly markets broke down because reasonable price discovery temporarily broke down. At no point was equity XYZ at a FMV of .01 and thus a segment of the tape needs to be edited out as erroneous.
Except it wasn't erroneous. It was, as far as we now know, a completely accurate reflection of the market for those equities at that time and place. It's not the algos per se, it's the concentration of volume under a limited set of algos which resulted in illiquidity. See Paul Kedrosky today on this point. But as a technical matter, the fascinating/terrifying aspect of yesterday is that pricing was correct!
I think this is a case where a little bit of paternalism is reasonable. Imagine someone new to trading, having just read Barry Ritholz's piece about the importance of stop-loss orders the other day. Before reading Barry's piece, he had no idea what a stop-loss order is. That person is still responsible for knowing that a sale can be triggered well below the stop-loss point, even though they will probably be surprised to discover that it is true. But the folks who run markets have an interest in not losing the customer for life because the stop-loss was triggered at less than 40% of the stop-loss point. That's simply too extreme to expect.
There's plenty of precedent for this. If you screw up drafting a contract to buy a house by leaving a zero off the price, the contract will likely be unenforceable. A person should know wtf they are doing when draft a contract to purchase a house. But sometimes people screw up, and sometimes it's in everyone's interest not to enforce that contract. In the same sense, when you purchase a security at a price that more closely falls under the definition of an error than a cheap investment/trade, it is not irrational to find that the purchase would be invalidated by the exchange.
Exchanges should use extreme caution when applying this logic, but yesterday's events qualify.
KD,
I enjoy your blog, but I think on this particular issue, your position and that of some of those who comment shows the serious disconnect between Wall St and Main St and why the majority of people look at Wall St with such contempt.
You can argue against the virtues and risks of stop loss orders all you want, but the notion that an individual could get busted and lose hundreds, thousands or even hundred of thousands on a stop loss on PG at 40 only to later find out it wasn't real and was only triggered by a computer error or glitch is horrifying.
You seem to be saying to that person "tough sh*t", that's what happens when you use a stop loss, but the stock never legitimately traded anywhere close to the price the order executed at and said individual ends up losing their shirt only to find out later it was a computer error, yet you are blaming the individual for having the stop loss?
Transor @ 3:07: i very much agree, but you have to realize that there is a big philosophical component of that argument that the vast majority of people will never understand or agree with.
Ken - the disconnect between Wall Street and Main Street is very real, and it's a result of the fact that Main Street is widely ignorant. Read the comments in this thread - the point is that the trades are not "Erroneous" - they are what they are. PG sold off BECAUSE of the idiots with stop orders! THEY ARE (at least in part) THE CAUSE!
If there are crazy computers doing stupid things - that's BETTER for all the humans in the market. If you're a human who puts in dumb orders - and market orders including stop orders are DUMB orders - you will pay the price for your lack of sophistication. sorry. that's how it is. Everyone needs to be smarter, or not complain when they lose. It's just like the housing market, in a way.
your claim that the stock never legitimately traded there is flat out wrong. see Transor Z's comment above. Price is the intersection of supply and demand. At THAT point in time, supply and demand balanced out at $40 in PG. NO ONE WAS FORCED TO TRADE AT $$40 in PG - at that time, or at any other time.
we should work more on education people about markets, and less on treating them as "victims" who need reimbursement
Keiran - thank you for the logical, well spoken comments. I fully understand your view.
I guess I don't understand why if the trade is broken on the buy side why the person is now short? If I bought AAPL at $199 yesterday and didn't close the position and whomever comes along and invalidates the trade then it's as if it never happened. Yes? But what if I had that hero order in and it got executed and with the stock back at $240 by close I decide hurray for me I sell and take my 40 points? Now they invalidate the buy and I'm short x number of shares at $240. I would argue that they would have to cancel the sell as well because the sell would never have occurred if not for the buy. I would think it grossly unfair to zero out the one and not the other. I am obviously missing something. Help me out here.
Daniel - i can't help you out, because that's exactly the problem. and algo guys will tell you that if they did manage to buy stock cheap, it's almost certain that they sold it before it regained all the losses too - so if their buy is canceled, they now have a loss on their sell.
Oh, I get it now. They buy at $199, sell at $210 and the stock is at $240. The buy is canceled and now they are short at $210 with the stock at $240.
"If there are crazy computers doing stupid things - that's BETTER for all the humans in the market."
Thank you so much for this - I'm so sick of people complaining about computers making things act irrationally. I may do algos at work, but in my personal account I pray that algos or other people will act irrationally - and the more irrationally they act the bigger potential I have to make money (both long and short term). It feels like a good portion of the people arguing against program trading either don't get this, or think that a market is only worth being in if vol is predictably always within some range.
If you're really investing for the long run, you should love things like this (assuming your trades don't get broken - but that's a whole other issue).
Adam> Thank you so much for this -
Adam> I'm so sick of people
Adam> complaining about computers
Adam> making things act
Adam> irrationally
I am in full agreement. One point however.
As one of the many stat-arb funds that exited the market yesterday (at 2:41PM), I realized that there was in fact something going on with the plumbing and put away my HFT hat and logged into my personal trading account at Fidelity (where I trade perhaps 20 times per year and don't care about their high $8/trade fees, and hold positions for sometimes years).
I wanted to buy GIM, FAX and FCO (all closed end income funds with international exposure that traded down well below fair value) - but due to system load (I presume), it was literally after 3PM before I was able to enter an order, and then it would literally take minutes to edit the orders.
I suspect that there were other trading accounts that retail investors tried to use with similar results (or lack of), and I think trying to get a retail broker on the phone would have been beyond futile.
So, in this instance, many people couldn't profit from the machines going crazy (which I don't think is an accurate representation of the events - but regardless) - and were only able to sit back and nervously watch huge unrealized losses build, or if they had stop losses - have incredibly large realized losses (perhaps unwound subsequently by the exchanges).
-PeterPeter
Thanks for spotlighting my comment as a red herring; I thought I was kidding!
The reasoning for calling the trades at next to nothing erroneous, could be that price the sellers got was not really the market price, the market was partly shut down and the trades were diverted to where there really was not a market.
I'd like to point out that stock trades are busted like this quite often. Usually it goes unnoticed by most because it happens in just a handful of low profile symbols and isn't in the context of a huge market decline. I was quite surprised by the 60% threshold which is by far the widest margin I can remember. So this should actually be quite encouraging in terms of holding people responsible for their trading decisions or errors.
Also, there clearly is going to be some level at which trades are busted. So, yes people should be more careful, and yes people should absolutely put more checks in their systems, but this is still going to happen. For example, a few years ago after-hours following an earnings announcement GOOG traded down to some crazy price like $50 on millions of shares. Whoever made the error technically had a loss over a billion dollars. That is a rather extreme scenario, but at some point an error is obvious and the exchanges all have rules granting them discretion in such matters. Clearly, the rules need to be much more well defined such that orders are rejected by the exchange or symbols are automatically halted to prevent these trades from happening in the first place.
There was one clear case of 'algos gone wild' near the close on Thursday that most didn't notice. HHH is an ETF consisting mostly of AMZN, YHOO, EBAY. Starting around 3:55 EST it took off from about $56 and started trading wildly up to $150. This lasted about 3 minutes on volume of several hundred thousand shares. I sold as much as I could and ironically was being slowed down by various safety features in my system. I was trying to sell it below $80 or $90 -- because I knew that anything higher would surely be busted -- and was frustrated because I kept getting filled above $120. However, what I didn't do is start buying up shares of AMZN and the like at any price to complete my 'arbitrage'. That wasn't at all necessary as there was zero risk of losing any money on the HHH sales and thus hedging was not only pointless but also risky because of the very real chance of busts. So what happened? The exchanges ruled to bust all trades over $60.08 -- well under a 10% threshold!!
So, to the retail investors out there -- market orders can get filled at very bad prices so be careful -- that's part of the game and you should know that. To the trading firms out there -- trades can and do get busted so be careful -- that's part of the game and you should know that.
great comments everyone - i take it as a point of pride that my readers, even the ones who disagree with me, make clear, eloquent, intelligent arguments in my comment threads.
basically, i think that if they are going to allow those trades to happen, they shouldn't cancel them. if 1c trades are clearly ludicrous- which they probably ARE! - then they shouldn't be allowed to happen at all! they should be prevented via curbs before they happen. PROACTIVELY.
The new talk seems to be of "intermarket sweep orders", or ISOs (which puts a whole new spin on ISO-compliant ...). I can't help but connect the distribution of liquidity to tight coupling among assets as Bookstaber has elucidated.
But the threshold for trade cancellation seems arbitrary, and will make for some interesting challenges. I think the appropriate name for this episode will be "The Great Recision".
I have almost no sympathy for people that lost money on stop loss orders. They should have known the risk.
What I was afraid of on Thursday was a margin call. I was buying all the way down until 12:52, when I hit my limit of 4x leverage. Because one of the ETFs that I had bought heavily showed a price over 40% too low, on paper, it looked like I had lost a fortune and was now 8x levered. Had a margin call been triggered intraday, I may have been forced to liquidate my portfolio at ridiculous prices. Is there anyone out there that was wiped out because of a margin call?
joshua - that would only be a problem if the ETFs didn't bounce back - but they did... in other words, if you get a margin call, it's not IMMEDIATE - you have at least several hours to make good... at least i THINK you do...
Joshua Kachner said...
I have almost no sympathy for people that lost money on stop loss orders. They should have known the risk.
What I was afraid of on Thursday was a margin call. I was buying all the way down until 12:52, .... Is there anyone out there that was wiped out because of a margin call?
Yes
This is from DavidDT at Trading to Win-- http://trading-to-win.blogspot.com/2010/05/tell-mr-president-to-stop-shell-game-it.html
Now – I took a hit today, big hit, Thinkorswim locked on me completely, my puts options orders (surprise) were executed at the worst possible prices, my long SP500 futures (/ES) trade was liquidated due to INTRADAY 50% margin raise (should I hold till close – I would have made very nice profit instead of hard to swallow loss.
that's a TRADER talking, who is obviously highly levered, trading highly levered products like e-mini SPX futures.
no retail investor is getting margin-called out of his positions intraday in a matter of 45 seconds - which is about how long the abnormal prices lasted. period.
Post a Comment