Yes, speed matters. However, the core HFT alpha can only be realized if you know what order type to send and when to send it. This process is completely dictated by how exchanges implemented REG NMS, Rule 610, order matching engine features, and very specific order types.
In theory, for a single price movement in the U.S. equities marketplace to occur, each exchange would have to show a gap of two or more ticks before a new price would be accepted on any exchange that established a more aggressive price level. With 13 exchanges, we know that HFTs are not waiting for a two-tick-wide national best bid and offer before placing their bets. Understanding this and mastering exceptions to Rule 610 became the name of the game, whether such features are adequately documented or, in many cases, not.
Prior to Rule 610, HFT scalping strategies could lock markets. Rule 610 changed the HFT game in such a significant manner that it can be thought of as a different phase of the HFT algorithmic trading tradition. The ban resulted in HFTs being forced to engage in “spam and cancel” strategies that repeatedly attempted to get to the top of the order queue on a price move. Such strategies would attempt to “step in the middle” to set a new aggressive price. This invariably locked away markets. Rule 610 demanded that such orders not be accepted at the entered price.
This activity caused immense load on exchanges, but in no way did exchanges want to discourage high-volume HFT order flow. To court HFTs, exchanges provided a number of specialized features to assist “spam and cancel” strategies, many of which are still operational today.
A common order matching engine feature that exchanges used to fulfill Rule 610 was to “price slide” the order. This practice modifies the price of an order that locked the market by, in the words of one exchange, ticking the order back in a “convenient” and “sensible” manner. When these orders were slid back, and did not have a high queue placement, the HFTs would first know that there was an order ahead in a better queue position, and second, cancel the order and retry.
While HFTs canceled their slid orders, traditional investor orders would typically just slide without being canceled. This causes the institutional orders to move to the back of the queue and away from the trading action. In this strategy, the HFTs would monopolize the top of the book, interacting with marketable orders, while the institutional-side orders would be at the bottom of the queue only to be executed when a large buyer or seller cleared the book.
To execute these spam-and-cancel strategies even more quickly, HFTs utilized specialized order confirmation information to detect being slid so they could quickly cancel the price-slid order. Exchanges also provided alternative cancel-back or “opt out” options that literally rejected orders that might have otherwise been placed in a disadvantaged queue position.
Many institutions have no idea that their orders are being slid away from the top of the book. In many instances, institutional clients and their brokers are not being adequately informed. In some instances institutions and brokers are not even receiving price-sliding information on exchange confirmations unless they specifically ask for it.
Many people believe that high cancel rates are an attempt to sniff out larger orders; however, many times it is these spam-and-cancel strategies jockeying for top-of-book status that just push institutional orders out of the way, only to trade when the market moves against them.
This is one early example of how seemingly straight-forward and appropriate order types are being leveraged not to take advantage of institutional flow, but to make institutional flow irrelevant to the point where the institutional order is disadvantaged.
This spam-and-cancel order-handling treatment is the precursor of special order types, which, for all practical purposes, exist to assist a specific type of sophisticated trader whose primary intent is to step ahead of all other customers.
In summary, Rule 610, which banned locked markets, dramatically changed microstructure to dictate when HFT scalping strategies were permitted to get to the top of queue. In practice, Rule 610 created an environment in which HFT scalping strategies needed to use a feedback loop of price-sliding information or plain rejects iteratively, sending and canceling large numbers of unexecuted transactions to get an order to “stick” at the best price at the top of an order book. Such strategies were lucrative but burdened exchanges with heavy transaction rates and canceled orders.
In the next article, we will explore how exchanges courted firms engaged in “spam and cancel” strategies by providing them specialized order types and order matching features that provided a more efficient and lucrative means of achieving superior queue position.
See: Locked Markets, Priority and Why HFTs Have an Advantage: Part 2: Hide & Light
For more on Haim Bodek's views on high-frequency trading, check out "For Superfast Stock Traders, a Way to Jump Ahead in Line," from the Wall Street Journal.
Comments | Post a Comment
43 Comments to "Locked Markets, Priority and Why HFTs Have an Advantage: Part I":
crammond1964
11 October 2012
good article and exchanges have denied this pratice ! but painfully obvious
Comments (251)
anonymole
11 October 2012
Thanks for this. Reading up on REG NMS R.611, if I get this right, the proliferation of market venues, as Larry Tabb points out, is exacerbating this "do not lock" directive the SEC enacted and effectively forces this sliding and hiding behavior on the individual exchanges. If BATS would lock DirectEdge which would lock NASDAQ, and so on, then the technique by which generic orders are slid into a price betterment position, to avoid the lock, has cracked open the window of opportunity thereby allowing the abuse of this order repositioning by HFTs which can attack the locked condition with specially treated orders.
Now, this locked condition is such that the spread between bid and offer is reduced to zero or less between markets yes? Which would theoretically enable HFT venue arbitrage then; and it is this condition the SEC was trying to avoid by the implementation of NMS R.611, correct?
Just supposing, if there was only a single exchange per market, then this "lock" condition would never arise thereby eliminating the ability of HFT to benefit from these fleeting locked markets. And so Larry's point about SRO proliferation is compounding the HFT problem - more venue combinations to provide more locked conditions with more books to force sliding orders.
If you or anyone could respond to correct or agree on these points I'm sure we would all appreciate it. Looking forward to your next article.
Comments (83)
Dave Cummings
11 October 2012
Blame the government!
When Reg. NMS was debated, several people very knowledgeable about market structure (including myself) argued against locked, crossed, and trade-through rules because of the side-effects caused by race conditions between fragmented markets.
I was told by the top SEC staffer that drafted the rule that even though it may not be needed it was left in Reg. NMS "in response to congressional pressure".
hbodek
11 October 2012
Dave,
I have thought quite a bit about the pros and cons of allowing locked markets. Though I am not well-versed in the history of the debate, I do think the problem of locked markets should be reconsidered from every angle. With regard to the history, I was told one major concern was investor confusion with locked markets due to investor inexperience with the maker/taker model and the reasons markets could be locked. On the other hand, I agree that a 0 width market has compelling economics for marketable orders. I am hoping we can reignite a discussion on the topic. Can you give us more insight into both sides of the argument prior to 2005?
Haim
Comments (20)
abekohen
11 October 2012
Written in 2009 by BMO (yes, after 2005) is an argument against locked markets:
http://qes.bmocm.com/papers/4_BMO_HiddenCosts.pdf
Comments (43)
arthur curry
11 October 2012
the BMO paper is very interesting. i am all for locked markets...after passive rebates are banned.
sarnuk
11 October 2012
Ban Maker Taker model. Force an order that would lock a market to trade. Eliminate price sliding. Eliminate order types that allow a participant to discriminate against who they trade with (Arca PL Select will allow passive ALO market maker orders to not trade with IOCs or orders bigger than it).Get back to basics. Smash the "pick up the rebate" algo market making module. Realign the market with investors. Stop perverting the most elegant and simple capitalist principal - supply = demand- best bid meets best offer. Eliminate self-trade prevention order modifiers. Everybody pays a fee to trade. No payment for order flow. Eliminate the complexity. Disclose all unknown order types. What is there that we do not know. Simplify. Simplify. Simplify.
Do the above. Please. Then no body is worried if some knucklehead wants to dig into a mountain to shave a microsecond. No need for a tax.
Just eliminate the game and get the markets back to investing.
Comments (160)
sarnuk
11 October 2012
The ironic thing? I would bet that even the founder of Island is shaking his head at how perverted it has all become.
Comments (160)
ltabb
11 October 2012
abe - that is a good article. I recommend that folks read it (only 7 or so pages). To paraphrase it basically says while the theory of a locked market is good - with the rebate model, folks sit on both sides of the market looking to collect rebates on both the bid and offer. While that is what a market maker does, it paper goes on to say that it only happens on cheap stocks with deep books. So basically they are just taking two rebate fees out of the market when there are natural buyers and sellers. Now this was written for the Canadian market when Pure went live.So its a bit dated but interesting none the same.
Comments (303)
Anonymous
12 October 2012
or we could just fix everything at once by moving away from a continuous market and adopting a series of calls and getting rid of time priority and exotic order types.
Why do we want a market structure that even allows things like locked/crossed quotes and race conditions?
why do we want to open ourselves to the systemic risk of a single participant flooding the market with quotes (willingly or otherwise) and impacting everyone else, or several participants potentially shutting the entire market down with DOS attacks? Do we really believe that relying on participants to not blow themselves up makes things any safer? Would it not make more sense for venues to throttle incoming orders rather than asking those accessing the "system" to play nice and not gum up the bandwidth? Is it unheard of to expect those that make up the system (the venues) to ensure stability of the system they make up?
Anonymous
12 October 2012
Is the previous Anonymous just being facetious, sarcastic or ironic? A call market in 2012?
Anonymous
12 October 2012
quite serious. See http://tabbforum.com/opinions/how-to-control-high-frequency-trading
abekohen
12 October 2012
Went home last night and scartched my head a lot trying to remember from my trading days in the late 80s and 90s why locked markets were frowned upon and then I remembered it was manipulation by NASDAQ dealers on index expiration prices. See:
http://www.finra.org/web/groups/industry/@ip/@reg/@notice/documents/notices/p004018.pdf
In those days we would fax a list to a NASDAQ dealer (or worse yet to our firm's OTC desk) of NASDAQ stocks to be bought or sold on the opening or closing print. They had huge incentives to lock the market taking advantage of outsize flows and their own inventory.
Comments (43)
Anonymous
12 October 2012
A call market? That is the most idiotic and unimplentable idea yet conceived. What next? A return to the thieving NYSE and AMEX specialists? Pay 2 million dollars for membership which is effectively a license to steal? Anybody can become an HFT. What prevents you?
Dave Cummings
12 October 2012
The big wholesalers like Knight benefit the wider the bid/ask spread is because they "internalize" at the NBBO. Wholesalers are the ones fighting locked markets. Investors benefit when they can buy or sell with zero spread.
Anonymous
12 October 2012
*** Comment Removed due to personal nature ***
Thanks
TabbFORUM Editor
sarnuk
12 October 2012
"Investors benefit when they can buy or sell for zero spread" Investors buy and sell at a price that they choose. Investors buy who buy at 10 cents did not save money because it was offered there and bidside was 9 cents any more than if bidside was at 3 cents. Spread is a nonsensical notion in modern order driven markets. Moreover, ask the listing firms like KO or PG or AMAT if they want to see locked markets. Do you think they want to see buyers post bids using ALO and sellers offer at the same price using ALO, and no trade take place? Is that liquidity? Is that capital and risk transfer? I do concede that video gamers who love to perfect level 3 of the pick up the rebate game would like that. Modern exchanges too, sadly. These arguments are surreal.
Comments (160)
hbodek
12 October 2012
I am sympathetic to Dave's comments on internalization and zero spread economics, though we likely differ on the best mechanisms to address each issue.
I support the idea of putting additional constraints on off-exchange internalization. Internalization practices in the stock market have contributed to the toxicity of "lit" exchanges, the of cost of which is indirectly passed to institutional investors through adverse selection. I'd like to see internalized orders subjected to aggressive competition on the "lit" exchanges, as is currently required in electronic options markets.
With regard to eliminating the ban on locked markets, the specific implementation details would still raise many of the order precedence and complexity issues that exist today (especially with regard to the interaction of post only vs. taker fee eligible orders). Though the zero spread economics of a locked market are arguable for simple market orders, the order handling procedures would need to be scrutinized from every angle for passive order posted by institutions. Alternatively, price improvement auctions also provide zero spread economics, a mechanism which if mandated for internalized order flow could also address the exchange toxicity problem.
Comments (20)
sarnuk
12 October 2012
I apologize for my comment: "these arguments are surreal". Bad form. I am happy we are all debating these issues.
Comments (160)
pskopp
12 October 2012
Sal> "Investors benefit when they can buy or sell for zero spread" Investors buy and sell at a price that they choose. Investors buy who buy at 10 cents did not save money because it was offered there and bidside was 9 cents any more than if bidside was at 3 cents.
The seller using a market order in your example I imagine would disagree.
Obviously there is an economic benefit to tight spreads in any trading environment where even a tiny fraction of the participants are using market orders.
> Do you think they want to see buyers post bids using ALO and sellers offer at the same price using ALO, and no trade take place? Is that liquidity?
What's wrong with that? If there are bids and offers at the same price, then even without a single trade going off, you know the market value of the equity is bound by the bid minus taker and reg fees up to the ask plus taker and reg fees. That in the most common case would put the market value in a range of under a penny, with posted orders on both sides of the book. If any ilsting firm took issue with that, I would contend they either dont understand equity markets, or they are likely in the business and are hurt by narrow spreads...
Comments (28)
jwells
12 October 2012
As a private investor, I like the idea of allowing crossed markets and zero spreads.
When I was a bank trader in the UK, admittedly many years ago, we used to see crossed markets from time to time. It was an opportunity for those that could take advantage and I see nothing wrong with that. When I came to the US, I was amazed at the extraordinary number of rules governing the markets. I thought that America was the citadel of the free market but it isn't by a long shot.
Having said that, over time I have developed an appreciation for the US system, especially the core idea of NMS which I think should be adopted in Europe ie centrally mandated consolidated feeds.
Reg NMS 2005 took logical steps by recognizing the power of technology and the internet. It formalized the new era of competing electronic marketplaces but kept some of the old practices.
Presumably, now that we have the transparency of electronic markets and more experience, we need to throw away many old rules and conventions and keep make improvements to the new system. Can Haim find some more rules that should be consigned to the scrap heap? I hope so. I am looking forward to the next tranche of this discussion.
Comments (11)
abekohen
15 October 2012
For me it boils down to having an orderly market while balancing competing claims to fairness. Each class, be it mom & pop investors, institutional investors, hedgers, HFTers, brokers, exchanges ... has a different concept of what is fair. It is impossible to satisfy all of them simultaneously. So perhaps the iterative regulatory process is best when it minimizes damage to the different constituent groups.
Comments (43)
crammond1964
15 October 2012
by ignoring "abuse " the problems soon escalate ; all parties have to stay within the required boundaries that the regulators have set up ; if not we have the chaos that we are in .
Comments (251)
zhuh
15 October 2012
This is a really interesting discussion. I still have a few questions about it, and wonder if anyone could shed light on them.
1. Is it a correct that exchanges slide the investor orders and do not automatically restore the original limit price? For example, say the current best offer is $20, and a limit buy order comes in at $20. To avoid a locked market, the exchange slides the buy order to $19.99. A moment later, the best offer becomes $20.01. In this case, does the exchange automatically slide the buy order back to $20? If not, then I can understand how HFT may jump to the head of the queue at $20 on the bid side. But if the original price of $20 is restored automatically, it's not clear how HFT can jump the queue.
2. Are exchanges allowed to slide an order, or does that require explicit consent from the investor who sends the order? Sliding a marketable limit order means a delay in trade, and that can be costly to the investor.
3. Allowing locked markets seems unlikely to solve the problem raised in this article. For example, say the best offer is $20 and it's now allowed to submit limit buy order at $20. In this case, what happens to a buy limit order priced at $20.01? If that order is slid, then the same problem raised in this article repeats.
I'm trying to understand this interesting issue (and I have no commercial interest to take either side in this debate). Thanks!
Comments (2)
hbodek
15 October 2012
Hi Zhuh,
Answers to your questions below:
1. Price sliding differs per exchange and there have been many modifications over the years. NASDAQ is probably the best example of an exchange where "spam and cancel" strategies have thrived and where certain price-slid orders are ticked back to the end of the queue permanently (default behavior of OUCH). As you note, on other exchanges price-slid orders are commonly ticked forward when the market "unlocks". Queue jumping and forward tick price sliding were not addressed in this article.
2. Smart order routers commonly avoid using limit orders subject to routing, instead using non-routable orders subject to price-sliding and rerouting at their own discretion. Exchanges permit a user to "opt out" of price-sliding if desired.
3. Correct, an aggressively priced limit order would most likely be repriced to the locking price, and there would still be aggressive competition for queue position on price moves. However, some of the race conditions associated with Rule 610 would no longer apply. I am still thinking through both sides of the argument myself!
Haim
Comments (20)
zhuh
15 October 2012
Haim: Thanks a lot for your response. It's interesting to know that exchanges differ in where they put "slid" orders in the queue. A natural next question is to what extent investors know about these practices, especially if they need to "opt out", rather than "opt in", to have their orders slid. I'm looking forward to your next article in this series. --Haoxiang
Comments (2)
lkovach
15 October 2012
Look for Part 2 Tuesday: To attract high-frequency traders, exchanges created special order types that would “hide and light” enabling HFTs to post orders that locked away markets despite the ban imposed by Rule 610. Meanwhile, institutional investors became further disenfranchised by an HFT-oriented market structure.
Comments (31)
lkovach
16 October 2012
Is the SEC finally stepping in? From the WSJ: SEC Keeps Wary Eye on Exchanges
Comments (31)
anonymole
16 October 2012
"I just let the computers get in the ring and bang each other around,"
http://online.wsj.com/article/SB10000872396390444657804578053153939092668.html
Comments (83)
kurtkujawa
18 October 2012
get rid of all the for profit exchanges and the fragmentation and you wouldn't have to worry about locked or crossed markets. As sarnuk said, it is pretty simple, get back to the basics. The problem over the years is one squeeky wheel wants to change the structure to benefit him, they make a change , then someone else complains and they make another. this keeps going on and on and on. Then they try and fix the problems with "bandaid solutions" only to complicate matters even more. 1 market where all orders go. No rebates.
Comments (171)
abekohen
18 October 2012
kurtkujawa, I guess you weren't trading when the NYSE specialists would rip you off for 25 cents a share, or the AMEX specialists who would hold your order for 5 minutes, or the NASDAQ market-makers who colluded on fixing minimum spreads. We have it now so much better and people are crying over fractions of a penny. Wow!
Comments (43)
kurtkujawa
18 October 2012
actually i have been trading since the 80's. Now don't misunderstand me, I am not saying that the NYSE or NASDAQ or the AMEX didn't have there fair share of thievs and problems, but at least there was depth to the markets and you could take an offer or hit a bid for what was displayed! Also, just because the spread was an 1/8 or a 1/4, didn't mean that where every share traded, that is just where the price was poated for size. I b/s millions of shares between the spreads. So i can tell you that you were either niave or a bad trader if you got ripped off for .25 a share when you traded in NY. Just because the spreads are tighter, doesn't mean your getting better prices and arent getting ripped off by all the front running HFT's costing you more than .25. It was much better when you could get a 20 to 20.25, 50k up market instead of all tis 100 share crap and fragmentation. Tell me how does that help anybody other then the HFT's, and the TCA salesmen that scare to many institutions into there suite of algo's.
Comments (171)
abekohen
18 October 2012
perhaps I was a bad trader, or perhaps I monitored in real-time all 500 stocks of the S&P500 I was buying, or the 800-1200 NYSE stocks I traded against NYA futures, and I was attuned to my slippage. What you saw on the screen was certainly now what you got, especially if it was a Spear, Leeds (SLK) specialist. Oh, here's comes a Super-DOT order, let's do a buddy trade in front of it. Oh, you have a mkt order for 5000 shares of PFE when the market is 3/8 - 5/8? Let's trade 5/8 and 3/4 ahead of you and the fill you at 7/8. Never mind a Sell-Short order! Limit? Nothing Done? Market? Let's take it down 3/8 before filling you.
Comments (43)
abekohen
18 October 2012
Rule of thumb for NYSE specialists was 80% of specialists are honest 80% of the time. AMEX? 0!
Comments (43)
kurtkujawa
18 October 2012
I agree on the 0% on the curb, maybe even the 80% rule, but i will still argue that all the fragmentation and 100 lot markets have increased the volatility and gaming. I for one, like to at least know who i am going to war against!
Comments (171)
abekohen
18 October 2012
I would rather lose 2 cents on 10,000 than 1 bit (an 1/8) on 10,000. That's just what my math teachers taught me.
Comments (43)
kurtkujawa
18 October 2012
id rather negotiate with a natural contra, trade 100k at a mutualy agreed upon price and shut the book out, but thats just me.
Comments (171)
kurtkujawa
18 October 2012
and not loose anything
Comments (171)
Anonymous
18 October 2012
most would - but even in it's heyday Posit never had more than 1% market share.....so while a pool like Posit Alert is a great starting point, you need a back up plan in the event contra liquidity doesn't exist in the pool. If you think allowing a broker to shop the order isn't rife with at least as many issues as trading downstairs then you need think again. At the end of the day, it is worth noting that trading impact costs are far lower today than they were 5 or 10 years ago. So while the noise may annoy - and needs to be addressed - the actually act of trading has become more efficient....no matter how many books you read saying otherwise.
kurtkujawa
18 October 2012
I disagree, ive beeen doing this for 20+ years, both on the sell side as a market maker and on the buy side. There is WAY more leakage now with all the dfferent venues. Just as before, you can make a bad call, but at least you new who it was and you could controll that the next time. Now with all the fragmentation and anonimity, trading on the exchanges is a toxic wasteland. Unfortunately it is the wourld we live in. To me the best case scenario would be to take the best of both worlds. 1 exchange or pipe or CLOB, call it what you like, where all orders flow so that there would be no exchange arbotrage or locked markets. Get rid of the rebates, this would result in deeper and more transparent markets while keeping spreads tighter.
Comments (171)
anonymole
18 October 2012
To truly reform Wall Street regulators and Congress must simplify. Look at how convoluted the tax code is and how this obscurity of implementation drives loopholes and malfeasance. SEC regulations are headed down this same crooked road. Dodd-Frank will be no different. One need only look to the U.S. Constitution for a model of purity of intent.
• Complexity breeds transgression.
• Simplicity breeds compliance.
• Complexity promotes concealment.
• Simplicity promotes transparency.
The answer to Wall Street reform is easy - simplify.
Comments (83)
kurtkujawa
18 October 2012
Agreed!!!!!
Comments (171)
lkovach
18 October 2012
Glad we can agree on something -- that's a start. If you haven't seen it yet, read part 2 of Haim's commentary: Hide & Light.
Comments (31)