You have been granted access to this page through First Click Free. Subsequent use of TabbFORUM will require logging in. If you don't have an account, registration is free.

Videos

  • Rail_thumb_merrin

    Is the Market Rigged? Retail Investors Think So: Merrin

    Whether the market is rigged is not the important question, says Liquidnet CEO Seth Merrin, who notes that the market structure debate is old news within the industry. The real question, he adds, is: Do retail ...
     
  • Rail_thumb_michael-lewis

    Michael Lewis Speaks With Malcolm Gladwell

    Following the release of "Flash Boys: A Wall Street Revolt," Michael Lewis spoke with Malcolm Gladwell at a Live Talks Los Angeles event held on April 10, 2014. ...
     
  • Rail_thumb_ronan

    IEX's Ronan Ryan Stirs 'Flash Boys' Debate

    Ronan Ryan, one of the central characters from Michael Lewis's popular book 'Flash Boys,' joined TradeTech in Paris via Skype connection. Richard Balarkas, CEO of Quendon Consulting, ...
     
 

More Video | Podcasts

Advertisement
Author_haim_linkedin
Haim Bodek

Decimus Capital Markets, LLC

More From
Haim Bodek

Spotlight-black
The Great HFT Debate

31 December 2013

2013 TOP STORIES: HFT Checkmate – The Alpha in Order Types

Despite REG NMS’s aim to strengthen the market structure and protect long-term investors, the explosion of special order types helped high-frequency traders win the game before most of the other players even understood the rules. Here are five order matching engine practices that served to preference HFTs over public investors.

This article originally was published Jan. 21, 2013.

The following is an excerpt from Haim Bodek’s ebook, “The Problem of HFT -- Collected Writings on High-Frequency Trading & Stock Market Structure Reform,” available on Amazon. A paper-based version is available here.

Over a good part of the past decade, special order types often were one of the primary topics of discussion between HFT firms and exchanges. Special order types became essential in assisting HFTs’ exploitation of the REG NMS regulatory framework introduced in 2007, which ironically was intended to strengthen the national market system. Exchanges introduced “innovations” such as preferential price-sliding and orders that “hide and light,” which assisted HFTs in circumventing impediments introduced by REG NMS that might have otherwise constrained the growth of HFT.

Such features in themselves might have served to benefit institutional investors if there had not been dramatic differences in order-handling treatment among different classes of order types, practices that created deep asymmetries between HFT-oriented order types and order types commonly employed by institutional investors. Instead, HFTs and exchanges undermined the very body of regulation meant to protect long-term investors from inferior executions in the highly fragmented US equities marketplace. Currently, the bulk of modern HFT volumes are executed with HFT-oriented special order types, accounting for a significant portion of the total US equity market volume.

In light of the growing significance of special order types since 2007, their absence from industry dialogue before 2012 is noteworthy. In fact, in the hundreds of articles, interviews, conference talks, and academic papers that addressed the topic of HFT, there was virtually no mention of the role of special order types in modern HFT until the first quarter of 2012, when regulatory scrutiny of HFT-oriented special order types became public knowledge (SEC Puts Exchanges on Notice Over Computer-Driven Trades,” Bloomberg News, April 4, 2012).

Over the period from 2007 to 2012, while an unprecedented flurry of new HFT-oriented special order types were either introduced into the marketplace or revised and updated with new features, scarcely a word was spoken about these developments in the unending HFT debate. There was, of course, no mention in industry dialogue of the importance to HFT strategies of exploiting unfair and discriminatory features embedded in special order types, features that gave HFTs preferential treatment over the public customers (and that were exploited by HFTs in combination with their speed advantage).

More specifically, these innovations resulted in a number of order matching engine practices that served to preference HFTs over the public investor:

  • Unfair order handling practices that permit HFTs to step ahead of investor orders in violation of price-time priority.
  • Unfair rebooking and repositioning of investor orders that permit HFTs to flip out of toxic trades.
  • Unfair conversion of investor orders eligible for maker rebates into unfavorable executions incurring taker fees.
  • Unfair insertion of HFT intermediaries in between legitimate customer-to-customer matching.
  • Unfair and discriminatory order handling of investor orders during sudden price movements.

These order matching engine practices either currently exist or have existed on nearly every major electronic exchange. The tendency for each electronic exchange to match feature for feature with rival exchanges resulted in similar functionality across market venues. Furthermore, although the specific practices noted above are well understood by HFTs and exchanges, being central to the “guaranteed economics” arrangement afforded HFTs, these order matching engine practices are, for the most part, undocumented.

The order matching engine abuses noted above represent primary alphas that assisted HFTs in growing their volumes to dominate more than half of US equity volume, features that have made traditional order types impossible to use without being subjected to order matching engine abuses.The total economic impact and the bulk of HFT edge-capture are tied to these features, in conjunction with the corresponding special order types that exploit the matching engine (order types that are noted in a later section).

For the most part, the rest of the market was focused on the wrong problem: speed. Most participants (including myself) who struggled to execute effectively on HFT-oriented exchanges assumed they were using the exchange correctly via the traditional vanilla professional order types. Most market participants just thought that they weren’t fast enough to compete against HFTs.

But HFTs Were Already Playing a New Game ...

Metaphorically speaking, if the market was a chess board, HFTs were using queens, rooks, knights, and bishops. In contrast, traditional investors weren’t even using pawns – they were using checkers pieces.

To take the chess metaphor further, most non-insider players didn’t realize that the checkers pieces they were using had no chance against any of the chess pieces, or perhaps didn’t realize that chess pieces existed at all.

For the most part, no one thought that the problem was that the game had changed and you couldn’t play checkers anymore. Many reputable and quite sophisticated firms that invested in the technological arms race were in essence simply trying to play checkers faster. But if you didn’t use the right pieces and play the right game, you were going to lose. You had to learn chess, and use the new pieces. There was no way around it. HFT was all about playing the right game with the right pieces -- and playing it very well against the public customer who was effectively playing the wrong game with the wrong pieces.

For five long years, the greater investment community was subjected to unnecessary transaction costs as they transferred “guaranteed economics” to HFT firms and exchanges through mechanisms unbeknownst to them. As long as HFT practices remained opaque, it was so much easier to believe HFTs were simply faster than it was to admit HFTs thrived through artificial means.

For more details on order types and how they have created an advantage for high-frequency trading firms, read Haim Bodek’s four-part series on TabbFORUM.

Spotlight-white-trans For more stories in the The Great HFT Debate Spotlight Series click here.

Comments | Post a Comment

47 Comments to "2013 TOP STORIES: HFT Checkmate – The Alpha in Order Types":
  • Missing
    abekohen

    22 January 2013

    Interesting points you make.

    I guess where I am confused is on the issue of what prevents non-HFTs from using these order types. What prevents a checkers player from playing chess? Inate ability? Spending time and effort to learn?

    When the NYSE was dominated by "specialists" and they'ld rip you off for 1/8s and 1/4s, then everyone else, who was not "the house" was disadvantaged. Can't anyone, who wants to, use "HFT-oriented" order types? Can't anyone, who wants to, become an HFT? 

  • Missing
    John Harris

    22 January 2013

    I would guess that something like this happened: after Regulation ATS opened the exchange market to for-profit models, leading to "demutualization," new exchange operators still faced the so-called "chicken or egg" problem. How could they attract passive orders or "starting liquidity" to their nascent markets? Perversely, the trade-through rule provided the answer, along with rules concerning the allocation of tape revenue. If new exchanges could convince just a handful of HFT firms to stand as initial liquidity providers - a term I use loosely - even though their market depth was poor in relation to the incumbent exchanges, they could still compete. The new exchanges, which weren't necessarily registered as such and didn't have to be under Reg ATS, adopted rebate-pricing models to attract their initial liquidity from firms who already knew how to game the SEC regulations. The nascent exchanges could "route out" any order depth these initial market makers couldn't handle and assure their initial customers of full fills - otherwise, they would never have been able to attract customers in the first place.

    This state of affairs begat initially a rebate race among the exchanges, including incumbents - the winners were the HFT firms. Once rebates were widely available, the new entrants had to do something else to compete, so they started crafting new order types and "undocumented matching engine" behaviors, to use Haim's phrase, to retain the patronage of the HFT crowd. These then spread to other exchanges.

    As a bond market person who merely observed these stock market machinations from afar, I undoubtedly have some of the details wrong and do not doubt that my scenario only approximates the truth. But to your question, Mr. Kohen, the checkers players never had a chance. They weren't at the table when exchanges made these deals with HFT firms. And, how could the checkers players know about undocumented order-handling practices?

  • Missing
    Marie@PublicNotions

    22 January 2013

    Complex order types, managed by Exchanges is distorting the efficiency  of Price/Time integrity.  Bats has over 2000 order types.   Leave the order type complexity to Brokers to manage, and keep the exchanges running a core matching engine ( or even auction engine) for price discovery. 

  • Comment_230146_210851315613283_100000652474653_678322_2285980_n
    crammond1964

    22 January 2013

    Exchanges perhaps were less than honest with the truth and advantages given to certain parties ?

     "fair and orderly markets " were probably not achieved or maintained during this period which puts an argument to whether exchanges should be given RIE status ?

  • Anon_avatar
    Anonymous

    22 January 2013

    One more argument for turning financial markets into non profit organizations.

  • Missing
    abekohen

    22 January 2013

    Mr. Harris, if I wanted to learn to play chess, I would start by googling "chess" and proceed to learn everything I could about playing chess. If I wanted to learn about order types I would select an exchange and proceed to learn all the rules of said exchange by reading its rule book and other materials.

    For example,

    http://www.batstrading.com/support/

    http://www.batstrading.com/resources/regulation/rule_book/BATS_Exchange_Rulebook.pdf

    and I would study Chapter XI extensively.

    Are you and Mr. Bodek saying that the publicly disseminated information is incomplete or undocumented? Or that it requires some intelligence to analyze and comprehend. If the former, you would have a valid complaint to the SEC. If the latter, greater effort brings greater rewards.

  • Comment_230146_210851315613283_100000652474653_678322_2285980_n
    crammond1964

    22 January 2013

    the problem lies that the regulators cannot interfere with exchanges . Their RIE status prevents any external intervention  ; even if the complaint is valid the exchanges currently are above the law and here lies the problem . When they ; exchanges ; eventually admit their error  I hope the fines are as hefty as our libor ones !

  • Comment_salarnuk
    sarnuk

    22 January 2013

    Reg NMS created a NBBO that was to be protected. Exchanges, in their efforts to not incur take fees by routing out, created price sliding. And they also created and proposed complex order types. Who asked for these order types? Who asked for Hide not Slide? Who asked for Post-only DAY ISO? Isn't and ISO an aggressive order? Is a Post-only DAY ISO a passive aggressive order then?

    It would seem to a casual observer that a good many of these order types get their magic in how they manipulate the matching engine around locked and crossed situations. Is this what the SEC envisioned? 

    Are all these order types documented? Are they documented with the SEC? Are they documented in the exchange rule books? Is it clearly laid out?

    Have these order types served public confidence? Have they distorted price-time priority?

    Lastly is it somewhat disingenuous for exchanges, who have created the complex order types, to then after the horses have left the barn, complain about market complexity?

    Why don't we fix all the matching engine "glitches", eliminate some of these passive aggressive order types, reduce the cap on takes from 30mils to 10 mils, and require exchanges to route out to clear a locked market, as opposed to "price-slide"?

  • Missing
    John Harris

    22 January 2013

    Mr. Kohen, you may have missed the point of Mr. Bodek's report: these exchanges handled orders in ways  contrary to their own rules and documentation. So, to use your analogy, you could google all you wish and still not have learned the applicable rules or pertinent details of "chess." And we don't need to speculate about whether Mr. Bodek really knows these details (though by all appearances he does, and quite well at that): several exchanges admitted that they handled orders in ways contrary to their own rules, over a period of many months, and that customers suffered losses as a result.

    Further, I would guess that in your own experience in our markets, you have encountered exactly the circumstances that Mr. Bodek alleges: cases where code either drifted away from its documentation or never accorded with its documentation. Sloppiness or laziness may explain some or even most of these discrepancies. Perhaps in only a small fraction of the cases are the discrepancies purposeful. But people proffering market data, matching, order handling, and reporting interfaces should be scrupulous about them. And exchanges particularly should know their own rules and affirm that their code implements their rules faithfully and reliably.

  • Missing
    abekohen

    22 January 2013

    Mr. Harris, I give the electronic exchanges the benefit of the doubt when it comes to having messed up price-sliding inadvertently. I don't assume nefarious behavior. Perhaps you are right that it was deliberate, but from my own experience I know how easy it is to make such coding errors and if the errors are small (1/2 million in 4 years distributed over a huge number of clients and orders) they are indeed hard to find. (Imagine tracking a $1/2 million error in GE's annual report - it's in the rounding error.) But please, if you have evidence to the contrary, or if Mr. Bodek (Inspector, what an appropriate name!!!) has evidence to the contrary it would be in the public interest to expose it.

    I would not extend the benefit of doubt to "legacy" exchanges for obvious reasons.

  • Comment_haim_linkedin
    hbodek

    22 January 2013

     

     

    At least for me, it was not clear that  invisible unexecutable orders that internally self-lock the BATS exchange were permissible until the following **very recent** regulatory filings were made which explained how order matching applied to such states for the purpose of justifying further matching engines changes. 
     
    August 14, 2012
    Self-Regulatory Organizations; BATS Exchange, Inc.; Notice of Filing and Immediate Effectiveness of Proposed Rule Change by BATS Exchange, Inc. to Amend BATS Rules Related to Price Sliding Functionality

    http://www.sec.gov/rules/sro/bats/2012/34-67657.pdf

    June 27, 2011

    Self-Regulatory Organizations; BATS Exchange, Inc.; Order Approving a Proposed Rule Change to Amend BATS Rule 11.9, entitled “Orders and Modifiers” and BATS Rule 11.13, entitled “Order Execution”

    http://www.sec.gov/rules/sro/bats/2011/34-64754.pdf

  • Comment_salarnuk
    sarnuk

    22 January 2013

    Haim, thanks for your works and your knowledge-sharing. The industry owes you a big thank you.

     

    Going back to what you just stated... does anyone reading this believe that exchanges allowing orders that self-lock their own quotes serves price discovery, or anything other than rebate arbitrage video games?

  • Missing
    John Harris

    22 January 2013

    We're already discussing "evidence to the contrary," Mr. Kohen - quite openly. We already have admissions from the subjects of the discussion, i.e., evidence. Does the evidence rise to the level of nefariousness as opposed to mere sloppiness? As I said above in discussing code drift generally, "Sloppiness or laziness may explain some or even most of these discrepancies." And in any event, I am not in a position to know whether these admitted, undocumented exchange features comported with the law. Those that the conduct harmed or that the law charges with policing these exchanges have (i) the necessary standing and investigative powers and (ii) legal or fiduciary obligations to get to the bottom of the matter. We're debating the applicable laws and regulations in this forum and whether enforcement of same is what it should be.

  • Missing
    mariegiangrande

    22 January 2013

    Let's paint the technical side of this:  13 venues all routing to each other under Reg NMS, speeds approaching  NANOSECONDS for receipt or market signals and triggering transactions and 2000 PLUS complex order types managed at  the EXCHANGE.  Do you think we may be approaching the 'edge of engineering possibilities' when it comes to 'managed, orderly markets?'    

  • Missing
    Marie@PublicNotions

    22 January 2013

    Let's paint the technical side of this:  13 venues all routing to each other under Reg NMS, speeds approaching  NANOSECONDS for the receipt of market signals and triggering trades. Now layer on  2000 PLUS complex order types managed at  the EXCHANGE.  Do you think we may be approaching the 'edge of engineering possibilities' when it comes to 'managed, orderly markets?'    

  • Anon_avatar
    Anonymous

    23 January 2013

    1. In the end using one or two examples of an un (or ill) documented order type is not really describing the broad market situation. The truth is it's all public information, all available in the documentation and the exchanges are there to answer any questions anyone has.

    2. All these order types exist because locking markets for years is being perceived as deregulation by the SEC, which it is not (actually it is rather efficient, dear client you can buy and sell at the same price). It would just solve all of these issues. HFTs are fine with that. IF you create this situation economic forces will just make sure smart people will find out all the details and be better than the ones who don't figure it out. Really don't understand the story here.

    3. Also, you fail to show why actually all of this disadvantages the institutional investor? Are you arguing for a pro-rata equity market then? if you remove these orders types why is the institutional investor so much better off ? expect for when not removing lockedmarket prohibition AND removing these order types markets likely to stay wider a lot longer.

    4. What do you propose exactly? I would propose to remove the local lockedmarket prohibition which topic started to get such a bad name after confusing the BATS/Nasdaq Flashorders with what DirectEdge was doing.

  • Comment_haim_linkedin
    hbodek

    28 January 2013

    Anonymous,

    Most of your questions are answered quite explicitly in my book. If you don't want to buy it, you can use the Amazon borrow feature to get a hold of it. Otherwise, here are some brief answers to your questions:

    1. Your two statements are contradictory, and I am getting tired of flippant statements like "it's all documented" when we have months of press citing insider sources voicing significant concerns at the level of exchange disclosure. For my views on where the current debate on disclosure actually is now, refer to the section in my book called "Did an entire industry fail to do its homework?" I am willing to discuss more detail on specific practices after I see some willpower from one or two of the other major exchanges to follow NASDAQ's lead in providing additional documentation on order matching engines practices and special order types.

    2. The story is that queue-jumping and all the other perks squirreled into matching engines that gave HFTs an edge is evidence of unfair and discriminatory treatment by exchanges who advantaged one class of participant above all others. It indeed was a situation too good to be true, and the industry is in the process of correcting the anomaly - that is why HFT-oriented order types are news. The "smart people" your refer to who found "out all the details" certainly did not receive instruction in the usage of such features from exchange documentation. Note that these "smart" people took a significant amount of regulatory risk when they chose to exploit order matching engine features that have only been recently scrutinized by regulators. Perhaps regulatory arbitrage isn't such a smart thing after all? Only time will tell.

    3. I am arguing for full disclosure of HFT-oriented order type functionality, interactions with other order types, intended usage cases, and order matching engine practices that serve to give HFTs an edge over the public customer. I am arguing that order matching engine abuses (such as queue jumping) that serve to preference HFTs be eliminated. I am arguing that Rule 610 and Rule 611 be made robust against the "smart" regulatory arbitragers you describe. I am happy with the direction that NASDAQ is taking to address the issues I describe. I want to see the other exchanges take similar steps.

    4. In my 10 point plan for reforming US market structure that is detailed in my book, I argue for a repeal of the ban on locked markets, concurring with the viewpoint expressed by Senator Ted Kaufman in his 8/5/2010 letter to the SEC. However if a repeal of the ban is implemented, I must stipulate that order matching engine practices will likely need to be further constrained to focus on the needs of long-term investors, especially with regard the conditions in which  maker/taker fees are allocated. Note however, that though I believe the SEC should conduct a pilot test to assess the impact of eliminating of the ban on locked markets, in no way do I see the arguments for eliminating the ban to justify the circumvention and undermining of Rule 610 by exchanges and HFTs who created a de facto two-tier system that advantaged firms employing HFT strategies over the bulk of institutional investors.

    Haim

  • Anon_avatar
    Anonymous

    28 January 2013

    Haim,

    Indeed i was not commenting on your book, but on this article. Not aware of any 'book' . I find it difficult to not comment on that though, drawing a parrallel between apparently all the answers to my questions being available in a book that i can buy and documented order types. At this point i guess we cannot cover point 1. anymore as it's described in your book. What is the purpose of this book for you? If it's about openness then why don't you make it freely available digitally? If it's about marketing for your firm,  you could argue the same or if it's about making money, yeah then you should leave it on there. I agree with you that your publications were news to the general public, but following that logic there can be a billion more articles written (and many are) on any technical topic and proivde more information, that doesn't make it 'evil' not being available yet in the retail community but a good addition to public knowledge.

    Also, if  you already get tired of arguments i would suggest to focus your energy somewhere else i think it's kind of the point of posting articles. You provide your viewpoint and others provide theirs. The path to knowledge of order types has always been available. The discussion should be whether for which type of participants and their needs this setup makes sense. You just can't argue the answers are not available. However, like you do, providing more information already i'm sure is appreciated.

    - Exchange Website

    - Exchange documentation

    - Exchange helpdeks, Exchange accountmanagers, every question gets answered.

    - Exchange test environments, check it out for yourself. In my opinion all professional parties (brokers,institutions with direct access, HFT have the obligation to know all there is to know when being active on these markets, especially when dealing with client orders

     

    2. As i mentioned I agree with removing the locked market bias of it being 'deregulation' to remove it. All of this unclarity is because of it and noone asked for it. 'HFT' is perfectly fine with this i'm sure.  What i still fail to see though is how you think institutions or some other group you represent will now suddenly be so much happier, but if they are that's fine.

     

    My main point on your article here is that  you throw together requests and providing for more information with the bias of 'badness' of HFT. There are many forms of regulatory Arbitrage, but using exchange functionality the way the account managers explain to you to use it on order types approved by the SEC i find really a bit strange. Why the strong bias? Due to the prohibition on locked markets the exchanges have to make choices how to handle locked situations, there is no way around that. using the term 'Queue jumping' is merely throwing oil on a fire and spice up your argument (and book). There is no such thing as 'queue jumping' . It's like saying if the exchanges introduce half ticks and a member starts using half ticks that he's queue jumping. Again the problem is the specific SEC rules the exchanges have to deal with.

     

    Summarizing I agree with that the current marketstructure should be adapted but i fail to understand what the purpose and merit is of heading the article for example: "HFT Checkmate". That debate is more about what is the function of participants with different trading horizons.

  • Missing
    abekohen

    28 January 2013

    I downloaded Mr. Bodek's book from Amazon for free since I am a Prime member (costs $$). Haven't read it yet. Interesting back and forth between anon. and Bodek. Seems like the information on order types is out there and it's not as easy as googling, but with reasonable effort it's accessible to ALL institutional traders.

  • Missing
    abekohen

    28 January 2013

    @mariegiangrande: When all else fails, divide & conquer. 2000 order types & nanoseconds is not a problem. If one computer cannot handle it, then just get more computers, hence divide & conquer.

  • Missing
    John Harris

    28 January 2013

    Here's the thing, Anonymous. You never want to be the person who defends the indefensible. In this forum and elsewhere, I have many times defended systematic trading. I oppose a government ban on systematic trading. I believe systematic trading is beneficial to mankind. So "talk your book" - that's fine, no problem - but for your own sake, don't insult our intelligence and call your own probity into question by suggesting that market participants should learn about the true order-handling practices of exchanges by talking to their help desks and account managers or trying out programs in their test environments. That's just nonsense - especially when these exchanges purport to handle orders one way but actually handle them another.

    And though he can defend himself perfectly well, I am grateful to Mr. Bodek for graciously, generously, and patiently sharing his knowledge and answering our questions. I learned something of value from him and I have no doubt that it took him many years of hard work to acquire the knowledge. Perhaps this wasn't your intention, Anonymous, but if your post isn't actually disrespectful and even insulting, it's pretty close to the line.

    Finally, if you didn't know about "the book," Anonymous, I am not sure how you missed it, presuming you actually read the article. Just look above - the information is there.

    I'll close now, because I am going to find my credit card so I can purchase Mr. Bodek's book. With gratitude.

  • Missing
    abekohen

    28 January 2013

    Can't answer for anonymous, but when I worked at DE Shaw, the quants would all read the latest academic research and see if they could monetize it. True capitalism. Risk-reward.

  • Missing
    John Harris

    28 January 2013

    There is an easy way to settle this issue. Let every exchange come into this forum, or some forum, and explain authoritatively and comprehensively how their matching processes work, and make themselves liable for matching orders exactly as described. When orders arrive, exactly how are they evaluated? Exactly how are orders prioritized for execution? What events kick off a match and what events do not? What happens to orders while a matching cycle is underway? When orders that arrived during a matching cycle are removed from cache for evaluation, exactly how are they prioritized for evaluation, if they are? Under what edge conditions are normal order-matching priorities in indeterminate states?

    No exchange has any legitimate reason for hiding any details about its order-handling process. Let them all spell out these details for all the world to see, and be held to them.

  • Missing
    abekohen

    28 January 2013

    WADR to Tabb and the Tabbforum, I don't think that this is the venue for exchanges to do such. But you could try to invite them. No one would be averse to that.

  • Comment_230146_210851315613283_100000652474653_678322_2285980_n
    crammond1964

    28 January 2013

    SADLY exchanges have been  guilty of this since 2003 . 

     However there is rumor that they are to be accountable at last ! I just hope they accept their

    punishment as they accepted their profits from certain HFT .

  • Anon_avatar
    Anonymous

    28 January 2013

    John,

    Those are all fair comments. In any case i did not intend to insult anyone, I did feel responding to referring to a book as an answer to my questions as well as 'becoming tired of an argument'. I believe those comments probably triggered you in your response. 

    I acknowledge completely the information that was shared in a pretty clear way and i also agree with the effects it has on the market as well as that it was 'news' as such. What i'm (trying) to comment on though is that these things are going along a continuous attack on this thing 'HFT'. If he's right on what happens with order types and locking markets that doesn't make him right that HFT is now checkmate or proven 'wrong' or 'evil'.

    Both sides of the story have to be viewed also when you attack.

    Thanks for your feedback.

  • Missing
    John Harris

    28 January 2013

    I've ordered Mr. Bodek's book and look forward to reading it. But I haven't gathered from anything Mr. Bodek has posted in TabbForum that he is "anti-HFT." What I have gathered is that he opposes undocumented exchange features and exchanges handling orders in ways contrary to what they have led the great mass of their customers to expect. Look, if exchanges want to give one trader or a group of traders an advantage, as far as I'm concerned, that's their right - they can do as they please. But to the extent they are promising one thing but doing another, they are wrong, at least ethically if not also contractually or legally.

    Executives at these exchanges read this forum. I've yet to see one objection from even their PR people that Mr. Bodek is wrong. At some point, the silence speaks for itself, don't you think?

  • Missing
    abekohen

    28 January 2013

    Perhaps the exchange officials feel that there is nothing to be gained from entering a pugilistic match here? Look, I hated NYSE specialists and OTC (NASDAQ) market-makers from day one in the late 80s when they were "stealing" multiples of 1 bit (12.5 cents) from me. They were the "house" and I was their prey. But with HFTs, anybody can become an HFT. It's not a private club anymore. And I would rather pay 1 cent on a 1 cent spread (with HFTers creating a penny spread), than 3 to 5 cents on a nickel or dime spread (without HFTers), but that's just me.

  • Missing
    John Harris

    28 January 2013

    I'm sure they have their reasons for not responding. Among those reasons could be that Mr. Bodek is right and they don't want to call additional attention to his remarks. Another could be that they don't want to explain their order-handling processes in public, for which - by the way - there cannot be an innocent explanation. It is true in a theoretical sense that "anyone can become an HFT," but as a practical matter not true. We do not have an open market inviting to one and all. We have what amounts to a hot Chelsea nightclub that requires would-be patrons to wait through a velvet rope line and then pass muster with a doorman whose standards for admission are discretionary and conflicted.

    Nominal or apparent spreads are one thing; actual executions are another. If some favored party gets an execution to which you were otherwise entitled and the market moves against you, the nominal spread means nothing.

    I'll repeat: I'm a fan of systematic trading. I think it is beneficial. But I'm also a fan of exchanges handling orders as advertised, especially since they enjoy, as SROs, delegated governmental powers.

  • Anon_avatar
    Anonymous

    28 January 2013

    Yeah i think these last 2 comments summarize my thoughts.

    - Describing the situation is good, for the public eye and for the professional eye. Mr. Bodek did a very good and clear job here.

    - Now that more people are aware of how it works the discussion is back to HFT vs Institutions vs Retail or short term vs longer term or does short-term provide tighter spreads. I"m sure it's clear a lot is being written about this and it's getting a bit more balanced in my opinion at least. Different topic.

    - I second abekohen point that it would not be strange (IMHO) for the exchange to think this given the various horizons involved. They just want their book to be visibly tight asap, which will help them in their flow and in the end provides very tight NBBO spreads to investors.

    - Circling back to Mr. Bodek and Mr. Harris point above i would argue that labeling an article : HFT checkmate and also looking at his original publication my perception is that an opinion on HFT is combined with the (again more properly than ever) describing the impact of order types and market structure in the public domain. I have difficulty moving away from that point.

  • Comment_haim_linkedin
    hbodek

    28 January 2013

    Thank you Mr. Harris and abekohen for getting my book. I sincerely appreciate it and look forward to your feedback. I hope you find it clarifies the nuances of my arguments, as well as where I draw the lines between the good, the bad, and the ugly with regard to HFT.

    And thanks to everyone for bringing the conversation into a civil debate. I'd like to add that I think it is important to point out what the regulators have been communicating about the topics I discuss above and which are further articulated in my book. Here are some important articles that should fill in some of the history and should provide some context for where things might evolve next.

    The regulatory scrutiny of undocumented order matching engines practices at exchanges that serve to advantage HFTs over institutional investors has been reported in the press for a year.

    Regulatory scrutiny of HFT-oriented order types first hit the press in Feb, 2012:

    http://www.businessweek.com/news/2012-02-25/sec-reviewing-u-dot-s-dot-trading-practices-after-decade-long-shift-to-automation#p1

    I believe the first mention of queue jumping being investigated by the SEC was April, 2012:

    "The SEC is examining whether such order types unfairly allow high-speed traders to jump ahead of other investors in an exchange's "order book," or the queue of buy and sell orders that are typically ranked by price and when they were received"."

    http://tech4b.blogspot.com/2012/04/sec-probes-ties-to-high-speed-traders.html

    In November 2012, it was announced that the OCIE division of SEC was expanding the special order type probe. According to a source in the SEC:

    "The order-type probe was a "come to Jesus moment" for regulators about how exchanges appear to have systematically developed trading advantages for certain sophisticated clients, often with insufficient regulatory oversight or understanding of products"

    http://online.wsj.com/article/SB10001424127887323622904578129210389143012.html

    Today's article in Trader's Magazine might explain some of the silence by exchanges with regard to the issues noted above.

    "Robert Khuzami, the U.S. Securities and Exchange Commission’s enforcement chief, said the agency is “very focused” on bringing cases against exchanges and traders when system or programming failures harm investors.

    The agency plans to bring more cases related to weak controls that result in “real harm to clients” in addition to those stemming from intentional misbehavior or rule violations, Khuzami said at a Securities Regulation Institute conference in Coronado, California."

    http://www.tradersmagazine.com/news/SEC-very-focused-on-policing-high-speed-traders-110784-1.html

     

    As I indicated earlier, NASDAQ has shown some leadership in becoming more transparent and less asymmetric in terms of feature set. That is the only exchange that has adequately responded in my opinion.

    P.S. I didn't pick the title "HFT checkmate," but I confess I like it!

  • Comment_230146_210851315613283_100000652474653_678322_2285980_n
    crammond1964

    29 January 2013

    bodek ; thankyou on behalf of many local day traders in London ; who now understand why exchanges have ignored their "screams " of dissent  . 

     Hide not slide was strangley not offered to us or even explained ; however if operated by us I believe would of been closed down immediately .  CHECKMATE  i could not agree more !  Finally glad to see folk no longer sitting on the fence .

  • Missing
    John Harris

    29 January 2013

    Interesting articles, Mr. Bodek. Given Khuzami's odd remark in the Traders Magazine piece about what CEOs know - or don't know - about the manufacture of mortgage derivatives, you may rest assured that nothing will come of this HFT investigation outside of some fines, even if the conduct merits harsher sanctions. As always, what the SEC purports to be, and what it actually is, are two different matters. 

  • Missing
    Marie@PublicNotions

    29 January 2013

    HOUSTON, WE HAVE A PROBLEM - There is another dimension of the problem that is being missed.. that is the technical complexity of managing complex order types when routing to achieve best price to other exchanges. While we have speeded up the market to nanoseconds and layered in complexity, we forgot to ask the engineers if we are pushing the technical limits of realtime networks. (Who knows maybe it's in their self interest not to speak about this.) . It's called RACE CONDITIONS and it puts in jeopardy price/time priorities that is THE CORE MISSION of the exchange. RACE Conditions are a problem in all real-time systems -- exacerbated when several of them are in communication with each other and are affected by their interactions with each other. See: Wikipedia: en.wikipedia.org/wiki/Race_condition A race condition or race hazard is a type of flaw in an electronic or software system where the output is dependent on the sequence or timing of other inputs.....

  • Missing
    Marie@PublicNotions

    29 January 2013

    Some times the answers are good, but the questions are crummy...Let's not ask if HFT take advantage of complex order types, let's ask: are Exchanges able to adhere to price/time priority if they are also handling complex orders in the midst of their routing obligations.. Let's ask: What are the points of failure in the system? Let's ask: if trading for alpha is all about finding imbalances, can a speculator cause perceived imbalances and how? Maybe the Open and Close..... and understanding the nuances of price/time adherence in the midst of RACE conditions happening every day.... or not happening ....because a programmer did decide the priority of handling the order ... instead of the matching engine.

  • Missing
    abekohen

    30 January 2013

    @marie: Race conditions have been around since the advent of real-time computers. One word: mutex.

  • Missing
    John Harris

    30 January 2013

    Mr. Kohen: the issue is not for how long race conditions have been known; it is that they are to be avoided. I believe her point is that complexity in order design, matching, and handling tends to lead to the manifestation of more race conditions. Nasdaq admitted that a race condition caused the failure of the Facebook IPO.

  • Missing
    abekohen

    30 January 2013

    Mr. Harris: Marie raised an issue about race conditions and complex order types. Where does the FB IPO come into the question? Did they use complex order types? NASDAQ OMX might have blamed the fiasco on a race condition, but that does not mean the cause of the fiasco was a race condition. But that's a non-squitur anyway. Yes, there's complexity. Yes, computers can handle complexity - better than humans. Yes, it requires skilled programmers. Yes, it requires lots of testing, testing, testing.

  • Missing
    Marie@PublicNotions

    30 January 2013

    Given the fragmented market with routing requirments, the testing needs to be done at a "market level' not at an indivicual exchange. Who is doing this? This will become a bigger and bigger issue....

  • Anon_avatar
    Anonymous

    30 January 2013

    ... and it regularly bugs, bugs, bugs. In the line of the G20 decision post crisis, not only structure"d derivatives, but really anything on the markets even remotely looking "complex" should probably be banned - and not only what touches the retail clients : I remember middle office people confessing they were virtually unable to make much of those structured products that led to the subprime crisis. I suspect the same goes for many of the 'complex' order types and I doubt reading the exchange rulebooks will suffice...

  • Missing
    John Harris

    30 January 2013

    Mr. Kohen: Here is one of the accounts I read of the race condition that affected the Facebook IPO:

    http://www.securitiestechnologymonitor.com/news/nasdaq-post-mortem-race-condition-facebook-30606-1.html

    It implicates order handling and matching. But in any event, she didn't limit her concerns to "complex order types" and neither did I. The trade-through rule practically invites race conditions. BATS attributed the failure of its IPO to a race condition and its CEO blamed market complexity as a contributing factor.

    Testing? See the comment from the Chief Information Officer of Direct Edge here:

    http://247wallst.com/2012/12/04/high-frequency-trading-a-grave-threat-to-the-markets-and-the-economy/

    From the deconstruction of the May 6 flash crash I learned that some matching engines - ARCA's in particular - had unconscionable designs with respect to how they handled certain orders (injecting them in the order book at far-off-the-market prices on the theory they would never be executed accidentally). No one is suggesting that human beings could match orders better. The issue is whether regulatory mandates or competition to help systematic traders skirt regulatory requirements are causing fragility.

    Let me repeat that I support systematic trading and think it's beneficial. But in response to Mr. Bodek's disclosures that exchanges may be violating their own rules as well as the law, I am detecting some inexplicable defensiveness from people who don't seem to have a dog in the fight. Perhaps I'm misperceiving some of the comments here and if so, I apologize. But otherwise, I see repeated defenses of what appears to be indefensible behavior and wondering why that might be the case.

  • Missing
    abekohen

    30 January 2013

    Mr. Harris: Let's stop mixing semaphores and metaphors. You are conflating a lot of things.

    Indefensible? Let's start from the top. HFT is not a villain, IMHO. They are players and a lot more honest than NYSE specialists and NASDAQ market makers of yesteryear. They do their homework and invest heavily to be competitive. In the great capitalist tradition of the USA, they earn their keep.

    Exchange software, like all software, is likely to have bugs. The best way to get rid of bugs is by extensive testing, by the developer, by a QA person, by a team leader, by clients who interact with that software. Testing should check for likely scenarios as well as unlikely scenarios. NYSE mentioned a FB IPO bug when they used an array of sizeof largest short integer (65,535 = 0xffff) and that was exceeded. Oddly enough, we encountered a similar bug 20 years earlier when WMT had a 2:1 stock split and we represented shares as a short integer. Beyond testing, contingency plans must be formulated for dealing with the unknown unknown. (No, that wasn't a typo.) My navigator app on my iPhone (WAZE) sometimes wants me to drive into the river - I don't!) Proper placement of mutexes and semaphores are the responsibility of the programmers and those who test the software.

    Structured derivatives, CDOs, CDOs squared, ... how are these related to the topic on hand?

    Sometimes a little education is more dangerous than none.

  • Missing
    John Harris

    30 January 2013

    Let's see if we can close on a productive note, perhaps even one of conditional agreement, Mr. Kohen: IF it is true that exchanges are handling or matching orders in ways contrary to their own rules or to federal laws or regulations, would you agree they should stop doing so? And IF it is proven that this contrary order-handling or -matching behavior was purposeful rather than erroneous, would you agree that sanctions are in order?

    Please take me at my word that these questions are not in the vein of "have you stopped beating your wife?" Further, I acknowledge you do not accept my posited conditions as true. I ask conjecturally and in the hope of understanding your argument.

  • Missing
    abekohen

    30 January 2013

    Mr. Harris: "IF it is true that exchanges are handling or matching orders in ways contrary to their own rules or to federal laws or regulations, would you agree they should stop doing so?" YES!!! "And IF it is proven that this contrary order-handling or -matching behavior was purposeful rather than erroneous, would you agree that sanctions are in order?" YES!!!

  • Missing
    John Harris

    30 January 2013

    Thank you, sir, and for the discussion generally. Your information about the memory-allocation bug was news to me.

  • Missing
    Marie@PublicNotions

    30 January 2013

    IS OUR SAND BOX CLEAN? Thank you Sirs for the discussion. I am so glad to know there are no SOES bandits in the market, that every exchange tests their software and follows their published statement of order handling, while the programmers lock in their processing cycles using MUTEX to avoid RACE CONDITIONS. My only persistant question is: if every exchange tests independently, and does not display their programming logic when handling, for instance sliders or market sweeps in face of more complicated order types (pegs, stops, etc) how do we know the system at large is managed and tested to PRICE/TIME PRIORITY? Maybe it's too many years of finding surprises when testing integrated systems.... HFT players know more than anyone about the intracate connections of the market .... if they have found a faulty logic to protect Price/time, I believe they would be bounded ethically and patriotically (probably not legally) to bring this forward. But should not the exchanges, as self regulated entities, collaborate on nation-wide testing of the system?

  • Comment_untitled
    vddobre

    31 January 2013

    " The best way to get rid of bugs is by extensive testing'

    Extensive testing yes; and also, avoid getting in a position likely to create problems; avoid situations that are prone to produce bugs. Stop unnecessarily complicating things. Make algos 1) simple enough (no marketing argument justifies creating software that is difficult to control, no matter your testing skills) and 2) submitted to regulatory approval (if only to avoid algo inflation).

    That will not completely ban the unknown unknown (or the Black Swans) but will limit the conditions favorable to its inception. (as to structured derivatives, that's just another example of a financial topic turned hyper complex after modern IT got involved).

    Btw given the potential consequences, I would liken trading software with air traffic control management systems rather than one of my mobile phone apps.

You must log in to comment.