TMC is an independent, primarily volunteer organization that relies on ad revenue to cover its operating costs. Please consider whitelisting TMC on your ad blocker or making a Paypal contribution here: paypal.me/SupportTMC

Discussion on High Frequency Trading and its Impact on the Markets

Discussion in 'TSLA Investor Discussions' started by hershey101, Apr 2, 2014.

  1. hershey101

    hershey101 Member

    Joined:
    Jan 22, 2013
    Messages:
    767
    Location:
    New York, New York, United States
    #1 hershey101, Apr 2, 2014
    Last edited by a moderator: Apr 4, 2014
    Hi all,
    So there has been a lot of talk about Michael Lewis's new book, Flash Boys, and I have even gotten a private message asking about my opinions on it.
    Just so everyone's aware, I'll start of with full disclosure that I do work for a HFT firm, one of the original players in the space in fact. However, I'm not going to attack or defend the business model, I'm here to make people aware of what HFTs do, how they work, and what their impact on the markets is. We at TMC are clearly an intelligent, and well-informed, group when it comes to Tesla Motors, and I feel like it is my duty to the community to make sure that they aren't swayed by the mainstream media and publicity stunts like Michael Lewis has done with his 60 minutes piece and his CNBC interview. To say that the "Markets are Rigged", and then to go and blame it on the High Frequency Trading firms is completely an oversimplification of a very complex situation, and in my eyes, a marketing move for him to sell his new book.

    Now that being said, here is what I posted on the short-term tread about the discussion (mods feel free to move that discussion here as it has nothing to do with Tesla):

    So here is what the other side has to say about it (Full disclosure, I'm a fan of Tower):
    http://nymag.com/daily/intelligencer/2014/04/9-gripes-from-a-high-frequency-trader.html

    Now IMHO, he makes some pretty good points about HFT and has valid critisms of the book, and the general take on the situation. I mean, lets face it, his big headline is "Markets are Rigged" or whatever the 60 minutes segment called it. To be fair, I think Gorton is being a slightly defensive about HFTs as well... Lets face it, everyone in finance is in it to make money. But I think electronic trading as definitely improved the market place, and I view what (HFTs) do as being positive in terms of liquidity provision. Its just that it comes at the costs of big banks and hedge funds.... At the end of the day though they still take a lot of money out of the retail investor's pockets... There is no reason why TDAmeritrade should be charging $8/trade and over $15 for options (I just did some CSIQ trades and was surprised by the costs!) when IB can do the same thing for $3 on a much better execution engine... Obviously, if your an HFT, you probably don't pay more than a few cents on a small trade, but to charge retail investors anything over $5 for any single trade (stock, option, etf, fx, future, forward, spread, whatever), is insanely overpriced (especially since most retailers dont trade in huge quantities).

    Do I think HFTs were a net positive when they came out? Yes, they provided liquidity. Today, any new HFTs that come out are probably not going to be doing social good since liquidity is plentiful, but that also means its much harder for them to survive and make money let alone succeed.... Do I think the founders of various HFTs a decade and half ago had social good as a goal? No, most likely not. They wanted to make money, and the social good it did was a nice by-product... That being said, I do believe certain companies/executives have/had done things that are morally questionable, if not illegal. But lets face the facts, thats true of any industry any where in the world. You have your good apples, and your bad. To judge an entire industry from the bad apples is a big mistake and very narrow minded...

    Also for those who don't know, here is a excerpt from the book:
    http://www.nytimes.com/2014/04/06/magazine/flash-boys-michael-lewis.html?_r=0

    - - - Updated - - -

    So, for those that want a more balanced view on the markets, or have questions about HFT, electronic trading. You can post here, and I'll try to answer them to the best of my abilities (and of course I encourage friendly discussion as long as everyone stays civil... :) )
     
  2. apvbguy

    apvbguy Banned

    Joined:
    Oct 30, 2013
    Messages:
    443
    Location:
    Ponte Vedra Beach FL (JACKSONVILLE)
    I think that there are 3 issues in the equation, 1. I agree that the HFTs do add liquidity and volume but all they do is turn the capital markets into casinos, 2. there is a certain amount of an edge the HFT can get and the may have an unfair advantage, before the markets went electronic the specialists and floor traders had an edge and were front running the market. lastly and this is the big secret that nobody mentions is how the exchanges kick back fees to the larger HFTs this just encourages more trading which back in my day was known as churning or wash trades
     
  3. woof

    woof Model S #P683 Blue 85 kWh

    Joined:
    Apr 30, 2009
    Messages:
    1,149
    I have a few questions about HFT, if you don't mind.

    1. Assuming the stories of moving comm links to be closer to the exchanges in order to lose a few mS of latency are true...what can a few mS really get? Why does it matter? Perhaps if you explain the basics of HFT it will be more clear.

    2. If mS matter and they really give some closer HFT houses an edge, wouldn't it makes sense for the exchanges to insert a delay to level the playing field? Or if not a delay at least some limits on transactions per second? Or increase the fees/commissions such that it naturally regulates the players.

    3. Why don't the fees/commissions suck the profit out of HFT?

    Thanks!
     
  4. Cosmacelf

    Cosmacelf Active Member

    Joined:
    Mar 6, 2013
    Messages:
    3,399
    Location:
    San Diego
    I have read Lewis' book and I just read your link, hershey. Gorton's defense is pretty weak, as is all HFT defenses I've read in the last couple of days. Here's my take, please correct me where I'm wrong!

    First, the 60 minutes piece was indeed over the top. Lewis' book is much more nuanced. Let's get past the marketing of a book and into the substance of the allegations.

    Lewis' book describes three ways that HFT firms make money. The easiest to understand is front running. When an order comes to the market, it get routed to a particular exchange (there are like 60 exchanges now). That exchange will try to cross the order then and there if it can, otherwise, it routes it to all the other exchanges. A front running HFT sees the order as it comes into the exchange and is sometimes able to purchase the stock the order is trying to purchase before the order gets executed. It then sells the stock to the original order for slightly more money. This practice should be illegal and the only reason it isn't is because regulators are either corrupted or asleep (take your pick).

    All HFT strategies rely on special high speed access to the exchanges' data feeds. The exchanges realized a while ago that being a few milliseconds closer to their matching engine computer gave HFTs an advantage, so they started selling rather expensive co-location rights to HFTs to enable HFTs to house their computers literally right beside their exchange computers to minimize latency delays. The exchanges also started creating new order types that HFTs could use. These order types are so specialized, only people with high speed co-located access could make use of them. Orders like Hide Not Slide (quite a way from the usual Buy Limit, or But Market orders!). HFTs also make use of, what I'll call, pretend orders. They put an offer on the wire, and then when an order comes to take it, they'll cancel the order before the trade can get executed. IMHO, this should also be illegal, but isn't.

    HFTs do other things as well - they may instead focus their energies on trying to discern when a large buyer or seller enters the market for a stock. If they can sniff out such a player, then can front run those buy/sell orders and make a little bit of money on each trade.

    HFTs as a rule never intentionally take a position in any stock longer than a very little bit of time. Ideally less than a second. They are thus not investors and, if done right, they incur very little risk.

    How much money are we talking about here? On a liquid stock, they may add $0.001 to $0.005 per share. On a 100 share order, it might cost the buyer/seller an extra $0.10 to $0.50. Note that the investor never even knows they are being "charged" this amount, since all they see is a stock price that is $0.0001 per share higher than it otherwise would have been.

    HFTs have a way of insinuating themselves into every defense sophisticated investors try to throw up in their way. Dark pools have even been compromised. When the HFT offers to pay to see order flow - ie. pays the exchange, and the broker for sending orders their way, it is hard for exchanges and brokers to ignore since, frankly, investors don't realize, at all, that they are getting a slightly worse deal on their execution prices.

    In their defense, HFTs will say, yes, yes, this may all be true, but we are providing increased liquidity and thus it is all worth it. BS. It is not real liquidity when the HFTs never hold a position. I could go on with shooting down the HFT's defenses, but this post is probably long enough as it is for now.
     
  5. hershey101

    hershey101 Member

    Joined:
    Jan 22, 2013
    Messages:
    767
    Location:
    New York, New York, United States
    I have a few questions about HFT, if you don't mind.

    1. Assuming the stories of moving comm links to be closer to the exchanges in order to lose a few mS of latency are true...what can a few mS really get? Why does it matter? Perhaps if you explain the basics of HFT it will be more clear.

    Yes, so what is described in the book was true at some point, and is still true a bit w/ respect to latency. But its really not as bad or evil as the narrative makes it seem. In general speed matters a lot when your doing things like index arbitrage, or stat. arb... there are lots of trading strategies, some of which are more latency senstive than others. HFTs that are in equity land are much more sensitive than those that trade in say, FX. So for example, suppose you have events (eg. quotes/trades etc.). and a strategy which predicts based some math, machine learning, statistics etc., what the price will do in the next 10 events. Well 3 ms won't matter if each event comes in at say 250ms, but if each event occurs at say, 500 microseconds, then the 3ms latency matters a lot.
    This is just an example, but you can see how latency sensitivity is not generic to all strategies, but variable on some scale based on what the strategy is doing..

    2. If mS matter and they really give some closer HFT houses an edge, wouldn't it makes sense for the exchanges to insert a delay to level the playing field? Or if not a delay at least some limits on transactions per second? Or increase the fees/commissions such that it naturally regulates the players.

    Yea, so what the book described was a problem in 2006-2007. Since then lot of measures have been taken to inset latency, and limit quotes/transactions per second. There is no central regulation on this, so its all exchange and product specific. Idk the specific rules in equities, but I deal w/ FX, and one of the FX venues for example, only provides quote updates every 250ms, another limits the number of orders you can send per sec. In US equities, there is really no longer the problem of front-running as described in the book. Yea, some minor arbitrage opportunities may exist, but people/companies that take advantage of these opportunities have to work very hard to do so. Arbitrage isn't as prevelant as it once was, and we are really talking fractions of pennies in most cases, so its not enough for anyone to run a sustainable business doing pure arb., eg. equity arb., stock arb. (Stat. arb. is completely different concept). Its more like just side pocket change that is collected while making the markets more efficient.
    For example, when electronic trading was first introduced in the FX world, you could do something really stupid like take your money (in USD), trade EUR/USD, EUR/JPY, JPY/USD, and end up with more money than u started with by going USD -> EUR -> JPY.
    This was essentially because of the flow of information across these countries was delayed and the currency pairs weren't properly priced. So you could start with $1, do this cycle, and end with $1.0001 (or something like that). But if you do it a hundred times, with a $1M or $10M, then you get some decent money every day. Now its almost impossible to do this because the differences are so small, and the number of electronic trading firms that are out there so large, that such arb. strategies don't work. So you gotta do something smarter. But yea, in 2006, you could probably get away with something like this. Is that stupid? Yea, but the people in doing these trades, made the markets more efficient so that it doesnt exist anymore.

    3. Why don't the fees/commissions suck the profit out of HFT?

    For the most part, you get money from exchanges for providing liquidity, or if not, you can negotiate really small fees eg. 0.1% of the transaction cost or something like that based on transaction volume (eg. if you say trade $10M/day w/ them or something like that)
    A lot of strategies look good on paper, and in back testing, but fail in production because of fees. Others don't. This is why you can't just run some simulation that works and plug it into a real system and have it make money....
    There are lots of factors to consider.

    I'll also point out that during the Flash Crash, research shows that lots of HFTs were liquidity providers, they were there to buy when people were selling. (just google flash crash liquidty provision research papers). So yea, it wasn't necessarily the HFTs that caused the crash, but rather some HFTs helped support prices and bring them back to normal levels....

    Begin Rant:

    The thing that gets me is that 3 decades ago or 4 decades ago, when you called your broker and wanted to order say 1000 shares of IBM, he made $50 or $100 on that trade, and could actually front run it. When we traded in fractional quantities, those fractions were what they collected. And for what? Being somewhat good at mental math, and a smart-talker. I mean lets face it, most of kids coming out of asia now a days can do the mental math that essentially made these brokers millions 4 decades ago.
    Now you have a bunch of highly intellectual kids, many with multiple PhDs (yes thats more than 1) in areas like Physics, CS, Statistics etc. who basically made the markets more efficient with their computers so you trade the same 1000 shares for $3 on IB instead of $100 and they do so by taking fractions of pennies as their service charge (or liquidty provision). and somehow they get more bad rep than people did 4 decades ago.
    At a certain point, it is a service that these guys provide, and they should be paid for it. Now, should these highly intelligent people with PhDs be working on something other than making money like doing ground breaking research in whatever other field they are interested in, and helping society that way? Yea probably, but what do I know, I'm just the guy w/ a master's degree :p (Big Bang Theory reference). And not everyone can be Elon and save the world :) Otherwise Elon woudn't be so special.

    End Rant.
     
  6. Cosmacelf

    Cosmacelf Active Member

    Joined:
    Mar 6, 2013
    Messages:
    3,399
    Location:
    San Diego
    Yes, an honest exchange would do that. Or at least a transparent one. In fact, that is exactly what the new IEX exchange is doing (special delays) and they are doing this to try to combat HFTs. The reason no one else does is that they have become defacto partners with the HFTs. The early competitive exchanges realized early on that they needed something to draw orders to their exchange, and rather than coming up with a good differentiator, they took the easy way out and started paying for order flow, funded by the HFTs. So the HFTs pay for expensive so-location services to the exchanges, and the exchanges use that money to attract order flow from brokers. The HFTs fleece the order flow, and everyone is happy. Even the fleeced investor since he doesn't even realize this is happening.

    They are doing so. Six years ago, HFTs were making a lot more money than they are today. The exchanges and brokers got smart, realized how much money the HFTs were making, and thus started charging them more for co-location and order flow. It is akin to an oil boom town where everything from motel rooms to groceries go up in price as soon as the oil money comes to town.
     
  7. hershey101

    hershey101 Member

    Joined:
    Jan 22, 2013
    Messages:
    767
    Location:
    New York, New York, United States
    Yea, I'm not saying Gorton's defense is air-tight but:
    Front running should be illegal? Probably. Do people still front-run. I'm sure they do. But I just feel like the book makes a humengous deal out of it, and its not like its a new thing. Its just being done in a newer, subtler way.

    Also, before we start talking about speed, we should really establish what you mean when you say high frequency. Trading strategies range anywhere between milliseconds to hours in terms of how long they hold a position... So its not really fair to say "all HFTs don't provide liquidity", it depends on what range your talking about.
    Just so we are clear, I'm not saying HFTs don't have their own set of problems, I'm just saying the problems are far more subtle and not as large as what's presented by Lewis and CNBC and the marketing of the book.

    Anyway, I did the rant on the previous reply, so I'll leave it at that for now.
     
  8. Cosmacelf

    Cosmacelf Active Member

    Joined:
    Mar 6, 2013
    Messages:
    3,399
    Location:
    San Diego
    The reason why spreads are smaller now than they were 30 years ago has zero to do with HFTs and everything to do with consumer computerized trading and competitive exchanges. Broker fees were slashed when eTrade and Schwab were created. Trading commissions were slashed when Nasdaq had to compete with a dozen other exchanges.

    - - - Updated - - -

    Anyways, I didn't mean to come out swinging on this thread (probably too late to say that). I am genuinely interested in what HFTs are doing these days. Obviously, hershey, you are doing something in the FX markets, which is quite different from the equity markets. Given how little regulation there is in the FX markets compared to the US equity markets, I can just imagine the kind of stuff HFTs can do there!

    BTW, all this is interesting in a sort of theoretical way (unless you are working for a HFT), but it doesn't affect investors all that much. It is a minor annoyance at most. Gotta keep that perspective.
     
  9. hershey101

    hershey101 Member

    Joined:
    Jan 22, 2013
    Messages:
    767
    Location:
    New York, New York, United States
    Thats not true. Yea there are more exchanges but things have a primary exchange they trade on. Usually the exchange where a particular security is listed is where its traded and the most liquid. This is the primary source of pricing on a security, and where other exchanges calculate their prices from. Suppose it was due to exchange competition. Then the primary exchange could always keep a larger spread, and the secondaries would end up w a large spread as well since you dont know where in the spread the true price resides.
    Furthermore, if your conjecture is correct, why are spreads on options so high? There are plenty of exchanges where options are traded. The real reason is, pricing options is a very difficult, and involved process. Hence not a lot of HFT trading takes place in options. Only the most sophisticated and mathematically advanced firms trade options.... We barely do options and we are one of the oldest. But I mean the exchanges are still there, so if its solely the existance of exchange competition that is the cause of spread reductions than options should also trade at penny spreads.

    - - - Updated - - -

    Yea this is the key point. Smart investors, retail, or institutional aren't that affected. Its the stupid institutional investors, or the lazy ones that get their money scalped. I mean if your smart your going to write an order routing engine on your end that does the right thing for yourself, regardless of what the middle-men do. If your a retail investor, your going to go w/ IB instead of ETrade because their fees are extremely low and their routing engine is excellent. If your a retail investor you don't care about paying half a penny extra on a trade when your already paying your ETrade guy $8 for the trade in the first place.

    From my prespective (I'm a developer), the lack of regulation in FX is a pain in the a**. It means everyone does things their own way and you gotta code the same things a dozen times, once for each exchange.
    I have also done research into making strategies for FX, and I can't get into the details in a public fourm (mainly because I'm too tired/sleepy not because its against rules or whatever), but I can say its far more involved than "oh lets just move closer to the exchange, see what the other guy wants to do and do it before him".
    Its more about analyzing relationships between pairs of currencies, and seeing how movements in A, B and C will affect D, E, and F (in terms of probabilities), and determining if the impact is large enough to get over trading costs (ideally, when you buy you buy at the ask, and when you sell you sell at the ask and you still make money in the middle)... and doing all of this obviously on a fast time horizon.
     
  10. Cosmacelf

    Cosmacelf Active Member

    Joined:
    Mar 6, 2013
    Messages:
    3,399
    Location:
    San Diego
    OK, then please tell me the actual mechanism that causes price spreads to shrink from the existence of HFTs?

    - - - Updated - - -

    Let's get the prices right - front running HFTs would cost the retail investor an extra $0.10 to $0.50 for a 100 share trade on top of the $8 in eTrade commissions.
     
  11. hershey101

    hershey101 Member

    Joined:
    Jan 22, 2013
    Messages:
    767
    Location:
    New York, New York, United States
    I'll answer the first part later, but can you explain the second part? $0.10 on trading what exactly? Equities, bonds, options, forex? and which venue? US, Canada, Brazil, Europe, Asia? Are you claiming the $0.001/share number just from the book, or is there any research that supports this?
     
  12. kishdude

    kishdude P #130

    Joined:
    Jul 11, 2012
    Messages:
    136
    Location:
    Glendale AZ
    HFT is gaming the system plain and simple. How can front running any order be fair? Where is the risk?
     
  13. Cosmacelf

    Cosmacelf Active Member

    Joined:
    Mar 6, 2013
    Messages:
    3,399
    Location:
    San Diego
    Yes, trading on the US equity markets - stocks. Yes, $0.001/share came from the book. And looking back at the book, I doubt these were rigorous numbers.
     
  14. mershaw2001

    mershaw2001 Member

    Joined:
    Apr 26, 2013
    Messages:
    559
    Location:
    Menlo Park
    Thanks to cosmacelf and hersey's discussion, I have learned a lot tonight. Please tell us more- your rants are most welcome because I've never been exposed to this side of the stock market.
     
  15. uselesslogin

    uselesslogin Enthusiast

    Joined:
    Jun 13, 2013
    Messages:
    1,299
    Location:
    Omaha, NE
    It certainly doesn't cost the retail investor anything on a 100 share trade. I am sure any one exchange will have 100 shares. It is only when one is trying to buy all available shares on multiple exchanges that the front running algo works. Institutions have figured it out so they can defeat the algo in general. I'm pretty sure HFT firms are not making as much money as they were 5 years ago.
     
  16. hershey101

    hershey101 Member

    Joined:
    Jan 22, 2013
    Messages:
    767
    Location:
    New York, New York, United States
    This ^^^^.. Although I'll clarify by saying HFTs aren't making as much money as 5 years ago doing stupid/dumb things like front running, or easy-bake strategies like mean reversion... There are some more complex strategies that result in certain companies actually making more money then 5 years ago, but yea, its a lot harder and a lot more sophisticated than what you see. That being said its not rocket science (although I guess as the Elon Musk fan club, no one should be afraid by a bit of rocket science), its often just a matter of lot of painfully tedeous work in analyzing data until you come up with the right way of looking at it to get some sort of insight on the markets.
    I highly recommend reading some research papers about HFT, market microstructure and statistics/machine learning to anyone who wants to learn more about this stuff. Also check out quantnet, its a great website.
     
  17. TSLAopt

    TSLAopt Active Member

    Joined:
    May 20, 2013
    Messages:
    1,009
    Location:
    Northern Cal

    Hershey,
    I work in the industry for a firm that has been one of the major pioneers of electronic trading.
    The biggest gripe with HFTs I think should be that it competes with market makers and pushes them out of business more and more. The Flash Crash was an example of what happens when market makers no longer exist... There needs to be something done to make the playing field between HFTs and market makers level again....

    From the HFT background perspective, what do you think of these two solutions for this?
    a)make HFTs register as market makers and follow the same rules/regulations MMs have to
    and/or
    b)make all exchanges have the built in provisions that IEX has
     
  18. Cosmacelf

    Cosmacelf Active Member

    Joined:
    Mar 6, 2013
    Messages:
    3,399
    Location:
    San Diego
    Quants and classic HFT are different. Quants do put market strategies into play. If you are analyzing the market to gain insight from it, you are being a quant. If you are using speed advantage to take no significant long/short position you are being a HFT.

    hershey, I'm still waiting for the explanation of how HFT decreases spreads :)
     
  19. ElSupreme

    ElSupreme Model S 03182

    Joined:
    Jan 13, 2012
    Messages:
    4,279
    Location:
    Atlanta, GA
    Listen to the second Radiolab story from this link. It will tell you eveything you need to know about a single ms of latency.
    http://www.radiolab.org/story/267124-speed/


    This whole episode is fantastic, perhaps my favorite Radiolab.
     
  20. woof

    woof Model S #P683 Blue 85 kWh

    Joined:
    Apr 30, 2009
    Messages:
    1,149
    Thanks for that! That really helped me understand (a bit). Love the Radiolab sfx.
     

Share This Page