News

Reducing High Frequency Trading By Regulating It Less

Market microstructure is a thing that I don’t really understand and that seems daunting to me so I’ll pass this along as tentatively as possible, but: I thought this piece was really good.* Again, not my area, so if you disagree just get furious at me in the comments, but I thought it might be fun to talk about it as a parable of financial regulation.

The background here is that there is a thing called high-frequency trading in which (i) people, and by “people” I mean “computers,” (ii) trade, and by “trade” I really mean more like “post bids and offers” – they trade, too, but the activity that they’re optimizing is posting bids and offers, (iii) frequently, and by “frequently” I mean “in tiny fractions of a second.” There are various worries about this thing, of which the two biggest are:
(1) computers are scary and
(2) the amount of resources devoted to this activity is staggering and probably out of proportion to its social benefit.

Worry (1) is hard to address** but maybe you’ll be a bit soothed to learn that humans can be scary too? No, I mean, fat fingering is a problem and seems to be a bigger problem with virtual fingers but let’s just bracket that and talk about worry (2).

Worry (2) can be phrased in two different ways. There’s the disinterested snooty way to phrase it, appropriate for those, like me, who aren’t exactly trading stocks a thousand times a day. This runs something like “wow, a lot of people are buying a lot of computers and building a lot of fiber optic cables*** and hiring all of the engineers in the world basically so one guy can trade stocks a nanosecond faster than the other guy with computers and cables and engineers, wouldn’t it be nice if those engineers were like engineering cancer treatments or whatever?”

The other, partisan and alarmist way of phrasing it, is “these rent-seekers are seeking my rents!,” which is appropriate if you trade lots of stocks, maybe? I don’t really get this; I suspect on balance HFTs offer improved liquidity to real-money market participants, but sometimes hose them in particularly hurtful and confusing ways, with those hosings perhaps especially concentrated at the worst possible times.

Anyway though former HFT guy Chris Stucchio tells the story in the first way, as a story of rent-seeking, and he isolates the rent that is being sought in the SEC rule prohibiting stocks from trading in increments of less than a penny:

The subpenny rule essentially acts as a price floor on liquidity – it is illegal to sell liquidity at a price lower than $0.01. … As with a classical minimum wage, two parties are harmed – the purchaser (who must pay extra) and the lower priced seller (who is pushed out of the market).

Similarly, at prices higher than $0.01, it makes price movements lumpy – on a bid ask spread of $0.05, it is illegal for someone to enter the market at price $0.049 or $0.045. Thus, at any price point, speculators are forced to compete on latency rather than on price. Price competition is only possible if one market maker is willing to offer a price at least $0.01 better than another, which is often not the case.

When price competition is impossible, market makers must compete for business via other methods – in this case latency.

His proposal is then to eliminate the subpenny rule, and lists some advantages and disadvantages****:

The first benefit is that the bid ask spread is likely to narrow (probably by about half a penny), thus making it cheaper and easier for speculators to ply their trade … [which] benefits price discovery (and presumably society in general). Realistically, this is probably not a major benefit.

A more significant benefit would be diverting labor from the latency arms race to more productive purposes. When I worked in HFT, my coworkers were extremely smart people, capable of doing many valuable things. I don’t believe the best use of such people’s labor is in reducing latency to make one HFT trade faster than another. While I don’t generally like to second guess a free market, I believe in this specific case the market mechanics line up in such a way as to make the most socially optimal option not individually optimal.

Coincidentally, he is not the only person thinking about slowing down the HFT arms race – yesterday David Merkel had another suggestion:

I have my own solution to high frequency trading: revamp all markets such that there is one auction per second in the trading day. Auctions happen at the top of each second: 9:30:00.000000… 9:30:01.000000… … 16:00:00.000000. Additionally, orders still standing at the start of any second may not be cancelled for the next second.

Auctions once per second. Click, click, click, click … 23,401 auctions per day offers more than enough flexibility to buyers and sellers. No truly economic commerce would be hindered by such an arrangement.

So if you like the idea of tamping down HFT, you could either eliminate the $0.01 minimum increment, or add a 0:01 minimum increment. How do you evaluate those proposals? Here is a Tyler Cowen post from three years ago that is skeptical about regulating HFT out of existence. It is short and all of it is good but let’s focus on this:

The more I read these debates, the more nervous I get about the idea of a financial products safety commission. Essentially on innovation we’re seeing a flipping of the burden of proof and I don’t think it is possible to easily fine-tune that flipping in a way to capture good innovations and rule out bad ones.

Right? How do you know the socially optimal amount of liquidity? How do you know if an auction once per second is the right number of auctions? It strikes me as a question not particularly capable of a priori answers though, I mean, that is a lot of auctions.

Similarly, how do you know that the right price for liquidity is $0.01 or an integral multiple thereof? Again, that doesn’t seem susceptible to a priori answers but you can sort of wave at interesting empirical answers by noticing that there are fewer stock splits and higher share prices these days, and by taking this chart from this Alphaville post as some evidence that there is a natural price for liquidity and it is not an integral number of pennies:

But that’s not the point. The point is, the subpenny rule reduces price competition and adds a relatively arbitrary regulation for traders to game: because you can’t be outbid by another bidder within the same penny increment, you get free money by just getting there first. One way to compensate for that waste might be to add another arbitrary regulation: if people are inefficiently competing to get that $0.01 at 9:45:00.5, because they’re forbidden from trading for $0.009, why not forbid them from trading at 9:45:00.5 and make everyone bid fairly at 9:45:00?

It’s not obvious that that’s a bad solution. But, compared to getting rid of the subpenny rule, it’s an inelegant solution – compensating for the ill effects of one arbitrary-number-based regulation by adding another arbitrary-number-based regulation – if you could achieve the same goals by just getting rid of both of them. (Again, there are probably ill effects of getting rid of both of them that I don’t know about, so commenter fury much appreciated.)

There are definitely lots of ways to go too far***** (and incentives for doing so) in claiming that the cause of all financial problems is financial regulation. But sometimes it’s a useful place to look. There’s lots of research suggesting that our financial system is not that great at reducing the cost of financial intermediation; part of the reason for that may be that the people involved in our financial system are really really good at focusing on regulatory regimes and finding ways to profit off of them.

Sometimes we celebrate that here: the intellectual daring, rigor and artistry deployed to game complex regulations is, when viewed in a certain light, breathtaking. And – this is just a subjective impression here – I suspect that the regulators rarely bring the same artistry and dedication to the game. Which is why, when there’s a possibility of achieving regulatory goals without playing, it’s probably worth considering.

High Frequency Trading – What’s broken and how to fix it [Chris Stucchio]
Should Stocks Trade in Increments of $.0001? [MR]
Stock splits and the volume-price correlations [FTAV]
Fewer Stock Splits, Record Share Prices [BW]

* It was in Write-Offs the other day but I was re-reminded of it by this good Marginal Revolution post this morning.

** If it worries you a lot, have I got a book recommendation for you! I did not love Robert Harris’s The Fear Index but if I were to give you a plot summary – which I won’t because it would contain spoilers – you would laugh really really hard because THE HEDGE FUND COMPUTERS ARE COMING TO GET YOU.

*** But!

**** An interesting possible disadvantage that he raises:

The other criticism is that it might be more difficult for traders to evaluate the markets. Instead of merely looking at the number of shares available at $10.00, a trader might need to add up the number of shares available at $10.0000, $10.0003, $10.0015 and $10.0029. In my view this will merely require minor modifications to behavior and the tools available – for example, trading programs might merely display a price of “no worse than $10.0029” which aggregates the prices between $10.0000 and $10.0029, and traders will need to look at cumulative price views rather than ladder views.

As someone who sometimes reads about stocks living in continuous time, this suggestion pleases me, though I suspect the technological fix is pricier than he lets on.

***** Start at like 3:15; it’s pretty silly.

73 comments
(hidden for your protection)
Show all comments

73 Responses to “Reducing High Frequency Trading By Regulating It Less”

  1. Im_a_Dude says:

    finally a picture of Matt's office.

  2. Guest says:

    How many trades can be executed by a HFT platform in the amount of time it takes to read this post with all source material included?

  3. Guest says:

    "Again, not my area, so if you disagree just get furious at me in the comments…"

    Should have said "Again, not my area, so if you disagree just get furious at me in the comments, where I will offer no defense"

  4. Also furious says:

    Second

    – Guy who really wanted to make 'Furious' comment in post but didn't get there in time

  5. Deleveraging says:

    I forced myself to read this post as punishment for having 3 slices of Pizza at lunch. As always, love the footnotes.

    Downloaded the book onto my Kindle.

  6. Guest says:

    If you connect all those dots, and then color it in a little, that appears to be a cornucopia spraying semen everywhere.

    -Its not magic, its science.

  7. ILoveLamp says:

    But what happens when they get rid of the penny altogether because it costs more in copper than it's worth? What happens to your little theory then, huh? Huh? Yeah, I thought so, bitch.

  8. getmeadrink says:

    Its only the main exchanges that have this rule right? last week I bought a bunch a market and got fills at 17.7583, 17.7587, 17.7586, 17.759,17.7585. WTF. its worth doing a market order just to make CGT forms simpler.

  9. BigTrade says:

    The big problem with HFT isn't price increments, it's the lifetime of a bid/ask, where trades go through on bids/asks that haven't gotten to all exchanges, therefore rendering the NBBO useless.That's how HFT make money, by seeing the market a fraction of a second before everybody. I guess a solution may be to have auctions at the top of each second and each bid/ask has a lifetime of at least half a second…

  10. Female HFT says:

    I fully enjoy it when you are in my space.

  11. evo4242 says:

    Reducing the tick size would certainly impact the strategy used by market makers, but ultimately they would adapt and still use technology/speed to gain an edge. Right now, they use technology to find ways to get to the front of the queue (ahead of retail traders) for passive orders, and to be faster than retail when sending aggressive (liquidity-taking) orders. Changing tick size would only impact the former edge, as there would be no queue to sit in front of. But what would happen is that algos would be fighting to become best bid/offer by changing price by $0.0001 or whatever was the minimum increment. They would be sending and cancelling order in milliseconds virtually continuously. So a manual retail trader would almost never get his limit order filled until the stock blew through his price — which is pretty much the same situation as now. Additionally, the exchanges would probably have great difficulty mananging all the increased order sending/cancelling.

    I agree that such a change would benefit people sending market orders in highly liquid low-priced stocks (say, SIRI), but it would actually hurt limit orders, as algos would always be able to jump in front of them without substantially improving the price.

    I suppose the theory is that competition among algos would make HFT generally less attractive, and most of the players would pull out. That could happen, esp. if volumes in general plummeted. But overall, introducing a system that rewards real-time data processing (the ability to read and react to constantly flickering price changes of hundredths of a penny) will likely increase HFT's advantage vs. retail (esp. limit order users), not reduce it.

  12. Avid Iconoclast says:

    Great stuff, Matt! Why couldn't they get rid of the penny rule AND institute auctions? Wouldn't that get rid of the "fighting over nanoseconds" issue while also helping liquidity?

    I think I might actually prefer reading a Matt post to a Bess post these days. In a good way!

  13. Guest says:

    The only thing that Exhibit 9 shows is the sub-penny rule in action! 1bps on $100 is 1 cent. 2bps on $50 is 1 cent. 4bps on $25 is 1cent. This chart shows that the most-liquid names are hitting hard up against the sub-penny rule wall and probably want to go much lower.

  14. Admiral Mike Taylor says:

    Matt,

    Fleet Week is here again. Looking forward to seeing you down by the docks in that pretty little dress. Should we meet at our usual place? xoxo

  15. Guest says:

    Do you think Matt gets himself jacked up by saying "I'm going to Mattack some graphs today!"? Because I want to live in a world where that happens.

  16. The Dealer says:

    Yeah, let me see here….first the mantra was was we need to move from 16ths to 100ths to increase liquidity. Now we have less visibly displayed liquidity at each price points and off exchange executions that take place at sub-pennies. Whatever benefit to the investor was generated by the decreased spread has been lost in the costs to attain lowest latency and increased spreads in less liquid stocks.

    Listen, in your top 100 or so stocks with low spreads and tremendous volume, speed and small spreads have been beneficial. However, when you try to buy trade in a low volume issue your paying a spread that is greater than you did with an old school market maker, even if he did scalp you for a few cents.

    You can thank your regulators for regulating away the participants that stabilized markets. Today you have a bunch of algorithms that only play when they want to and will drain liquidity as soon as CNBC broadcasts a protestor throwing a molotov cocktail in Athens or Lisbon.

  17. guest says:

    Matt, your summary of the article is, as usual, significantly more comprehensive (and more masturbatory) than the entire body of works of the original author, who at one point actually had something valuable to say. I award you no points, and may God have mercy on your soul.

  18. Duncan says:

    Articles on market microstructure should ideally be written by people with some idea of the structure of the market.
    At least, you could find someone who knows something about the U.S. markets to reference. Is that too much to ask?

    There are so many things tarded about Stuccio's essay I hardly know where to start, but let's start with the argument that HFT's are socially wasteful.
    No, let's start with him completely ignoring the effect of maker/taker pricing on the market.
    No, let's start with penny stocks actually being priced in subpennies.
    No, let's start with 30% of trading in NMS securities occuring off-market at subpenny prices and with no quoting.
    No, let's start with consideration of the latency in the SIP, and the limitations of co-location when there are multiple market centers.

    Give me a chance to stop hyperventilating and I'll start with one of those.
    HFT's aren't a problem; they have substantially lowered the transaction costs of your 401k manager, even if you don't trade yourself.

  19. 401k manager says:

    Your 401k manager does some research on a stock and figures out its fundamental value is $35. The current trading price is $30. A passive investor needs to sell some of that stock for liquidity reasons, and has a limit order to sell at $30. After the 401k manager executes a few buys, the HFT figures out what is going on, takes all the remaining inventory from the seller at $30, and drives the price up to $35. As a result, the HFT captures about a $5 spread, the passive investor is no better or worse off than before, while the 401k manager has no remaining alpha and 401k investors are left with no alpha. The HFT free rides off the 401k manager's efforts to determine fundamental value. In equilibrium, 401k investors have less reason to try to identify fundamental value, and so less information becomes incorporated into markets, i.e. markets become less efficient. Investors need to be able to capture rents from information gathering for information gathering to continue to occur. HFTs free ride off the information gathering process and reduce alpha available to average investors.

  20. Duke of URL says:

    disinterested = unbiased, impartial
    uninterested = not interested

    Learn proper English you jerk

  21. Finkle is Einhorn says:

    90% of comments come from Matt today or something?

  22. jerk nut says:

    1. try midpoint peg orders u fuckstick
    2. the real arms race is the es-spy basis trade, $MM to go from NJ to CHY in under 8ms
    3. the HFT firms are gonna fuck themselves anyway, as they have spent $10+MM to get the "fastest" shit, only to find out that volumes in equities and all CME produtcs is down 20%. oops

  23. obvious says:

    Brain surgery is a thing that I don’t really understand and that seems daunting to me but I will anyway write a long piece about the subtle intricacies of it.

    First off, brains are pretty big and also I guess they say that most people never use more than 10% of their brains so there is a large margin of error for you when you do a surgery. Now another thing I know is that the brain is in this skull thing so you'll need like a saw or something or at least a drill. If you use a drill, make sure that you are drilling over part of the unused 90% in case the bit goes in too far.

    <more rambling musings and heuristics that have nothing whatsoever to do with the reality of the highly complex subject of the post>

    So really I think that the laws about brain surgery need to be relaxed and it is amazing what you can do with a sharp scalpel and some carefully applied novocaine.

  24. Guest says:

    Note that some economists have proposed increasing the tick – not to one second, but to one day. I think it is a serious question as to what the utility to society is of allowing intra-day trading. Given that most investors check their investments, and the news, daily, the people who are primarily served by intraday trading are – not surprisingly – traders. But that comes at everyone else's expense. It could be done without regulation, simply by setting up an exchange where stock trades only once a day, by auction. A radical solution, to be sure, but interesting to consider.

  25. Guest says:

    Its like a nerd slap fight up in here.

  26. Bubbagump says:

    Captive order flow and regulatory arbitrage through fractured marketplace and invisible orders. If you want quotes from the HFT firms saying this you can just search them up. They make their money because they can fade and trade ahead better than you can. Other ones make cash due to large amounts of captive flow. Stop yapping HFT about crap noone here seems to know about. Profits are way down over last 2 years, theres a long list of firms already under, some of them are now selling their algos to buyside. HFT is getting screwed because the guys who aren't quite fast enough or rich enough to buy location or flow are selling execution services to smart investors. HFT and payment for flow as well as dark pools have killed the traditional market maker. Canada at least is looking at the value of requiring price improvement on this, but until we dump and no one is there to catch the knife, no one is going to care.

  27. HFguy says:

    "Market microstructure is a thing that I don’t really understand"

    I shrudder to think about the size of epitome if you did understand
    L. Tolstoy

  28. IMO: If one-cent increments and/or timed auctions are needed to fix the problems arising from HFT, so be it. The incentives to cheat, sometimes at TBTF levels, will drive brilliant people with vast resources to exceed the capacity of any non-fascist regulator to effectively oversee. Structural modifications are required. Surely we can agree that markets are made "imperfect" by many mechanisms — in this case, the value(?!) lost to "friction" ought to be viewed — perhaps with regret –as necessary measures. Market perfection is a straw man. Why demand of financial markets when we accept sometimes-draconian mkt imperfections in other commodities? Philip Stern

  29. Good day! I could have sworn I’ve visited this site before but after browsing through some of the posts I realized it’s new to
    me. Regardless, I’m certainly happy I stumbled upon it and I’ll be book-marking it and checking back regularly!

  30. Hi there, the whole thing is going well here and ofcourse every one is sharing information, that’s genuinely excellent, keep up writing.