Wednesday, February 13, 2013

Why HFT is Not a Problem

Many people are concerned about the fact that some people can trade more frequently than they can due to having a large infrastructure of high speed computers and connections to various exchanges.  It all seems like regular people can get ripped off, and so some kind of governor is need via transaction taxes, or some limiter on the trade frequency.

This is all unnecessary. If you are concerned about being gamed by high frequency traders (aka HFT), there's a pretty simple way to avoid this.  Simply trade using Volume Weighted Average Price (VWAP) orders. This order gives you the volume weighted average price for a period, usually a day, and on InteractiveBrokers it's available pretty cheaply, say for half a cent a share.  Brokers have whittled this fee pretty much down to the exchange fees and spread, taking into account the fact that they trade these things 'algorithmically', so you actually benefit from HFT as they aren't simply and stupidly crossing the market every half hour. Most HFT actually are simply common sense applied to generating a light touch by trying to emulate a market-maker (eg, posting buy orders at the bid) as opposed to something sinister.

Say you decide on Sunday, after doing all sorts of research over the weekend, that you want to buy IBM on Monday. If you are buying less than 1% of ADV, your effect on the price is positive to be sure, but spread out over the day, not quick enough or with a high enough certainty that one would want to front run you (ie, it's not sufficiently greater than the spread to be worth buying in the morning and selling at close). Further, any information that might give a HFT a millisecond advantage within they day will be randomly because you can't predict the sign of such innovations hours in advance, so HFT shenanigans are irrelevant to you, they aren't biased in any direction.

Widows and orphans can execute VWAP orders now, and then benefit from any benefits in a perceived case where trading is simply done at a central auction once a day. Yet with this solution no one is constrained, no one has to monitor or regulate anyone. The solution exists currently and is available to everyone who want to avoid HFT.  Real investors shouldn't be day trading, but making longer term decisions based on a longer term analysis of the trends, and in that case they can trade as if HFT didn't exist without any new rules.  

38 comments:

Tel said...

Correct me if I'm wrong, but VWAP is not available as a standard exchange order -- you need to hire someone else's HFT engine in order to get access to this type of trade method.

In effect, you are just paying protection money to one HFT algorithm so you are safe from other HFT algorithms. OK, if you can't win the fight yourself then protection money is not the worst choice, but hey, I'm paying fees to the exchange in the first place. Why do I have to pay multiple parties here?

What's more, the arms race continues because implementation of VWAP might not in practice be as perfect as the theory suggests. A poor implementation then leaves a small chink in the armour for other HFT algorithms to dig out their profit. So then the retail traders are left shopping around for protection -- one again, better to pay fees to the exchange, especially if there is an easy way for the exchange to block such behaviour.

Eric Falkenstein said...

I'm not clear on two points.

"I'm paying fees to the exchange in the first place": how so with VWAP? You don't pay an exchange fee to execute a VWAP order, the broker bundles that into his fee.

A poor implementation then leaves a small chink in the armour for other HFT algorithms to dig out their profit." I'm not sure I see how this hurts you. If VWAP was 201.13 for IBM that day, that's what you pay, plus 0.5 cents or something to the broker. It doesn't matter what they ended up buying it for (eg, 201.09) on their particular implementation. It only hurts you if, by sending an order, you think you will actually increase the price by more than the 'old' way one used to make these trades, which is implausible.

Mercury said...

Many *more* people are concerned about the fact that some people can utilize expensive technology to see orders before they are publicly displayed, manipulate price/time priority and overwhelm market infrastructure with massive quote volume for the sole purpose seizing up the order matching process or spiking the price.

When things are running smoothly it is true that your average retail trader typically won’t be negatively impacted by HFT although A) yes you can attract the interest of HFT algos with an order size well below 1% of ADV and B) if enough order flow is collectively doing the VWAP thing there are ways for algos to game and manipulate that too (and the broker to whom you gave your order is in the best position to effect that). Of the hundreds (thousands?) of order types now available (an absurd situation in and of itself and more or less the definition of regulatory capture) there is no order type that guarantees VWAP and no broker will guarantee you VWAP unless he’s sure he can rook you some other way. It’s all on a “best efforts” basis. But no, if Joe Retail places an order to buy 200 IBM and gets filled 11 cents above VWAP it’s not the end of the world.

Also, once prudent stop-loss orders can now be suicidal as a sudden, HFT-induced “flash-crash” can cause you to dump your long-held position at much lower prices even if the stock rebounds back to where it was a second later. Then you’re at the mercy of regulators, hoping they will wave their wand, grant a “do-over” and erase the swath of trades over that time period.

On the retail side HFT’s negative impact is ultimately one more of perception than actual harm. On the institutional side it’s more complicated but overall I think HFT in combination with market fragmentation has reduced liquidity and overall trust in the market. And no, yesterday’s volume is not synonymous with today’s liquidity which has a habit of suddenly drying up in an HFT dominated market when things get ugly.

Since you can’t alter the laws of physics or put the technology toothpaste back in the tube there aren’t many simple ways (as opposed to a rat’s nest of regulations) to restrict HFT. One of them however might be a tax that would be small enough to be a de minimis expense for retail orders, not a big deal for most institutional orders but prohibitively expensive for the most pernicious forms of HFT. Why not implement a small fee on cancelled orders (as opposed to transactions)? That would certainly put a damper on many HFT shenanigans but I fail to see how it would have much of an impact on market liquidity (or any other desirable quality) which is HFT’s dubious claim to fame.

I think there are reasonable arguments that a non-luddite can make against at least certain types of HFT trading activity. At the bottom of this page there is a restriction labeled “Please prove you’re not a robot” which is there to prevent certain types of commercial activity which are legal but don’t add value to this venue. That is pretty close to the spirit of my stance on HFT.

Finally, my appologies...this is not intended to be a "speed and spread" argument.

Anonymous said...

You are aware of course that as soon as these orders are entered they cannot be cancelled and may end up costing more at the end.

Eric Falkenstein said...

Merc: I've done thousands of VWAP orders, and they are basically at VWAP, not biased. They make money via the fee I see, and which has been competed pretty much down to their all-in cost for the broker. The only way this hurts me is if I think that this kind of order has a greater trade impact (ie, moving the VWAP adversely) than alternatives now or in the past (eg, when you used to call up a broker). I don't see any evidence of that, and I monitor that closely.

As per liquidity drying up when you need it, that has always been the case (remember 1987?), so the question is, is this more common? I haven't seen any evidence for that.

Adding Tobin taxes or captchas just would make it less competitive, and trading now is a lot cheaper than before the bots, as brokers used to charge 1/4 on Microsoft in the 1990's, which then was a large stock. Further, commissions in the old days were much higher too. Everyone takes spread compression and lower commissions for granted, but it's because of competition.

Anonymous said...

The fact that HFT is good at trading high volumes quickly is not a problem. However, allowing sub penny trading gives them an unfair advantage over retail investors. Today basically the only way to get filled on a limit order is for the market to trade right through you.

Either sub penny trading should not be allowed, or it should be available to everyone.

Anonymous said...

Ok, you're wrong, VWAP is on a best-efforts basis and by no means guaranteed. Please support your claim that brokers will give you the day's VWAP. If you do a VWAP order, over 30 min, it is often not the same as the actual VWAP at that time. I've also don't thousands of those trades as a fund manager...

Eric Falkenstein said...

Their best efforts are not biased, ie, the average of (VWAPbesteffor-VWAP) is not biased for buys or sells, that is, it' not 1 penny higher for buys on average, or hasn't been for me.

Anonymous said...

I misunderstood your prior statement, thanks for clarifying. Love your blog, thanks.

Mercury said...

EF: I guess I’ll defer to your current assessment of VWAP trading since it has been a couple years since I’ve placed large VWAP orders myself with any kind of size or regularity. VWAP is such a benchmark now that it makes sense that competition has produced a way to get very close to that price cheaply. But I think it’s also easier to pull off in a slow and steady, low-vol market which is what we’ve had recently. I would still be inclined to stay away from this strategy for less liquid names and/or a high vol market.

In an HFT dominated market I think there is a much smaller percentage of trading activity that has any connection to valuation (correctly or incorrectly perceived). So, when the market plunges and goes crazy the smart move for most HFT algos is probably to shut down and wait for a more comfortable environment. An HFT algo isn’t going to be the market participant that says: “Holy crap! IBM at $100? I can’t NOT buy here!” Ideally perhaps you’d want all the “traditional”, valuation and speculative capital plus all the HFT capital in the market together but I think the later has driven away the former to a considerable extent. The next big market correction will generate some important data in this regard one way or the other (as to how fickle liquidity is).

Some sort of HFT barrier or disincentive might diminish competition….but competition for what exactly I’m not so sure. An algo based around rapid-fire quote stuffing and order cancellation just doesn’t seem to overlap well with the concept of price discovery as it is commonly understood. And spreads and commissions have declined because of technology and disintermediation in general, not specifically because of HFT.

At this point, for both retail and institutional investors (which I am not ashamed to say I think market structure should be biased toward), lower and lower transaction costs yield less and less incremental benefit. Almost all the incremental cost savings is realized by high turnover strategies and of course HFT is high turn to the max. If HFT is in fact “pulling its weight” by bringing more net liquidity to the market that’s one thing but if not I don’t see why it should be accommodated much less allowed to write the rules.

I’m not sure what your exact niche is but I don’t really have a beef with program trading per se.

Unfiltered said...

HFT trader here. I trade derivatives on a foreign exchange but am pretty familiar with the issues in U.S. equities. In general it seems to me that people criticizing HFT are comparing it to some kind of fantasy world in which all trading happened at mid-market, for fair value, with nothing paid away to market makers. And they seem to think that a transaction tax or other speed curbs would take us back to that world. This kind of underlying premise is revealed in inquiries like, "I'm paying fees to the exchange in the first place. Why do I have to pay multiple parties here?" Uhhhhh....

Furthermore, competition in the HFT world is very intense and the profits earned by the large algo trading shops are way down due to reduced volatility and trading activity, so the idea that there's some scam being pulled and algos are being used to rake obscene profits out of the markets doesn't really square with the recent financial performance of computerized trading firms.

@Mercury
"Many *more* people are concerned about the fact that some people can utilize expensive technology to see orders before they are publicly displayed, manipulate price/time priority and overwhelm market infrastructure with massive quote volume for the sole purpose seizing up the order matching process or spiking the price."

If you're referring to flash orders, these make up an exceedingly small portion of equity market volume, and are used entirely at the discretion of people placing them, usually to allow them to execute a large amount of size in a name quickly while reducing slippage.

I'm not sure what you mean by "manipulating price/time priority" unless you are referring to the use of some of the more esoteric order types, which I would agree may have gotten out of hand on some of the exchanges. (This could be one area where a bit of SEC regulation could do some good: they could draw up a list of, say, 20 order types and make those the only ones allowed across the board.) But every order type is available to every trader on an exchange, and the rules are applied to them equally regardless of who placed the order, so I don't see the unfairness there.

The whole quote stuffing thing is a bit of an overblown concern. I think most of the crop circles Nanex is so fond of publishing are just cases of test algos being put into the market with something slightly off, which causes them to enter order placement/cancellation loops for a time. I guess it would be possible for someone to make money by adding latency at one exchange by spamming it with requests, but I've never heard of it being done. The CME definitely comes down very hard on excessive messaging volume and has all sorts of policies in place to discourage it.

Unfiltered said...

@Mercury
"The next big market correction will generate some important data in this regard one way or the other (as to how fickle liquidity is)."

We already basically had a test of this in the flash crash. Liquidity was restored within a matter of minutes. Was there a point in the pre-computerized history of markets where it would have been restored in seconds, or guaranteed never to have disappeared in the first place?

"An algo based around rapid-fire quote stuffing and order cancellation just doesn’t seem to overlap well with the concept of price discovery as it is commonly understood."

Setting aside the loaded term "quote stuffing" for the moment, the problem then is with the common (an apt adjective in this case) understanding of price discovery. I mean, it just seems kind of clear to me that competition for market-making profits across smaller grained ticks requires a higher number of cancelled vs. filled orders. Why exactly that is ipso facto a bad thing has never been clear to me.

@Anonymous
"Today basically the only way to get filled on a limit order is for the market to trade right through you."

This is simply wrong, for starters, but if it were true then how would the HFT market makers ever make any money? Could it be because they have a better handle on value over the short timeframes on which the difference between being filled at the bid or the offer is relevant?

I don't get it, on the one hand, it's supposed to be "wrong" to make money on split-second, miniscule changes in the price of a stock. But on the other hand, we're supposed to worry about whether retail & institutional investors get their fills at the bid or the offer in 1-cent wide markets?

Mercury said...

Unfiltered:
Check out “hide and light” order types in particular for examples of how superior, protected quotation queue positions can be place-held and then reclaimed after sliding out of sight into “hidden” status. I suppose you can consider the explosion of esoteric order types since 2007 (many designed to get around Reg NMS and whose precise features are hard if not impossible to find explicitly stated anywhere) as competition in action that will ultimately lead to “superior” order types (whatever that might mean) but really, how are hundreds (instead of the previous handful) of order types now available suddenly such a boon for the market? Knowing nothing else it should be obvious to an outside observer that such a situation is just an invitation for abuse and trust erosion.

After decades of technological progress the market has become less transparent since Reg NMS and the HFT that had grown up around it and again, it seems hard to consider that a good thing.

Paul Revere said...

Do you have ANY proof to your claim that most HFT are simple common sense
algo's? Any data to back up your OUTRAGEOUS claim that VWAP is all the
insurance you need against ad algo's?

it is writers such as yourself who are creating problems for those of us who
understand the dynamics caused by nanosecond trading computers, expert networks,
and a newly fragmented market where each exchange has its own RLP and complex
fee structure.

First off, VWAP won't do anything against a locked market or queue slide.
Second, VWAP won't help when the connected HFT firm (GETCo, VIRTU) are able to
flip the fee from the taker side and make the maker pay it, all the while
partially filling lots and offering sub-penny pricing.

If you are ok with ignoring bad algo's and not enforcing the rules we have paid
the SEC to come up with and pass, then I expect you'll turn your head over when
your house is robbed.

The problem is that widows and orphans you talk about are not making their own
trading decisions. Some MBA, CFA jibroni is sitting his office making these
trades and poo pooing HFT saying VWAP creates an environment where regulation
isn't needed. And what does he care, he gets his 1% and bets with your money,
all the while ignoring HFT and marginalizing his fiduciary responsibility.

Furthermore, you can know when HFT will be most prevalent. The fact that you
state the opposite shows your lack of research into this area as bad algo's and
the negative effects introduced by them are visible on energy releases, major
economic releases, European close, cash outcry close, and during the last hour
trading when they are rebalancing.

Please stop writing about HFT. Your limited exposure and understanding is
skirting our effort to bring retail back to the market through the
reestablishment of trust and a regulator who actually regulates and doesn't cut
corners by buying surveillance software from the very firms they seek monitor.
Thanks for nothing big guy!

Anonymous said...

I am thinking about considering a career in the trading of stocks using the computer.

Gabe said...

Easy there Paul –
I think everyone here is making a good faith effort to assess this issue based on their own knowledge and experience. Retail investors remain skittish about the stock market for a whole host of reasons and HFT is probably not a primary concern for the vast majority of them.

Anonymous said...

Unfiltered said:
"@Anonymous
"Today basically the only way to get filled on a limit order is for the market to trade right through you."

This is simply wrong, for starters, but if it were true then how would the HFT market makers ever make any money? Could it be because they have a better handle on value over the short timeframes on which the difference between being filled at the bid or the offer is relevant?"

HFT market makers make money by frontrunning my $10.00 bid and filling a market order at $10.0001. I can't trade sub pennies, you can, why? I have no problem with your assertion that you have an edge in short term valuation. And as both a retail investor and professional trader yes I am worried about at what price I get my fills. I'm not surprised you're not so worried.

Eric Falkenstein said...

Well, a lot of assertions, and I don't have the time to address them. But the key to my argument here is that the essence of market making shenanigans is front-running orders, or the worry that they have a speed advantage that is unfair. In the front-running case, their ability to front run a VWAP order is pretty limited, much more than any intraday order type, or any order type pre-internet. In the micro-second advantage scenario it's all irrelevant to a day VWAP, because your trade presumably isn't based on that information, and it will be randomly distributed around zero so on average won't hurt or help you.

Anonymous said...

Everyone has a speed advantage or disadvantage compared to some other market participants. That is impossible to change and not inherently unfair. All I'm saying is let's apply the same rules to all participants. If you can trade at sub penny levels, I want that ability to.

And in many scenarios VWAP is not an appropriate order type.

Anonymous said...

A lot of people seem to not know what frontrunning is. No, nobody is front-running you. This is certain.

Also, what is the counterfactual to the ridiculous claims people like Paul make? Has there ever been a world in which you could freely reap the rebates and nobody else wanted in on that game? Of course not. Would we be in such a world if HFTs were suddenly extinct? Of course not. Is the rebate even relevant at all for the long-term investor when compared to the size of price impact? Of course not.

All it comes down to is entitled people whining because their mediocre intraday strats got outcompeted. Or even worse, entitled ex-market makers whose golden goose was beheaded by a smarter, faster HFT...let's not even get into the myriad of ways they screwed over investors by frontrunning, colluding, etc.

Eric Falkenstein said...

Haim Bodek. It's like Bob Rubin complaining about crony capitalism.

Paul Revere said...

I'm sure you're a good guy but your information and your belief that you're right are wrong.

Haim was an automated trader, he didn't have special order types, he didn't stuff exchanges, he didn't lobby for the industry. Not all HFT are bad, just the ones that skirt the rules and destroy market integrity while ripping off our fellow citizens. Haim did not operate his firm the way we see most people operating them today and certainly not in the way you are implying. I know this because I know him.

Nothing is wrong with automated trading or electronic markets. The problem is with special order types we can't use, a total lack of oversight, and the buddy system being used by the HFT lobby and the SEC. Haim is a prime example of how those expert-networks ruin careers but this gets lost when people make claims such as "Haim Bodek. It's like Bob Rubin complaining about crony capitalism" without ANY way of backing it up, forcing people like me to spend time explaining things are that public knowledge and detracting from the substance of what I was saying.

I was addressing the misinformation that laid in your Widows/Orphans in the last paragraph. VWAP order won't stop you from being sub-pennied or from having your lot partially fill. The clear statement you made will give lesser minds a false sense of security and that is no way to treat our fellow citizens.

Using Hide-And-Light, Hide-Not-Slide, and Inter-market Sweeps and Locked-Market are front running when your order is passed over by the internalizer. This about holding up and enforcing the rules passed by the entity we all fund with our taxes. It's very clear and shouldn't be challenged. I am supporter of guns but giving a gun to every person won't prevent murder, the rules (laws) must be enforced. Just because guns are available to everyone doesn't mean their safe, in the same way having a VWAP order doesn't mean your safe and that regulation is not needed.

At this point, your order has identifying characteristics which super fast computers analyze to determine how much you're buying, if you cancelled, what you lot sizes are, etc. THAT'S FRONT RUNNING. And if you don't believe me, scroll all the way to last comment on this post and you'll find all the information that supports me: http://tinyurl.com/Casey-Report-HFT

This information is part of the "Guaranteed Economics" between HFT, brokerages, and exchanges. HFT get what they want and they generate their volume at the location who supplies them with the structure and order types required to be profitable in an environment where prices of real-world assets are priced by machines that interact at a speed far beyond the ability of our human minds.

Notice that since going public with Special Order Types, GETCO's revenue released this week shows they earned 82% less in the nine months ended September 30, 2012 compared with a year earlier and has been on steady decline since its peak 1 year after the introduction of REG-NMS: http://tinyurl.com/GETCO-Revenue

Furthermore, VWAP won't do anything when someone is running Trade-Throughs, covered in rule 611 of Reg-NMS. The SEC admitted in the 500+ page Reg-NMS document they can not enforce because of the latency inherent in technology and space-time travel. TM was formed in 2007. He wasn't there when the HFT lobby pushed the exchanges to form the structure that would be the foundation for the insider HFT firms.

In a due time he'll be vindicated and I assure you I'll be back here to remind you. Thank you for responding to my initial comment. Please refrain from slander unless you have facts I can check to back it up, nobody wins when we behave like that. Good luck entering your VWAP during the next flash crash when the price you see is 5 minutes stale.

I hope you respond to this as I believe this comment section on this post will give people enough information to draw their own informed conclusions.

Paul Revere said...

@Gabe Comment noted. I get testy, which is not something to be proud of, when pieces like this are released. I see this stuff all day. I make money off the dislocations caused by Reg-NMS. I don't believe only honest people work in finance and I do not believe our regulators can protect us. They ignore Congress insider trading yet within 36 hours they seized the accounts of the Heinz insider traders. When I read pieces like this and think that my nephews could read it an gain a false sense of security I, very childishly, get over zealous with my attacks. To everyone here, I apologize for my tough approach.

Unknown said...

HFT used to be exotic and reserved for only a few large hedge funds. The HFT programs would try to detect very short-term trends or anomalies and profit off of them. Today, most of HFT is directed at the execution of trades (that is you have a trade idea, perhaps based on fundamental information, and you use HFT technology to buy or sell at the best possible price). Many of these algos are off-the-shelf and can be purchased by smaller investment funds. The price of the technology will decrease and it will become more and more accessible. The goal is to be able to buy or sell - no matter who you are - at a fair price. Seems like a reasonable goal to me. Cam Harvey

Unfiltered said...

@Mercury

I think there's a point where some of these order types start to blur the line between providing flexible order functionality and actively assisting HFT firms in reducing latency by moving parts of their algorithms into the matching engine itself. Attempts to cartelize exchange flow by big, established players would be nothing new, and is definitely something the SEC should guard against. I'd be interested to read a defense of "hide and light" orders before taking a definitive view on the subject, but as I mentioned it's one area where I think some simple, straightforward regulations would be effective and possibly justified.

Of course, that may also be because I work on derivatives exchanges which, so far, have been pretty good about resisting the proliferation of some of these order types. As with so many people airing opinions about HFT and regulation of it, I do have a dog in this fight!

@Anonymous

"HFT market makers make money by frontrunning my $10.00 bid and filling a market order at $10.0001. I can't trade sub pennies, you can, why?"

Putting a small bid order slightly in front of your giant one is not "front-running", but I agree with your argument that the same tick schedule ought to be available to all market participants. In general I think the SEC needs to be more flexible on the tick size issue, allowing sub-pennies in certain highly liquid names and possibly implementing wider ticks on those with less volume. (They've already started looking into the latter.)

Eric Falkenstein said...

All these order types are different, and so addressing them one by one is beyond my patience, but for instance I don't see why hide-and-slide is so bothersome. It's available to everyone playing that game. Not grandma, sure, but she should be pretty orthogonal to such tactics. Do we really want our standard that everyone should make the same amount of money in High Frequency Trading--not investing--regardless of smarts or capital?

In any case, these order types all fall under my point that if you buy a vwap, and all the vwap producers use the same order types, then you benefit from them. The only reason you would think you don't is collusion. Collusion is a big problem, but mainly in the bad old days (eg, Cristie and Schulz) when no one complained because there was no competition, no Haim Bodeks who fail and then blame the industry: no one failed because it was a much tighter club back then.

Trading grew under exchanges that innovated by providing such rules, so by revealed preference this is what the market wants. If you don't like it you can trade other markets or asset classes. It's rather narcissistic to think that because you don't like something preferred by most people, no one should be able to do it. Perhaps you don't like alcohol, should we pass a law forbidding it?

If you think adding sand to the gears will help unsophisticated traders relative to those with knowledge and infrastructure, you are being naive. It just moves the choke point back, say from competition on reacting to market prices to getting one's new price cancelled the quickest, or getting the new bid price as close as possible to the agreed-upon closing time.

Aaron Brown said...

I agree with Eric's post, and I'm sorry that there's so much anger from obviously well-informed people who in my opinion are inflating highly complex complaints of at best theoretical validity into major problems and treating technical issues as points of fundamental justice.

A few broader points.

A lot of you seem to dislike complex order types. The alternative to complex orders is complex behavior. Most of these order types can be duplicated by complex conditional combinations of vanilla orders issued algoritmically. The few that can't, because they change priority, would be replaced by attempts to approximate them if they were outlawed. Basically, you can write a computer program to say what you will do in any future situation, or submit an order that does it for you automatically. I don't claim the result is exactly the same, but it's pretty close, and not obviously better or worse for anyone. It's similar to the question of why some derivatives are so complex, they're complex so one transaction can replace an active trading strategy of simpler instruments, with negligible difference in final result.

Letting people bid for advantages like colocation or order information is not inherently less fair than trying to get everyone to have exactly the same information and access. If there's a popular concert or new product you could hold an auction so the people willing to pay the most got it, or you could charge a low fixed price and let the people willing to wait in line the longest get it, or you could have a lottery to give everyone who wants it an equal chance regardless of how much they want it. All of these are fair systems. The real issue is that most people have no idea what to do with the information and access high frequency traders get. If we made exchanges give it to everyone free (which is actually impossible anyway) the high frequency traders would have the same advantage they do now, and lower costs. If we tried to make sure no one got it, it would just change the game because some other information or technical factor would determine profits.

I see what Eric means about Haim Bodek, but it's unfair. Everyone who writes on HFT seems to hold their business model up as simple honesty and ask for all their competitors to be shut down. The two millisecond trader hates the one millisecond trader, the guy who guesses institutional order flow from public orders thinks dark pools are the work of the devil, the institutional trader who wants to move a large block thinks it's scandalous that she can't do it all at the price from before her order. Haim is actually better than average on that score, although not innocent. But he's one of the few insiders who actually published useful information on the subject (he has a highly informative short e-book) and has a position that is intelligible, non-contradictory and practical (and, yes, in his interests). So while the charge has some directional validity, it singles out one of the less egregious offenders.

Eric Falkenstein said...

I concede that Bodek is no Rubin (who should be in the Hall of Fame for shamelessness). He's still a hypocrite.

Aaron Brown said...

In the Valley of the HFT Critics, the self-consistent informed hypocrite who provides useful information is king.

Tel said...

I notice that the introductory argument of many of the postings here is calling the other guy dumb. Well, IMHO it isn't the smart guy who goes around calling himself smart.

Back to the topic...

"Everyone has a speed advantage or disadvantage compared to some other market participants. That is impossible to change and not inherently unfair."

As I've pointed out earlier, it is possible to change this, should the exchange choose to do so. You merely create a schedule of time-ticks such that the queue does not not resolve until a time-tick comes around. When the time-tick triggers, it freezes the queue, resolves everything it can, and then unfreezes and waits until the next time-tick. By randomizing the time-ticks you create white noise which effectively destroys high frequency information. You can choose to destroy more or destroy less, or whatever spectral characteristics you decide is appropriate.

It can be done.

Maybe some would argue that such high frequency information is to valuable to destroy, I personally don't believe there is anything intrinsically useful on the millisecond timescale that can't operate on much slower timescales. The price discovery mechanism is intended to link to the real physical world at some stage and the physical economy operates slowly.

"Letting people bid for advantages like colocation or order information is not inherently less fair than trying to get everyone to have exactly the same information and access."

Well I think that's pretty much a matter of opinion, and market participants are entitled to their own judgement and no doubt they will divide into various camps. Each person wants what's best for himself or herself, and I'm not pretending to think differently on that score. Of course I don't want someone to have an advantage over me, why would I want that?

At least I can see that many of the posts above do agree (sometimes by implication) that the retail traders are paying a small premium -- either a direct fee paid for access to HFT equipment (e.g. to place VWAP orders), or an indirect premium by not using those facilities. We have come to some sort of consensus on that score at least. Now we merely argue about justifications.

"Retail investors remain skittish about the stock market for a whole host of reasons and HFT is probably not a primary concern for the vast majority of them."

I agree. There's a host of reasons right now, but speaking for myself, once I have maybe four or five reasons why doing XYZ is a bad idea, I don't put a lot of effort into figuring out which is the most important reason.

Eric Falkenstein said...

Tel: If you are correct, an exchange could implement that sort of mechanism, and if people thought it generated 'fairer' prices, more trades would go up there. Trade orders you don't like proliferate because they increase trades; most people do like them.

Tex said...

For the sake of argument, let's stipulate the HFT's are robbing the small investor on every trade. How much are they stealing? A penny? Two pennies?

Compare that to the old NYSE specialist system where small investors paid an 1/8th or 1/4 BID/ASK spread on every trade. Specialists that used to live on the spread are exstinct. They were taking anywhere from $1 billion to $3 billion per year out of the market.

Bottom line is that even if HFT's are stealing candy from babies, it is NOT as bad as the old system. Anybody want to go back to NYSE specialists?

Aaron Brown said...

Tel, I did not express myself well when I said auctioning off advantages is not inherently more or less fair than trying to ensure equal access. I meant what you wrote, that fairness in market access is a matter of opinion. Some of the attacks on HFT (not yours) seem to assume making everyone equal is the only fair thing.

I don't think your randomized tick scheme eliminates speed advantages. You still have people entering and canceling orders based on both news events and other orders (I'm not sure how much of the intratick order book you plan to reveal but if it's a lot people will trade on it and post orders to modify it; if it's a little then people who are good at predicting it have an advantage). When the tick comes, the faster guy will always have a more up-to-date set of orders than a slower guy. The faster guy gets fewer advantages during the day, but they're bigger advantages, because it's an entire tick faster.

I might leave the house two minutes before you, but get the same train, and arrive at the same time. But once in a while, I'll get there a lot earlier, because I just made a train that you just missed. Random or scheduled trains have the same effect.

Tel said...

"When the tick comes, the faster guy will always have a more up-to-date set of orders than a slower guy"

For starters you force them to place orders they really are intending to live with, rather than placing an order, then yanking it, then placing it, then yanking it.

Suppose the tick follows a Poisson arrival process (the stateless distribution) so any time you place an order there's a chance it executes, which is strictly proportional to the time the order sits. So the faster guy can look at other orders in queue and maybe adjust his own order but only to bring it towards the ideal price -- which should be the same price everyone gets on that round.

I argue that the exchange should execute the following on every tick:

http://www.asx.com.au/products/calculate-open-close-prices.htm

Note that only a single price is the result of each time-tick, you don't need a "market maker" and the exchange tries its very best to fairly split the bid-ask spread between participants.

Yes, some participants will react faster to genuine news events, but those people have to actually sit and read the news, and think about what it means -- not on a millisecond timescale they don't.

Tel: If you are correct, an exchange could implement that sort of mechanism

Well maybe they will. Actually the ASX already has a similar random opening time each day, all I'm suggesting is a simple extension of what is already implemented. Probably HFT has only become a concern in the past few years. If the HFT algos were subtle and did a little bit here and there, no one would worry, but as they become more bold and take over a higher percentage of the trade volume, they piss more people off.

The exchanges have already made a few counter-moves against the HFT such as imposing limits, and the Germans look like they will legislatively impose minimum hold times. Personally, I think those are the wrong approach but it shows people are willing to try stuff. I think you have to inject white noise somewhere in order to destroy the high frequency information. Minimum hold times don't destroy information (unless you get imposed a random minimum hold time).

By the way, the concept of deliberate and systematic information destruction is not entirely alien to human affairs. Try going to a regular auction and attempt to increment the bid by $0.0001 see how they react. eBay won't stand for it, nor will any other auctioneer.

Tel said...

Micah Hauptman, Financial Campaign Coordinator, Public Citizen, Washington, District of Columbia

http://www.sec.gov/comments/4-652/4652-20.pdf

Not only do they want to tax transactions, but also they want to tax cancellations. How about that?

Perhaps they colour their language a bit, but they do bring up some important points. The "Project Express" fiber to reduce the latency from London to New York from 65 milliseconds to 60 milliseconds for an exclusive clientele cannot have been cheap -- thus basic economics tells you those extra 5 ms are worth a lot to someone.

Tel said...

Niall H. O'Malley Managing Director, Portfolio Manager Blue Point Investment Management, LLC

http:// www.sec.gov/comments/s7-02-10/s70210-383.pdf

That's a bit more level-headed but points out similar issues. I note that link has been going around.

Another guy (below) actually blames the regulators for creating the problem in the first place (possibly true):

http://www.sec.gov/comments/s7-36-11/s73611-76.htm

"When regulators finally run all corporations offshore how are you going to regulate anything??"

I agree that government regulators are typically ham fisted in the way they deal with things. Interesting the justifications you see coming from HFT advocates above, "you need us" because of liquidity, market making, etc, and the other justification, "it's only a small rake" or something along those lines.

Well, government regulators are using the exact same justifications to support their own involvement. "You need us" because with government regulators the market would be anarchy or something, and anyhow the Tobin tax is only wafer thin, so take cut number 990 and shut up.

The other argument is that HFT has reduced transaction prices, but really computer involvement in general has reduced transaction prices, just like it has in all other trading (e.g. online auctions, web banking, sending letters by email, etc). The HFT algorithms can't claim the credit for general technological advancement.

Aaron Brown said...

Tel, the people you cite are against computerized trading, intermediation and profit, not HFT. I think at least the first two of these represent the majority of HFT critics among the minority with any coherent point of view.

I think you support minor rule changes to reduce the advantage of pure gaming.

I don't disagree that minor changes might improve markets a little. Other than the extremists who believe everything done by any market is perfect, or hopelessly naive people who think intermediaries like brokers and exchanges act in the public interest, no one can disagree.

Personally, I don't take seriously any prediction of the effect of minor rule changes. I don't think anyone knows. I support getting rid of all complex regulation, especially NMS, but not because I know what will happen, because I believe in freedom both as a right and the long-term best general policy. A few simple transparent principle-based rules without regulatory discretion could protect basic honesty.

If you think you know of some more complex targeted rules that could make markets function a bit better, I can't disagree. But to get them forced on exchanges you'll need the help of dangerous people and I predict the attempt would end badly (with unlimited historical evidence to support my position).

So while HFT might be more bad than good in theory, and changes to slow and simplify markets might reduce costs, improve stability and liquidity, and be more fair, that's not practically relevant to the current political debate.

Anonymous said...

I am posting as a retail guy looking into what I am hearing from others who do not participate in the markets but are repeating claims on the dangers of HFT.
Personally I do not find any danger to my own trading. Spreads are almost always close- I take positions at the ask and trying initially to gain some breathing room from purchase for intra-day or swing trades. I usually buy 1000 share lots. If I see "front running" I don't play. I am often wrong about market direction but don't see how I am being effected unless the programs are seeing my purchase and running the trade against the retail idiot.. I see some lightning fast spikes from time to time but they occur at predictable places and I usually take them to be stops being run.. I trade against them If I can.
Am I too dense to notice how I am being affected by these HFT techniques?

Thanks