Should High-Frequency Trading Be Banned?

http://www.freakonomics.com/2011/03/28/should-high-frequency-trading-be-banned-one-nobel-winner-thinks-so/

by cwan • 15 years ago

20 62 15 years ago

MarkPNeyer• 15 years ago

How about a tax on capital gains that is inversely proportional to the time the financial instrument is held?

The biggest cost HFT imposes on the world is the brain drain - there are so many super intelligent people working in the arms race that is electronic trading. Imagine if all of those smart people were doing something productive instead of helping people buy and sell options 1 microsecond faster.

tom_b• 15 years ago

I believe you are presuming that all those super intelligent people are hacking on high frequency trading algos because of the big $$$ being made for it.

I think hacking on those things would be great fun. Others might as well.

I do like the idea of putting some kind of brake (or field-leveling force?) on the actual act of high speed trading - without any real economic insight of my own, it seem flaky/shady.

baberuth4• 15 years ago

I find the "talented people not fulfilling their potential" argument terrible. That's often cited as one of the costs of HFT, but its really true of a lot of human pursuits.

It just so happens that the prevailing sentiment is that finance, and HFT in particular are particularly immmoral, so its a popular notion; I find it as "bad" that hordes of talented engineers work on optimizing ad revenue for Google.

Humans will always be underemployed from some perspective ("Oh but she could be doing X"). Penalizing them for not operating in a way that is consistent with your (or anyone's) world view is far more sinister.

pg4• 15 years ago

The fact that smart people's time is also wasted on other types of work doesn't mean there is no such thing as wasting time.

Do you really believe it's meaningless to talk about time being misspent? Would you call it sinister for example to criticize someone for spending their time murdering people at random?

yummyfajitas5• 15 years ago

There is a significant difference between optimizing ads and HFT.

By optimizing ad revenues, you are increasing the number of people who find a product which they find valuable. This is useful - some people are better off than they would be otherwise.

A significant portion of HFT is simply about winning a race. The first to trade gets the benefits. Doing a better job of HFT merely redistributes the benefits of trade rather than creating them.

(Of course, there are other more useful aspects of HFT. Providing liquidity where none existed before is useful, and providing a better bid when you lost the race (i.e., if you lost the race to bid $10.01 first, you can win the race by bidding $10.02) is also useful. But I don't think the criticism of wasted effort is off the mark.)

euroclydon• 15 years ago

It makes a lot more sense if you think about it not as talent being wasted, but as some activities being unwisely lucrative. I don't know much about HFT, but if it's not part of the mainstream purpose of the exchange, then it should be throttled.

andrewf• 15 years ago

Imagine if all of those smart people were doing something productive instead of helping people buy and sell options 1 microsecond faster.

From each according to his ability?

dclaysmith• 15 years ago

Or a $.001 tax per transaction.

bartwe• 15 years ago

Depends on how you define 'High-Frequency Trading'

mmavnn• 15 years ago

How about turned based trading? Once an hour, on the hour, results published at half past...

pasbesoin• 15 years ago

Something I've favored considering. Put the exchange on a stepped clock. Randomize order processing within each step.

I don't care if a lot of people lose their financial jobs. Through their own actions as much as resulting events, they've demonstrated themselves to be harmful parasites.

Modern society depends upon a degree of stability and predictability. Current market evolution has been heading in the opposite direction. Time to reign it in to its role as tool, not goal.

grandalf10• 15 years ago

Absolutely, and all traders should be required to use rotary dial phones to even the playing field further. While we're at it, speaking into the phone at faster than 200 words per minute should also be banned.

dasil003• 15 years ago

This is some kind of reverse slippery slope argument that just doesn't hold up to common sense.

grandalf10• 15 years ago

Care to explain or should I just take your word for it?

wdewind• 15 years ago

I see your point, but it would be a pretty easy line to draw that traders need to hold stocks for more than a few seconds/minutes without disrupting much but the HFT aspect of the market right?

edit: or even that the trades must be ordered by a human. There are really easy dichotomies to draw here.

yummyfajitas5• 15 years ago

This would probably increase transaction costs for the rest of the market.

If I can liquidate a position whenever I feel like it, I'm happy to provide liquidity even with a $0.01 spread. If you impose restrictions like "you must hold this stock at least 60 seconds even if it tanks", you have considerably increased my risk. Now I'll hold out for a $0.05-0.10 spread before I bother to offer liquidity.

This raises the cost for others to trade from $0.01/share to $0.05-0.10/share.

grandalf10• 15 years ago

What good would that do?

Suppose we're looking at a world with one stock.

Suppose I have 100 hypotheses about that stock, each of which is relevant to some time frame. Some are hypotheses about the next few seconds, some are hypotheses about the next few minutes, hours, days, etc.

What is the harm of trying to execute trades to profit from those hypotheses? If there is no counterparty, none of the trades will go through regardless of their timeframe.

If there is a counterparty who decides (for example) to participate in several transactions with me over the span of several minutes (or seconds), we should assume that counterparty is self-interested and is making trades that satisfy its goals.

If on the other hand the counterparty is a firm whose algorithms and trading infrastructure was build in the 1990s and is somehow tripped up by my strategy, how is that any different from a lay investor being tripped up by a misinterpretation of (say) a low end of day closing price?

All market participants are either speculative or based on fundamentals. If you're the latter, you don't care much about the ultra short-term. If you're the former, then you have some strategy for speculation, which may take as input the behavior of other market participants.

In my view, since there is no reason why a rational strategy creator would prefer long term strategies to short term ones (expected risk adjusted returns being equal), there is no reason why HFT should be viewed with disdain. Much as computers replaced rotary phones on the trading floor, HFT is the next iteration of technological improvement in trading.

HFT happens to be a specialty area at present, but there's no reason that any adept speculator would avoid high frequency trades if she had sufficient algorithmic and tech sophistication to make them.

tzs• 15 years ago

I'm leery of HFT, but not out of concern of a level playing field. Consider a market for some commodity. Start with the simplest market--people who produce the commodity sell directly to those who use it. Let's also assume external factors that affect the supply and demand for the commodity are stable over a long period of time.

After a number of transactions have been done, and the information about these has spread among the buyers and sellers, the market should converge on some price for that commodity, which will then be stable.

When some real world event happens that affects supply (a late snow wipes out newly planted crops, for instance) or demand (Justin Bieber turns vegetarian causing millions of teen girls to give up meat), the market will adjust to a new price.

A key thing to note here is that the price depends on real world events and the knowledge the actual produces and consumers of the good have of those events.

Now introduce intermediaries who trade in the commodity, but do not product it or consume it. They speed up the rate of transactions, and help the flow of price information, and so arguable help the market settle on the correct and stable price--basically the price you'd have in the direct market adjusted to take into account the increased transaction cost of having an intermediary.

Since the correct price economically ultimately depends on the factors that affect supply and demand (the weather, Justin Bieber, and so on), my concern is that if trading by intermediaries becomes too fast you can get to a point where the price is not being driven by the real world factors, but rather by traders reacting to other traders.

In other words, could HFT make the feedback loops that allows a market to set prices too strong? Feedback as a control mechanism in general is a good and useful thing, but too much feedback and it can become self sustaining, and the output of the system depends less on external inputs than it should (or even not at all).

Until there's solid analysis showing that HFT can't lead to the bad kind of feedback, I'd rather see it approached cautiously.

grandalf10• 15 years ago

I think your argument is essentially an argument against speculation. All of the fears you cite apply equally to any form of speculation.

The problem is, many things count as speculation -- a farmer who produces a commodity buying insurance against a poor crop is ultimately a futures contract.

The way I think of HFT is that it's like increasing the fidelity of a recording from 11 kilohertz to 44.1 kilohertz. There are a lot more intermediate data points, but the underlying rationality (or irrationality) of pricing trends will be the same.

Before HFT, it was more likely that a bid/ask spread would be higher ... reflecting the risk aversion of the market making firm... With HFT, that spread decreases b/c the liquidity is offered as part of a small timeframe strategy which may or may not pay off.

If you weren't concerned with that short timeframe, then it makes no difference to you whether your order was filled by one party (a traditional market maker) or 100 parties (each a different HFT algorithm).

I think to assuage your fear I'd probably just say that it's very difficult to move large numbers of shares at a price that's all that different from the previous price... So any HFT algorithm will necessarily be dealing either with largely uncontroversial prices or very small share volumes.

Considering that, I think the overall risk of rogue algorithms leading to a harmful feedback loop is not significantly increased by HFT, and HFT was mostly a scapegoat in the recent crash.

intended3• 15 years ago

This is already the case - see your garden variety bubble. Human beings create similar loops with speculation and lack of fact checking.

Fact checking is harder still, if you are stuck in herd behavior. It takes quite a bit to take the risk of standing against the crowd, especially when they seem to be making hand over fist by following the new mantra.

cynicalkane5• 15 years ago

A while ago someone made the interesting point that the costs of HFT are not being accounted for, from a free-market perspective. Exchanges typically charge for completed trades, not for orders, so traders can flood exchanges with very large amounts of orders. In practice, traders that flood exchanges will face penalties--but that only slows the arms race.

The natural economic question is: if HFT traders are making more orders than is efficient, who loses? The answer almost certainly has to be the exchange, and those traders who are losing the time race to HFT's order-spamming advantage. But when Joe the Consumer places an order on the exchange, he always get the best price--it's hard to imagine a way Joe, or even Joe's Mutual Fund, can anything but benefit from HFT.

grandalf10• 15 years ago

Exactly. HFT disempowers market makers b/c all those orders create extra liquidity with worse spread for market makers.

The anti-HFT sentiment is similar to the opposition to electronic exchanges and demands that all pricing be centrally routed -- the old school exchanges realize they have a sweet deal and so they try to fight off other sources of trades/liquidity/pricing.

HockeyPlayer3• 15 years ago

You are confused. Most HFT shops are market makers.

baltcode4• 15 years ago

So they should have a very small fee for orders as well, proportional to its actual cost, but not high enough to dissuade people experimenting with pricing as long as they don't cost other people anything.

yummyfajitas5• 15 years ago

Typically there is a (very low) order fee - off the top of my head, I'm not sure if it's levied by the broker or matching engine.

There is also a limit on the number of orders one can place. Your broker will typically demand fill rates of at least 0.5-1% and will impose limits on the number of orders you can place in a rolling window.

intended3• 15 years ago

I know a way people suffer because of HFT - the flash crash(es). Should the market fall, the HFT leave the scene. Joe Consumer suddenly finds that he has no one on the other end of the trading counter. The market is essentially an illusion. http://en.wikipedia.org/wiki/2010_Flash_Crash The discussion page is actually quite interesting, and makes for an educational thread in itself.

Short summary - HFT left the table after creating a mess. A large sale in e-minis started a trading pattern between where algorithms kept trading back and forth between each other, causing the prices to drop. After a point there was No liquidity and stub trades, which were never supposed to see the light of day, got executed. This resulted in some stocks going for a dollar, while others for many times market value. Sometime after this markets trading was brought to a halt and the regulators stepped in. A large portion of the trades made were nullified. Effectively the markets were reset, order was restored manually, before it was brought back online. At this point the market returned to equilibrium and started recovering.

From the Wikipedia article, the SEC report on HFT is below:

Impact of High Frequency traders: Regulators found they exacerbated price declines. As noted above, regulators found that high frequency traders exacerbated price declines as they sold aggressively to eliminate their long positions and as they withdrew from the markets in the face of uncertainty.[8][9][10][11] Other theories postulate that the actions of high frequency traders (HFTs) was the underlying cause of the flash crash. One hypothesis, based on the analysis of bid-offer data by Nanex, Llc., is that HFTs send non-executable orders (orders that are outside the bid-offer spread) to exchanges in batches. Though the purpose of these orders is unknown, some experts speculate that their purpose is to increase noise, clog exchanges, and outwit competitors.[26] However, other experts believe that deliberate market manipulation is unlikely because there is no practical way in which the HFTs can profit from these orders, and it is more likely that these orders are designed to test latency times and to detect early price trends.[27] Whatever the reasons behind the possible existence of these orders, this theory postulates that they exacerbated the crash by overloading the exchanges on May 6.[26][27] On September 3, 2010 the regulators probing the crash concluded: "that quote-stuffing – placing and then almost immediately cancelling large numbers of rapid-fire orders to buy or sell stocks – was not a “major factor” in the turmoil." [28] Some have put forth the theory High-frequency trading actually have been a major factor in minimizing and reversing the flash crash.[29]

Do note that the person defending the HFT is a HFT trader himself - Jim Simons of quantitative hedge fund Renaissance Technologies. His contention is that HFT saved the day by brining liquidity back. I think this claim is simply wrong and his motivations dubious at best. If no HFT funds were there, and this had to be manually checked, the flash crash would not have happened.

They have since added circuit breakers to the exchange to prevent re-occurrences.

wglb2• 15 years ago

You are malinformed.

Only some of HFT left the table. The ones that stayed at the table helped it come back from the crash. And came out ahead.

dalore• 15 years ago

Why can the banks have high-frequency trading, yet still not have instant transfers between accounts?

secretasiandan• 15 years ago

One creates directly attributable profits with little, if any, ongoing exposure. The other creates little, if any directly attributable profits.

Furthermore, if the transfer is from one bank to another, the sender wants to hold onto the funds for as long as possible to earn more interest.

Lastly, most retail banks probably still run ancient code that is not as conducive to innovation.

CJefferson• 15 years ago

My main annoyance with HFT is that it clearly is a "game". A few times algorithms have misbehaved badly, whole blocks of trades have been rolled back.

baltcode4• 15 years ago

Well, if the algos can make money they should also be allowed to lose money. How else would you have efficiency?

baltcode4• 15 years ago

I don't get it. If the flash crash brought down the market drastically for 20 minutes, only to pull it back up, this only points to an opportunity for a more intelligent algorithm to be allowed to correct that and make money in the process. I can understand this as being an example of the state of the art being pretty dumb/gullible to multiplying its own noise, but I don't understand why this would be bad. Didn't the algos selling during the low and buying stocks later lose money?

yummyfajitas5• 15 years ago

Didn't the algos selling during the low and buying stocks later lose money?

Not necessarily, many trades were broken (mostly trades near the low).

If you sold low and bought halfway to the high, you probably made money since your sale was broken (but not the buy).

baltcode4• 15 years ago

So the problem was that the trades were broken, not HFT per se.

Consultuing• 15 years ago

Let me begin by saying that I likely do not understand even half of what HFT is.

For me, they key part that I fail to understand is exactly what value HFT adds to non speculative market participants. The only explanation I've heard so far is that HFT adds "liquidity" to the market, but I don't get how. Let V be the value HFT adds to non speculative players.

If V is above zero, then allow HFT to continue. If V is zero or negative, my proposal would be not to ban HFT, but to restrict it to transactions made between HFT players. Then let them play in their casino all they want until one player is left. If this suggestion cannot be implemented because HFT needs the rest of the market to make profits, then I highly suspect that V is less or below zero and that HFT is really one of the Casino owners.

Should High-Frequency Trading Be Banned?

Related Stories