Hacker News Reader

Comment by patio11 | Hacker News Reader

Healthy markets are markets where everyone has access to the same information.

If I want to sell 100,000 shares of Yelp (which is a multi-million dollar transaction), the fact that several million dollars of Yelp is about to hit the market is a really, really important bit of information. I have a reason to hide that from the rest of the market at least until I can complete my sale. HFTs profit by ascertaining my intentions faster than I can complete my intended action and then broadcasting my intentions to other market participants via the ticker tape, whereas previously my intentions were proprietary information locked in my head.

An argument against HFT is, necessarily, an argument that I should be allowed to benefit from material non-public information about the near-term state of trading in Yelp.

View on HN

That's private information. It is an important distinction. You're saying the market should have a right to know whether I plan on selling my shares next year.

I'm saying that if anybody can successfully predict your intentions for next year based on your observable actions to date then they have a right to the fruit of those labors, and that you don't have a cognizable interest in "Prevent that person from telling the rest of the world about my intentions for next year", even though that might lead to you making less money than you would but for our friend who reads actions well.

Maybe my TCS background and dabbling in mechanism design has biased me, but it seems to be that truthfulness is a desirable property from the perspective of society: It would be beneficial to have a system that is designed in a way that allows participants to state their intentions and desires truthfully, without any incentive to lie.

Auction theory is all about achieving that for simple, single round auctions. Why not redesign the stock market in a similar way as well? I suppose it could make for a nice PhD thesis (or several).

>I'm saying that if anybody can successfully predict your intentions for next year based on your observable actions to date then they have a right to the fruit of those labors

I don't see why this is so. If I have decided through some analysis that I am going to sell X shares at some date, why shouldn't that information be private? Front-running my behavior is essentially front-running my analysis. It would be unethical to hack my computer and read my analysis for yourself, so how is analyzing my behavior to get at that same information any different?

If you're buying a million dollars of Yelp (as posed in this hypothetical), then it's going to take some time (microseconds, minutes, months, etc) for people to realize that. For as long as no one catches on, the person making this large order has a tremendous advantage because supply-demand is misunderstood by everyone else. The market can't fairly price the stock as a result. The HFTs are saying they'll pay the money to get order info in nano-seconds so they can design algorithms to recognize these large orders and react appropriately. They make money being a middleman but the benefit for the rest of the market would be seeing a price change and also for small orders to be filled at a fairer price to supply-demand.

>For as long as no one catches on, the person making this large order has a tremendous advantage because supply-demand is misunderstood by everyone else

Why is the market entitled to knowledge that I'm currently in the process of making a big order? This knowledge being disseminated clearly doesn't benefit me. Should I not be entitled to the sole benefit of my reasoning in deciding to make a big order? Furthermore, how does this knowledge being disseminated mid-order rather than after the full order has completed benefit the market as a whole?

Not that I think any of this outright determines how the argument should go, but a couple points:

Your knowledge that you're placing an order of size X, and why you're placing an order of size X, is always going to be more precise than what's gleaned from observing your behavior, and thus you'll still have an advantage.

"Furthermore, how does this knowledge being disseminated mid-order rather than after the full order has completed benefit the market as a whole?"

For a large enough order in a sufficiently illiquid market, "after the full order has completed" could be days later. More things being priced in means more accurate prices, and that seems to be of some benefit to "the market as a whole." Though to be sure, the longer time frame isn't terribly relevant to the "predatory traders" across exchanges that Chris had been discussing.

Edited to add: Also, the market doesn't learn anything about you in particular - orders are anonymous. What we learn directly is "a trade happened for X lots at $Y", along with possibly which side was the aggressor (depending on the exchange).

It's private information right up until the millisecond that your trade hits the exchanges. Then it's public information.

Which the HFT firms have earlier access to and greater ability to respond to. Can this legitimately be called fair?

They're paying market rate for that information; you can too if you want to. In the previous version of wall street, it couldn’t be bought—you needed to be part of the old boys club. Are you going to argue that was more “fair”?

These aren't bribes to get the information quickly; there are actual costs involved that must be covered to transmit and process the info with high frequency. The alternative would be saying they'll only publish the information slowly (SIP, or say the morning newspaper), in which case there's an imbalance of knowledge between people placing huge orders and everyone else.

Suppose we had no weather forecasts and meteorologists hoarded the information. And suppose store owners could only adjust prices each morning. Before a hurricane, meteorologists could buy up every last emergency supply within an afternoon from the stores and then sell them the next day after the hurricane at 10 times the original price. Is that fair?

I don't think I understand the point you're making in your first paragraph.

For your meteorology example, if such information didn't have life or death consequences, I might call it fair. But the contexts of the two scenarios are different. The stock market is inherently a bet on future prices, and so using information obtained by yourself through your own efforts cannot be seen as unfair--it is an expected property of the system. Using HFT techniques to profit off of frontrunning your demand is unfair. The "market" for items needed for survival isn't analogous.

They get early access because they pay to be colocated on the exchange. Literally anyone can do this. Can you think of a more fair system than this?

A system where no one can do this or that eliminates the ability to profit off of such a position. Some have suggested a minimum limit on how long one can hold a stock. Seems reasonable to me.

What do you think would happen to spreads if there was a non-trivial minimum limit on how long one could hold a stock?

Obviously they would increase. There seems to be an assumption that reduced spreads at any scale benefits the economy as a whole. On the scale of cents, I don't see how this is necessarily so.

On the other hand, I can imagine there being a trivial limit as far as the wider economy is concerned, but still eliminating the benefit that colocation gives HFT firms. Do you disagree?

If you don't think that increasing spreads by a few cents would hurt the economy than why are you worried about the penny a share the HFT firms are making?

Fairness is a fundamental bedrock of systems like this. Even if the amount of money HFT firms extract does not have a measurable effect on the economy, it harms trust which can have serious consequences.

But people only think it's unfair because people like Michael Lewis have done such a bad job explaining it to them!

They think that HFT makes them materially worse off when it, in fact, does the exact opposite. We should just cave to that misconception?

But if a firm is making a big sell of a stock, as a representative of institutional investors, then it is making regular folks worse off.

Misconceptions should certainly be eliminated, but the fact that HFT firms require colocation does not pass the smell test of fairness. No amount of explanations can eliminate that.

Regular folks aren't buying into firms with proprietary knowledge that deeply care about trading speed. They're buying into index funds or dumb mutual funds. These sorts of funds are the ones that benefit from more accurate and faster updating prices at the expense of the Bill Ackmans & David Einhorns of the world.

I don't know enough to question your point, so I'll take your word for it. But the smell test is still strong here, even if HFT is neutral or beneficial to regular folk. You'll have a hard time convincing people with a complicated argument in the face of this obvious signal for unfairness.

I agree with you about having a hard time. This is exactly what's so frustrating about the Michael Lewis book that kicked off the present round of controversy. See this comment: https://news.ycombinator.com/item?id=7548124

It may be fair in some sense, but it also leads to an arms race that is inherently wasteful.

>If I want to sell 100,000 shares of Yelp (which is a multi-million dollar transaction), the fact that several million dollars of Yelp is about to hit the market is a really, really important bit of information. I have a reason to hide that from the rest of the market at least until I can complete my sale. HFTs profit by ascertaining my intentions faster than I can complete my intended action and then broadcasting my intentions to other market participants via the ticker tape, whereas previously my intentions were proprietary information locked in my head.

Let me see if I understand your example. I want to dump 100k@100$ and yet the current book doesn't have that many buy orders (otherwise the order would just complete). The book has buy 1k@100 and sell 1k@105 So I try to be sneaky and just clean out the book with 1k@100, selling 100 shares each to 10 retail investors and wait for more 100$ retail investors to show up by placing another 1k@100 sell order. The HFT sees this quickly and puts in orders for buy at 90$ and sell at 95$. Now the next retail investor sees this and just buys at 95 instead of 100, saving 5. I need to then sell at 90 instead of 100, losing 10. The HFT firm makes 5.

So the HFT firm is taking money from informed investors and taking some for itself and giving some to people who have no special information about the market like retail investors. If that is the case the only problem with HFTs are second order. That (much like bitcoin) they spend a bunch of resources with busywork that could be at least partially eliminated by more decimal points in pricing trades. And that they may discourage some informed investors from adding information to the market because they can't profit from it, reversing their benefit.

Your saying you should be able to front run the market because otherwise its unfair that I know what decision I'm going to make before you know what decision I'm going to make?

No; he’s saying that you have to choose: if you really believe that the same information should be equally available to everyone, you need to admit HFT that reacts to that information as it becomes available.

It’s perfectly reasonable to argue that shouldn’t be allowed, but then you are (implicitly) arguing that either (a) complete market information should not be publicly available or (b) there should be some mechanism which prevents some individuals from acting on that information as quickly as they would like to.

We're talking about different things, actually responding quickly to information is not the problem, gaining access to it before other people can is, which is what Michael Lewis's book was all about. Nobody is saying Algorithmic Trading is wrong, just front-running in HFT that's the problem.

Look more carefully at Lewis' scenario: there are two kinds of entities that have access to secret information: informed institutional traders and HFTs. Lewis is effectively arguing that there should be only one.

Reply on news.ycombinator.com