[harryh]

What you think is happening in the world is not what's actually happening.

What's happening:

1) A large block of shares is traded on exchange #1.

2) HFTer notices #1 and very quickly updates the price at which they are willing to buy/sell on exchange #2. No one has any sure knowledge of trades on their way to exchange #2. They can guess. They can infer. But they aren't seeing any actual trades before they hit the exchange.

[001sky]

Please explain how you would do this in an open outcry market.

You're talking about front-running the trade inbetween the "sell" decision and the receipt of the signal at the exchange. In otherwords, your bid/offer is in bad faith. I can advertise a car for sale and then when you walk in remove the price and mark up the trade. You cannot do this in an open outcry market. Because it would be subject to retaliation (of various kinds).

Providing bid/ask indication in bad faith is NOT liquidity. Liquidity is not layering a trade to double the volume.

HFT does not increase liquidity. It increases volume, at best. But so does front-running. The bait and switch issue and the front-running issue not interchangeble, but they are closely linked.

Market micro-structure is difficult to discuss because most people are completely ignorant of the mechanics. Its easy to pass of volume for liquidity and to disquise front-running as "market making" or speculation. It is clearly distinct.

[harryh]

Market makers are in the business of selling liquidity. They're saying "maybe no one else wants to buy/sell this stock but here I am and I always have a bid/ask on the table." You don't have to trade with them. But a lot of people do every day.

Why do you think those people are trading with market makers if they're doing such shady things?

I don't think you really understand the business that market makers are in. They aren't in the business of trying to front run people. They're in the business of making a penny a share on all the trades that go back and forth. But they face a big risk. When they're busy buying/selling for $50.00/$50.01 what happens when a big & smart hedge fund guy comes along who has new information and knows the price should really be $50.10. They try to buy up as many shares as possible at $50.00 and suddenly the market maker is fucked because they're no longer market making in a stable market.

So the market maker has this big challenge where they have to do a good as job as possible detecting the big & smart hedge fund guy and quickly adjusting the price so they don't get squashed. If they can't do a very good job they have to increase spreads so they make more money in the steady state to make up for their bigger losses. If they do a really good job in this detecting they can lower spreads.

It turns out that HFTers are really good market makers. You can see this because they've successfully competed against each other to the point that they can lower spreads all the way to a penny (a 10x reduction of where they used to be!).

When you look at it this way you can see that market makers aren't trying to screw the big traders, they're trying to avoid getting screwed so that they can better serve everyone else by offering lower prices (lower spreads).

The nice side effect of all of this is that price information gets better communicated to the market faster. If you happen to be randomly selling your MSFT when Mr. Big & Smart shows up you're more likely to get the best possible price due to all the information being correctly communicated to the market.

[001sky]

You're missing the point that front running is capital efficient and ~zero risk. There is no discussion of hedging strategy and cost, etc in your reply. It seems plausible that (unobserved) front-running profits could easily subsidize (observed) low-spreads as loss leaders.

[harryh]

What front running? Are you talking about arbitrage between exchanges or something else?

[001sky]

Anything that starts after "buy/sell" electrons are entered into a book [#] and includes a pricing mark change before "fill" hit my account. This set of trades is broader than what you might imaging, but specifically I'm interested in any trade that (1) has information incorporation; and (2) dependent logic based on (1) and (3) requires sequential execution (ie, fronting). It is not enough to say these trades are not observed. They have to be shown to be impossible (or unprofitable). There is reasonable evidence that they are both possible and profitable in Lewis's book.

# This includes both voluntary and involuntary disclosure.

[harryh]

OK, well I haven't read the book (only the NYT excerpt so far) but I am highly skeptical that such trades are happening. I will read the book though, and reserve judgement until then.