Flash Boys & HFT

[Lusso1]

I don't understand why just being able to place trades quickly is such a big deal?

In the book Flash Boys, do the HFTs get inside info somehow to decide (via algorithm) how to place their trades?

If not, what is their advantage?

[TheMentat]

Haven't read the book, but my understanding from 60 minutes is that the hedgies would see the order flow in New Jersey, and use a proprietary wire to beat the orders to the exchange with a "front running" order of their own.

[jyl]

Not familiar with the book.

The trades are determined by computer, in reaction to data, if everyone gets the data at the same time then he who processes the information and executes the trade, before others have reacted, can get a slightly better price. Might only be cents per share, but do this for hundreds of different stocks and tens of millions of shares every day, using leverage (borrowed money) to magnify the returns.

[masraum]

I don't know about that stuff myself, but the network at work uses NTP (Network Time Protocol) to keep the time on all computers and equipment in sync EXCEPT for the traders gear. The traders use PTP, precision time protocol.

Network Time Protocol - Wikipedia, the free encyclopedia - Precision Time Protocol - Wikipedia, the free encyclopedia

NTP can usually maintain time to within tens of milliseconds over the public Internet, and can achieve better than one millisecond accuracy in local area networks under ideal conditions.

(PTP) ... it achieves clock accuracy in the sub-microsecond range.

So, NTP can get you accurate to better than .001 sec
and PTP can get you accurate to better than .000001 sec

[pavulon]

computers watching for big trades race to the exchanges, buy that stock then sell it to the original order at a slight mark-up making a ton on volumes.

On A 'Rigged' Wall Street, Milliseconds Make All The Difference : NPR

[jwasbury]

Originally Posted by pavulon

computers watching for big trades race to the exchanges, buy that stock then sell it to the original order at a slight mark-up making a ton on volumes.

^the means is different, but the end is similar to what NYSE Specialist firms and Nasdaq Market Maker firms have done for ages.

[Bully]

Jwasbury, HFT is different. Market makers are charged with running a book, that means that they MUST make a market in their stock/s in good markets and bad.
HFT's are front running the order flow pure and simple. Been in the biz 30+ years, if I did it I'd lose my license, but because it is an algorithm it is legal, so far, this is going to heat up quickly

[Willem Fick]

There are three key ways in which HFTs make money:

The first way is to beat the rush. HFTs don't get insider info, they get the same info at the same time. They however run on the knowledge that the trader that can act on a piece of information first can make most money. Because of the speed at which this all is happening human intervention is simply not viable, therefore the tendency is towards extremely complex trading strategies built into high performance trading applications, and even microprocessors.

A second way is based on the fact that liquidity is good for exchanges as they profit from trade related charges. Some exchanges are therefore prepared to "reward" participants that bring liquidity to the market, as this positively influences their own bottom line.

How this plays out is that certain participants (Supplemental Liquidity Providers) are in essence paid to trade in the form of a trade rebate. This means that they do not so much profit by buying at one price and selling at a slightly higher price but rather by buying at one price and selling on at the same. As ridiculous as it may seem they can in theory even profit by selling at a slight loss, as long as the loss is less than the rebate. The amounts of money they are paid is minuscule, but as they are doing this at frequencies that boggle the mind, these small amounts add up quickly and therefore SLP business is booming.

Lastly, as the same stock is traded in multiple venues, a trading firm can use an HFT algorithm in conjunction with an order router to buy stock at one price in one venue, and sell the same stock in another venue at a higher price, as long as I can do it quickly enough before the difference in price closes up between venues.

[jwasbury]

Willem Fick has taken the time to type it out clearly.

HFT don't have inside info...they just have the capability to act faster on the same info that is available to all market participants. They co-locate their servers next to the exchange servers and pay to have hard wired local network connections between them so they have the lowest possible latency.

And, as Willem correctly noted, exchanges today pay a rebate to "price makers" (those who provide liquidity to the exchange) and charge a fee to "price takers" (those who take liquidity from the exchange). The HFTs are typically liquidity providers/price makers, so they get paid rebates from the exchanges.

The maker-taker pricing is not a new thing. It was introduced in 1997 by Island ECN as a way to attract liquidity away from the primary exchanges. The advent of competition in the exchange space has lowered the cost of trading, and along with HFT activity has lead to tightened Bid-offer spreads.

If the argument against HFT is that it's not "real" activity and is detrimental to John Q. public retail investor, I don't see how it's any different ("less real") than specialist firms stepping in to provide liquidity and "price support", or an underwriter stepping in during an IPO to support a price if market demand is otherwise not present.

[masraum]

I just read the book. Great read with lots of fascinating info.

This seems like a decent summary of some of the situation.
http://www.theatlantic.com/business/archive/2014/04/everything-you-need-to-know-about-high-frequency-trading/360411/

Also some very interesting stuff here.
http://www.nanex.net/NxResearch/

It seems like the situation is one in which they do take/make money from the system, but their actions, can provide some benefit. I've read several articles on it and some of the arguments for it seem oversimplified to me. I don't think it's a black and white thing, but several shades of gray.

[trader220]

Round about 8 years ago I was approached by several HFT firms who wanted to give me a consultants fee because I sat on some committees at the exchanges I was a member of. They all wanted the same info. The off site location of the exchange servers. The closer they were the better, when you're fighting over fractions of a second.

As others have pointed out, their strategy was essentially just a front running strategy. They would send and cancel thousands of orders a second trying to fish out the real orders and trade them to either A) capture fractions of a penny per share, or B) trade for a wash and collect the rebate (payment for orderflow) money.

Yes it adds up quickly.

The original guys back in the 1980's were known as SOES bandits. They figured out then how to front run orders on the SOES system. ( Small Order Execution System). The markets were much much wider and slower in those days. Once you had the keys to the kingdom it was easy to front run orders for big profits.

There was a day even before that when front running was totally legal. On many of the Futures Exchanges you could be both a market maker and a floor broker. You could execute trades for your own account and for customer accounts. You could actually front run your own orders. Guys on the COMEX in the 70's and early 80's could literally make a fortune with no risk at all if they had really good institutional clients. Those were the days. Then again thats when the $2 dollar broker existed on the NYSE too. Two bucks to execute, LOL.... My grandfather was an NYSE specialist for 50 years. He retired in the early 1970's and I was still a kid. When I graduated and went to work on the floor he was pretty sick and the only advice he ever gave me was to "get yourself a good pair of shoes". I remember many other stories he would tell long before I became a trader in the 1980's.

[Willem Fick]

Since my original post above I also read the book, but was astonished by how it took bad research and presented it as fact. A very nice read that explains HFT properly and debunks some of the myths created by Flash Boys is "Flash Boys: Not So Fast" by Peter Kovac.