Taibbi: NYSE ends transparency to protect Goldman Sachs

[rj1]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

Can someone put this in English for me? I kind of get the whole "split-second trading" thing, but I don't see how that's illegal unless they're going back in time, and it only represents a penny here and there more in profit, which over time adds up, but isn't $100 million per day.

From Barry Ritholtz:

http://www.ritholtz.com/blog/2009/07...r-trading-day/

[Sharky]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

Can someone put this in English for me? I kind of get the whole "split-second trading" thing, but I don't see how that's illegal unless they're going back in time, and it only represents a penny here and there more in profit, which over time adds up, but isn't $100 million per day.

The penny-here-and-there thing is *per share.* The latest NYSE report shows Goldman trading 2.8B shares per week, or 560M shares per trading day:

http://2.bp.blogspot.com/_FM71j6-VkN...NYSE+fixed.PNG

So if they make two cents per share, that's around $10M/day. $100M/day would be 20 cents/share. Not impossible, I guess, but it seems a bit of a stretch.

The way this might be illegal is if they're front-running trades. In other words, they find out that someone else is about to place an order, so the place an order first. On the buy side, the combined orders would move the market higher. Then Goldman sells, and makes a profit at the now-higher price.

[rj1]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

The penny-here-and-there thing is *per share.* The latest NYSE report shows Goldman trading 2.8B shares per week, or 560M shares per trading day:

http://2.bp.blogspot.com/_FM71j6-VkN...NYSE+fixed.PNG

So if they make two cents per share, that's around $10M/day. $100M/day would be 20 cents/share. Not impossible, I guess, but it seems a bit of a stretch.

The way this might be illegal is if they're front-running trades. In other words, they find out that someone else is about to place an order, so the place an order first. On the buy side, the combined orders would move the market higher. Then Goldman sells, and makes a profit at the now-higher price.

Well doesn't everyone know front-running goes on? Don't the floor brokers always see the big corporate orders coming in? I remember seeing people complain about this on CNBC like 6 months ago.

[ThePythonicCow]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

Can someone put this in English for me? I kind of get the whole "split-second trading" thing, but I don't see how that's illegal unless they're going back in time, and it only represents a penny here and there more in profit, which over time adds up, but isn't $100 million per day.

There is a nicely detailed and well written paper explaining these high speed trading techniques at http://www.themistrading.com/article...t_12-17-08.pdf

Roughly speaking, it's computer versus computer.

Consider the following example, adapted roughly from the paper at the above link. A more traditional algorithmic computer (ALGO) program might be asked to buy say 1000 shares of XYZ at no more than $20.10 per share. Such programs might break that into 10 orders of 100 shares each, one at a time. Currently (this very second) XYZ is at say $20.00 per share. The co-located high speed trading (HST) computer is constantly watching the order stream and issuing its own orders.

For example, one thing unusual thing the HST computer can do is issue an order that is out of the money, and if that order is not instantly matched (by some old fashioned algorithmic computer trying to fill a big automated order say) then the HST can cancel the order it just gave. This is called a "ping." So when the HST sees the first ALGO order to buy 100 shares at the current price of $20.00, it will suspect it has found an algorithmic buyer, and probe to see what is the highest price it can sell to ALGO. Walking down failed offers to sell from say $20.20 down, it will get the first successful sell at $20.10, the upper limit at which ALGO will buy. With this knowledge, HST can rush in and buy 100 share lots at any price below $20.10 and instantly resell them to ALGO at $20.10.

There are lots of games one can play like this, when one has the bigger faster computer located closer to (meaning right inside) the stock exchange data center.

Volatility and trading volumes have jumped dramatically in the last year, with likely over half of all successful (actually filled) trades being HST trades, not to mention the much larger volume of failed "pings" (see above.)

It's difficult enough to make money at the Stock Exchange Blackjack Tables when the house is more crooked, more experienced, and better funded than yourself. But when they replace the dealer with a robot that can sneak a look at your hand and count cards perfectly, it is a pure suckers game.

[ax]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

There is a nicely detailed and well written paper explaining these high speed trading techniques at http://www.themistrading.com/article...t_12-17-08.pdf

Roughly speaking, it's computer versus computer.

Consider the following example, adapted roughly from the paper at the above link. A more traditional algorithmic computer (ALGO) program might be asked to buy say 1000 shares of XYZ at no more than $20.10 per share. Such programs might break that into 10 orders of 100 shares each, one at a time. Currently (this very second) XYZ is at say $20.00 per share. The co-located high speed trading (HST) computer is constantly watching the order stream and issuing its own orders.

For example, one thing unusual thing the HST computer can do is issue an order that is out of the money, and if that order is not instantly matched (by some old fashioned algorithmic computer trying to fill a big automated order say) then the HST can cancel the order it just gave. This is called a "ping." So when the HST sees the first ALGO order to buy 100 shares at the current price of $20.00, it will suspect it has found an algorithmic buyer, and probe to see what is the highest price it can sell to ALGO. Walking down failed offers to sell from say $20.20 down, it will get the first successful sell at $20.10, the upper limit at which ALGO will buy. With this knowledge, HST can rush in and buy 100 share lots at any price below $20.10 and instantly resell them to ALGO at $20.10.

There are lots of games one can play like this, when one has the bigger faster computer located closer to (meaning right inside) the stock exchange data center.

Volatility and trading volumes have jumped dramatically in the last year, with likely over half of all successful (actually filled) trades being HST trades, not to mention the much larger volume of failed "pings" (see above.)

It's difficult enough to make money at the Stock Exchange Blackjack Tables when the house is more crooked, more experienced, and better funded than yourself. But when they replace the dealer with a robot that can sneak a look at your hand and count cards perfectly, it is a pure suckers game.

Thank you, Cow, for this analysis. Hasn't anyone felt like a pawn in the GS game while trying to fill orders before? I saw 1800 contracts fill at the strike limit I placed last week, yet not one went to me.

[rj1]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

There is a nicely detailed and well written paper explaining these high speed trading techniques at http://www.themistrading.com/article...t_12-17-08.pdf

Roughly speaking, it's computer versus computer.

Consider the following example, adapted roughly from the paper at the above link. A more traditional algorithmic computer (ALGO) program might be asked to buy say 1000 shares of XYZ at no more than $20.10 per share. Such programs might break that into 10 orders of 100 shares each, one at a time. Currently (this very second) XYZ is at say $20.00 per share. The co-located high speed trading (HST) computer is constantly watching the order stream and issuing its own orders.

For example, one thing unusual thing the HST computer can do is issue an order that is out of the money, and if that order is not instantly matched (by some old fashioned algorithmic computer trying to fill a big automated order say) then the HST can cancel the order it just gave. This is called a "ping." So when the HST sees the first ALGO order to buy 100 shares at the current price of $20.00, it will suspect it has found an algorithmic buyer, and probe to see what is the highest price it can sell to ALGO. Walking down failed offers to sell from say $20.20 down, it will get the first successful sell at $20.10, the upper limit at which ALGO will buy. With this knowledge, HST can rush in and buy 100 share lots at any price below $20.10 and instantly resell them to ALGO at $20.10.

There are lots of games one can play like this, when one has the bigger faster computer located closer to (meaning right inside) the stock exchange data center.

Volatility and trading volumes have jumped dramatically in the last year, with likely over half of all successful (actually filled) trades being HST trades, not to mention the much larger volume of failed "pings" (see above.)

It's difficult enough to make money at the Stock Exchange Blackjack Tables when the house is more crooked, more experienced, and better funded than yourself. But when they replace the dealer with a robot that can sneak a look at your hand and count cards perfectly, it is a pure suckers game.

Thanks.

So pretty much in order for this stuff to work, the NYSE has to be in on it?

[ThePythonicCow]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

Thank you, Cow, for this analysis. Hasn't anyone felt like a pawn in the GS game while trying to fill orders before? I saw 1800 contracts fill at the strike limit I placed last week, yet not one went to me.

You're welcome, ax. And thanks for the confirmation. I don't actually trade in those large volumes myself, so all I knew is what I read in that paper.

[ThePythonicCow]

Re: Taibbi: NYSE ends transparency to protect Goldman Sachs

Advanced Trading has more discussion of the role that High Speed Trading is playing in the market these days, at The Real Story of Trading Software Espionage.

This article provides a few more statistics estimating the size of this trading volume, and concludes with the following, which is consistent with the conclusions that we have reached above on this thread:

There's no doubt that Goldman Sachs, or any other proprietary trading firm, could indeed lose tens of millions of dollars from its proprietary trading if their strategies are stolen—and that is very serious. The competitors that obtain access to these trading secrets could (and would) use it to front run or trade against it, ruining even the most well-planned tactics. This news story contains many very important sub-plots: trading espionage, the necessity for a trading firm to have sophisticated security systems built around its technology, the requirements for risk management, and even the potential for proprietary trading software to be targeted on a wider scale for terrorist activity; but more than anything else it highlights the critical role played by high frequency prop trading in this new market.