Author Michael Lewis, author of many poplar books such as Moneyball, The Big Short and The Blind Side is releasing his latest book 'Flash Boys' which is about High-Speed trading firms and Algos. This book is promising to 'shake up' the Wall Street with its revelations. Was looking up on machine generated quotes that fill up our screens, and came across this informative article. Some excerpts :
http://nymag.com/daily/intelligencer/2014/03/michael-lewis-flash-boys-could-shake-wall-street.html
High Frequency Trading (HFT) is the use of computer algorithms to rapidly trade stocks. Highly sophisticated proprietary strategies are programmed to move in and out of trades in timeframes as little as fractions of a second. It is a business dominated by a few giants as it is a sandbox that costs many millions to play in.
The computers have completely replaced the human specialists, or designated market makers as they are referred to now. The vast majority of the bid/offers we see on quote systems come from High Frequency Trading systems and not from traders.
For some perspective, in 1999 at the height of the tech craze, there were about 1,000 quotes per second crossing the tape. Fast forward to 2013 and that number has risen exponentially to 2,000,000 per second. And yet there are fewer market participants today and actually less trading. All this noise comes from the High Frequency guys trying to game each other or fool traders. Today, 90 to 95 percent of all quotes emanate from High Frequency machines…… This doesn’t imply share volumes just quotes traveling on the tape.
HFT’s are the new market makers without the traditional affirmative obligation of designated market makers to keep markets orderly. When uncertainty enters the picture, they cancel their orders and liquidity disappears. Without traditional market makers to step in and be the buyer of last resort, prices can fall quickly as we saw in the flash crash in May 2010. HFT’s big advantage is co-location or speed which helps keep their bids and offers at the front of the order queue. HFT’s goal on the big liquid stocks is simply to make the penny spread between the bid and the offer, and to make exchange rebates. (Explained below) When HFT’s buy stock, they make sure that there are other real bid’s to buy behind (waiting to buy) them. If they can’t sell on the offer, they can push a button and sell to one of the other bidders at the same price they bought in at and scratch the trade. So the High Frequency Traders of today are just like the old line human market makers except the speed at which they operate and the information advantages they have means they are only going to play when very high probability setups exist. HTF always want to make sure there are plenty of real bids or offers in line behind them to “lean” on so if markets start to gyrate they can exit the trade flat.
On a day to day basis the HFT’s are certainly there on the big liquid names to buy and sell so long as you are willing to pay their penny toll, which is not much different than when beating hearts were in charge. Things get dicey when a market dislocation occurs and then bids dry up. With no affirmative obligation to be buyers of last resort, if some big macro news event causes markets to shudder, then the HFT’s simply pack their bags and there are no underlying bids in the markets.
The flash crash of May 6, 2010 is the best example of what can happen. After the “fat finger” trade by mutual fund group Waddell & Reed and markets went into a dive, HFT’s went hiding under rocks and the market for a short period was without any bids and big price dislocations occurred as evidenced by the nearly 10 percent free-fall. I think it fair to say that without the “affirmative obligations” of the human system, HFT’s run like chickens and during times of duress greatly exacerbate market declines. HFT’s were a big reason the flash crash became the flash crash.
Despite the enormous edge High Frequency firms enjoy they are starting to show significantly lower profitability. Exchange volumes from 2011 to 2012 are down from 7.8 billion shares per day to 6.5 billion or 17 percent. Another contributor is that, as has been stressed, High Frequency Trading is all about co-location or speed. These giant dominant HFT firms, like GETCO, Citadel, and Virtu Financial are all being forced to constantly upgrade their systems in the arms race to be fastest, to collate information closer and closer to the speed of light. GETCO reported in a 2012 SEC filing it spent $37 million upgrading or “building new trading strategies”. Lower volumes hurt the HFT’s because as fewer orders enter the markets there are just that many less opportunities to “scalp” the bid/offer spread. Another issue which gets overlooked in the media is that there are fewer traders exposing themselves to being gamed by the HFT’s.
Traders who have limit orders floating out there in the marketplace get “picked off” or gamed by the HFT’s. It is called “adverse selection” risk and here is how it works. My example focuses on the S&P E–mini contracts and the release of the Friday morning monthly unemployment reports. When the number came out at 8:30 EST there were hundreds of bid limit orders to buy the S&P E-mini’s representing thousands of contracts. The actual employment number was very bad, meaning claims were higher than anticipated, and the market gapped down over 7 points instantly. But even faster were the High Frequency algorithms which hit hundreds of bids (they sold stock to the bids) before the clock had moved one second……it was still 8:30 sharp not even close to 8:30.01. Traders with limit orders could not withdraw their orders before they were filled by high frequency computers in a matter of micro seconds. In the blink of an eye, the HFT’s bought back shares that they had sold at much lower prices reaping millions in profits. This is information arbitrage where High Frequency Traders take advantage of just released data and profit from it. They react far quicker than any human possibly can. Given this advantage, traders have much less incentive today to place limit orders. The adverse selection risk is simply too high.
Another example... Say Bank of America (BAC) is trading $13.93 bid and $13.94 offer. The retail investor has very little chance of being able to buy on the bid at $13.93. That person is typically at the back of the queue, behind all the HFT’s. The only time they get filled is when the quote rolls over them (the bid becomes the offer). In most cases, for the retail investor it is better to buy BAC at the $13.94 offer, get filled and not risk the market moving and not get filled at all……just pay what amounts to a one penny “toll” to get filled. In the old days of human NYSE market makers, or specialists as they were known, the retail investor had a much better chance of buying on the bid.
Where the HFT’s become really pernicious is on smaller more thinly traded securities when the bid/offer spread is wide, say 30 or even 50 cents. Say the bid/offer is $25.50/$25.80. The High Frequency computer is programmed to “step in front of” the real buyer willing to pay $25.50 and bid $25.51 or a penny more than the bona-fide buyer at $25.50. This “penny hopping” pushes the real buyer back in the queue so he will not get filled. If the real buyer cancels their bid, (the computers recognize this) then the HFT will withdraw its bid to buy as well. If the HFT gets filled at this price by paying up a penny more they will hold the shares and make an offer to sell at somewhere below the $25.80 offer price so they can once again be first in the selling queue. Now that the HFT owns the shares, machines will watch to make sure that the real buyer is still there so that if they can’t sell the shares within their timeframe they always can instantaneously sell to the real buyer at only a loss of a cent. They have a bid to “lean” on…….with potential loss of only a cent and a possible profit of nearly 30 cents.
It’s important to note that the great majority for High Frequency action occurs in only the top three dozen or so most liquid stocks. These are the “thickest”, meaning computers have the most bids and offers to “lean” on if markets collapse. The computers make money almost every time trading these stocks. Depending on who you believe 50 to 70% of all trading volume comes from the HFT’s yet 90 or 95 percent of all quotes come from them. HFT’s are constantly putting out bogus or fake quotes. This “quote stuffing” as it is referred forces other machines to slow down to assimilate the data thus maybe giving the “stuffer” a brief advantage. It is also used to flush out and see which bids are real. Also, an uninitiated trader may be fooled by thinking demand for a stock is there, when it is just noise.
The omniscience of the high frequency computers can see your every move and the land mines are everywhere. The structure of the markets has permanently changed. A human hand that can react in seconds is outclassed by machines operating in milliseconds and even microseconds. It is neither all good nor all bad. It is just the way it is.
http://nymag.com/daily/intelligencer/2014/03/michael-lewis-flash-boys-could-shake-wall-street.html
High Frequency Trading (HFT) is the use of computer algorithms to rapidly trade stocks. Highly sophisticated proprietary strategies are programmed to move in and out of trades in timeframes as little as fractions of a second. It is a business dominated by a few giants as it is a sandbox that costs many millions to play in.
The computers have completely replaced the human specialists, or designated market makers as they are referred to now. The vast majority of the bid/offers we see on quote systems come from High Frequency Trading systems and not from traders.
For some perspective, in 1999 at the height of the tech craze, there were about 1,000 quotes per second crossing the tape. Fast forward to 2013 and that number has risen exponentially to 2,000,000 per second. And yet there are fewer market participants today and actually less trading. All this noise comes from the High Frequency guys trying to game each other or fool traders. Today, 90 to 95 percent of all quotes emanate from High Frequency machines…… This doesn’t imply share volumes just quotes traveling on the tape.
HFT’s are the new market makers without the traditional affirmative obligation of designated market makers to keep markets orderly. When uncertainty enters the picture, they cancel their orders and liquidity disappears. Without traditional market makers to step in and be the buyer of last resort, prices can fall quickly as we saw in the flash crash in May 2010. HFT’s big advantage is co-location or speed which helps keep their bids and offers at the front of the order queue. HFT’s goal on the big liquid stocks is simply to make the penny spread between the bid and the offer, and to make exchange rebates. (Explained below) When HFT’s buy stock, they make sure that there are other real bid’s to buy behind (waiting to buy) them. If they can’t sell on the offer, they can push a button and sell to one of the other bidders at the same price they bought in at and scratch the trade. So the High Frequency Traders of today are just like the old line human market makers except the speed at which they operate and the information advantages they have means they are only going to play when very high probability setups exist. HTF always want to make sure there are plenty of real bids or offers in line behind them to “lean” on so if markets start to gyrate they can exit the trade flat.
On a day to day basis the HFT’s are certainly there on the big liquid names to buy and sell so long as you are willing to pay their penny toll, which is not much different than when beating hearts were in charge. Things get dicey when a market dislocation occurs and then bids dry up. With no affirmative obligation to be buyers of last resort, if some big macro news event causes markets to shudder, then the HFT’s simply pack their bags and there are no underlying bids in the markets.
The flash crash of May 6, 2010 is the best example of what can happen. After the “fat finger” trade by mutual fund group Waddell & Reed and markets went into a dive, HFT’s went hiding under rocks and the market for a short period was without any bids and big price dislocations occurred as evidenced by the nearly 10 percent free-fall. I think it fair to say that without the “affirmative obligations” of the human system, HFT’s run like chickens and during times of duress greatly exacerbate market declines. HFT’s were a big reason the flash crash became the flash crash.
Despite the enormous edge High Frequency firms enjoy they are starting to show significantly lower profitability. Exchange volumes from 2011 to 2012 are down from 7.8 billion shares per day to 6.5 billion or 17 percent. Another contributor is that, as has been stressed, High Frequency Trading is all about co-location or speed. These giant dominant HFT firms, like GETCO, Citadel, and Virtu Financial are all being forced to constantly upgrade their systems in the arms race to be fastest, to collate information closer and closer to the speed of light. GETCO reported in a 2012 SEC filing it spent $37 million upgrading or “building new trading strategies”. Lower volumes hurt the HFT’s because as fewer orders enter the markets there are just that many less opportunities to “scalp” the bid/offer spread. Another issue which gets overlooked in the media is that there are fewer traders exposing themselves to being gamed by the HFT’s.
Traders who have limit orders floating out there in the marketplace get “picked off” or gamed by the HFT’s. It is called “adverse selection” risk and here is how it works. My example focuses on the S&P E–mini contracts and the release of the Friday morning monthly unemployment reports. When the number came out at 8:30 EST there were hundreds of bid limit orders to buy the S&P E-mini’s representing thousands of contracts. The actual employment number was very bad, meaning claims were higher than anticipated, and the market gapped down over 7 points instantly. But even faster were the High Frequency algorithms which hit hundreds of bids (they sold stock to the bids) before the clock had moved one second……it was still 8:30 sharp not even close to 8:30.01. Traders with limit orders could not withdraw their orders before they were filled by high frequency computers in a matter of micro seconds. In the blink of an eye, the HFT’s bought back shares that they had sold at much lower prices reaping millions in profits. This is information arbitrage where High Frequency Traders take advantage of just released data and profit from it. They react far quicker than any human possibly can. Given this advantage, traders have much less incentive today to place limit orders. The adverse selection risk is simply too high.
Another example... Say Bank of America (BAC) is trading $13.93 bid and $13.94 offer. The retail investor has very little chance of being able to buy on the bid at $13.93. That person is typically at the back of the queue, behind all the HFT’s. The only time they get filled is when the quote rolls over them (the bid becomes the offer). In most cases, for the retail investor it is better to buy BAC at the $13.94 offer, get filled and not risk the market moving and not get filled at all……just pay what amounts to a one penny “toll” to get filled. In the old days of human NYSE market makers, or specialists as they were known, the retail investor had a much better chance of buying on the bid.
Where the HFT’s become really pernicious is on smaller more thinly traded securities when the bid/offer spread is wide, say 30 or even 50 cents. Say the bid/offer is $25.50/$25.80. The High Frequency computer is programmed to “step in front of” the real buyer willing to pay $25.50 and bid $25.51 or a penny more than the bona-fide buyer at $25.50. This “penny hopping” pushes the real buyer back in the queue so he will not get filled. If the real buyer cancels their bid, (the computers recognize this) then the HFT will withdraw its bid to buy as well. If the HFT gets filled at this price by paying up a penny more they will hold the shares and make an offer to sell at somewhere below the $25.80 offer price so they can once again be first in the selling queue. Now that the HFT owns the shares, machines will watch to make sure that the real buyer is still there so that if they can’t sell the shares within their timeframe they always can instantaneously sell to the real buyer at only a loss of a cent. They have a bid to “lean” on…….with potential loss of only a cent and a possible profit of nearly 30 cents.
It’s important to note that the great majority for High Frequency action occurs in only the top three dozen or so most liquid stocks. These are the “thickest”, meaning computers have the most bids and offers to “lean” on if markets collapse. The computers make money almost every time trading these stocks. Depending on who you believe 50 to 70% of all trading volume comes from the HFT’s yet 90 or 95 percent of all quotes come from them. HFT’s are constantly putting out bogus or fake quotes. This “quote stuffing” as it is referred forces other machines to slow down to assimilate the data thus maybe giving the “stuffer” a brief advantage. It is also used to flush out and see which bids are real. Also, an uninitiated trader may be fooled by thinking demand for a stock is there, when it is just noise.
The omniscience of the high frequency computers can see your every move and the land mines are everywhere. The structure of the markets has permanently changed. A human hand that can react in seconds is outclassed by machines operating in milliseconds and even microseconds. It is neither all good nor all bad. It is just the way it is.
.His books are based on non-fiction,hard-hitting truth.
