- 3 weeks ago
Category
🤖
TechTranscript
00:03The 5th of February 2018. At 3 p.m. local time, the Dow Jones plummets as though the
00:09stock exchange has been hit by lightning. It loses more than 1,500 points in a matter of
00:15minutes. Never before in the history of the index have American stocks lost so much in
00:20a single day. Approximately $2,000 billion vanish in a puff of smoke. And New York is
00:27not alone. The collapse spreads like wildfire. To Frankfurt, Sydney, Tokyo. Panic erupts.
00:34Nobody knows what has caused this disaster. There is no discernible real world economy
00:40reason. There are no new economic predictions. No unemployment statistics. The only explanation
00:45that is offered sounds strange. Apparently, the trading computers were too fast.
01:05How is it possible that a fringe development of automated stock trading, the astronomic acceleration
01:11of computer systems, could almost cause a worldwide financial crisis? In Berlin, Gerhard Schick campaigns
01:21for more transparency in the financial sector.
01:44Essentially, high-frequency trading significantly reduces the time that a stock is held before
01:50being sold. Sometimes, it's just milliseconds or even microseconds. High-frequency trading is
01:57the opposite of long-term investment, which is what stock markets were actually designed
02:02for. Access to stocks also became a lot faster. The quality of the investment was no longer
02:08the deciding motivation for buying or selling. All that mattered was who could react the fastest.
02:13High-frequency trading is performed exclusively by computers, not people. These computer systems are
02:21completely autonomous. Essentially, they are warring bots. High-frequency traders place their
02:27computers as close as possible to stock exchange computer systems, a process known as co-location.
02:34Proximity and high-speed connections are a deciding factor. For example, a pension fund wants to buy
02:40600,000 shares at a price of $9.50. Their order goes out to various stock exchanges. When the order
02:47arrives at the first stock exchange, a high-frequency trading system detects it and immediately places
02:53its own orders for the same shares with all the other exchanges, faster than the pension fund can.
03:00Thanks to its high-speed systems, the high-frequency trader is able to buy up all the shares at a
03:06price of
03:06$9.50, then sell them to the pension fund for $9.51. The high-frequency trader has earned
03:14$6,000 within a few milliseconds and with no attached risk.
03:25High-frequency traders do not have to be active in the markets permanently. Their advantage is most effective
03:32when crises, disasters or political decisions cause strong price fluctuations.
03:44Essentially, high-frequency traders benefit from crises and are constantly waiting for the next one to occur.
03:52In the beginning, the intention was to use greater predictability and faster infrastructure to increase the stability of financial markets.
04:00In fact, the opposite has occurred. Automation has made them more vulnerable.
04:07After all, there is no electronic system that cannot be hacked.
04:19I had a bot, you know, a trading engine. It was running in Frankfurt remotely.
04:27But it was being controlled from London.
04:33What happened in this particular case is that an internet-related outage caused us to lose control over the engine
04:43that was in Frankfurt.
04:46And I remember, for a couple of hours, trying to validate that the engine itself was demobilized.
05:08It turned out my engine was actually live.
05:16I had actually designed a feature that said if you break connection with the mothership, stop trading.
05:26But that feature had a parameter. And it was off.
05:31I had this engine that was basically trading. It could have lost, you know, it could have lost $30 million,
05:38right?
05:38It lost a few hundred thousand dollars. But it was a terrifying experience.
05:46Heim Bodek is a trader and commerce software developer in Stamford, Connecticut.
05:51He played an important role in the development of automated stock trading.
05:56His idea of using computer algorithms to process information as efficiently as possible is one of the foundations of modern
06:05stock trading.
06:10A second prerequisite for success is a high-speed data connection.
06:16The French businessman, Stéphane Teek, knows all about fast data transfer.
06:22His companies, McKay Brothers and Quincy Data, are based in Paris and California
06:28and operate the fastest data connection between New York and Chicago.
06:33The homes of America's most important stock exchanges.
06:41Stéphane Teek and Heim Bodek are players in a global battle for the best computer software and the fastest data
06:48networks.
06:52This contest began a long time ago, in the casinos of Las Vegas.
06:59At the end of the 1960s, a new generation of scientists recognizes that it is possible to use computers to
07:07increase its luck at the tables.
07:10Pioneers like Edward Thorpe and Blair Hull develop algorithms that allow them to count cards when playing blackjack and to
07:18make predictions.
07:19Doin Farmer takes a different approach.
07:22In the 1970s, the physicist makes roulette his field of expertise.
07:29Farmer and his team begin to play.
07:37They soon discover that it is possible to predict the course of the ball if one knows its current position
07:43and speed.
07:44And then we embarked on building a computer that would allow us to do this.
07:51Now this was 1976 when we got started.
07:54So you have to realize at that point there were no personal computers or anything like that.
08:02In Pennsylvania's Allegheny Mountains, Stéphane Teek's company is working on making the fastest data connection between New York and Chicago
08:10even faster.
08:12Nos réseaux sont conceptuellement très simples.
08:15Ils démarrent par une fibre.
08:17Cette fibre va entre le data center et la première tour.
08:22Elle arrive jusqu'à une radio.
08:24Au moment de la radio, l'information est encodée dans des signaux radio qui sont envoyés jusqu'à une autre
08:32tour.
08:33C'est reçu là et c'est répété de manière analogique.
08:38C'est-à-dire qu'il n'y a aucun traitement qui est fait.
08:40Et donc c'est répété jusqu'à la prochaine tour.
08:42Il faut être absolument parfait tout le long de la chaîne pour avoir une chance de garder un signal propre
08:49depuis Chicago jusqu'au New Jersey sur 1200 km sans avoir besoin de faire aucun traitement dessus.
08:58C'est un petit peu comme lorsqu'il y a une course de Formule 1, on doit s'arrêter au
09:03stand
09:04et le temps d'arrêt au stand est souvent ce qui décide de la course.
09:08Donc nous, notre temps d'arrêt au stand, c'est le passage à travers l'équipement radio sur une tour.
09:18An algorithm is a rule for answering questions.
09:23Algorithms provide the foundation for farmers' bets at the casino.
09:34A computer is required to process algorithms quickly.
09:38At the time, however, these are large, expensive and slow.
09:45This begins to change in the mid-1970s.
09:49The 6502 microprocessor had just come out.
09:53The same microprocessor that the guys from Apple used to build the Apple computer.
09:58Doyne Farmer's team designs a computer around the same processor.
10:04The initial version of it was under the armpit.
10:08There was a pack of AA batteries under the other armpit.
10:13Switches in the shoes, wires running up and down, an antenna around the shoulders.
10:19The antenna would transmit the signal to a second person who would then place the bets.
10:26From the moment the croupier spins the ball until the ball, until the bets get closed.
10:31There's typically about 10 seconds that pass, 10 to 15 seconds.
10:37So we had to be able to take the data for the rotor and the ball and get the prediction
10:46four to six seconds ahead of when the bets got closed.
10:50So that the bettor had enough time to reach over and place the bets.
10:55So we had to have the prediction coming pretty fast.
11:00The computer we had ran very slow by modern standards.
11:03But because it was, you know, carefully coded machine code, we were able to get the answer out in a
11:11matter of a fraction of a second.
11:15Doyne Farmer's computer became the stuff of legends.
11:18It wasn't just any computer.
11:21It was the first computer built into clothing.
11:24While Steve Jobs and Steve Wozniak were creating the original desktop computer, Doyne Farmer was building one in a shoe.
11:33Farmer himself becomes an icon for people working in the field of computer-based predictions.
11:40Farmer and his team eventually decide that their winnings at the roulette wheel do not justify the risks they are
11:47taking.
11:52They return to more serious scientific fields.
11:56Farmer finds himself at Los Alamos, where he investigates methods of predicting the weather, among other things.
12:05Then the world changes, or at least the world of money.
12:12Now let's go to the financial desk in New York and to Bill Hurley to find out what's happening on
12:17Wall Street.
12:19Sorry, there's only one word to describe what's happening, and that is panic.
12:22The Dow Jones Industrial Average is off almost 200 points.
12:26It is down by 191.
12:28Had been down to fall right now.
12:30Volume, 345 million shares.
12:32Incredible day on Wall Street.
12:34The Dow Jones Industrial Average right now down 271.53 points.
12:40Putting the Dow below the 2,000 mark.
12:43The Dow right now at 1975.20.
12:48During the 1987 stock market crash, investors attempt to dump their shares in a panic.
12:54Their brokers eventually begin to ignore the frantically ringing telephone.
12:59The authorities decide that more trading should be left to computers, as people simply are not reliable enough.
13:09The New York Stock Exchange remains America's largest, but Chicago becomes the center of new financial technologies.
13:17The Hull Trading Company is one of the pioneers in the field of big data and market predictions.
13:25In the early 1990s, the company is considered a measuring stick by many traders.
13:39The Hull Trading Company's founder Blair Hull had originally put his mathematical skills to the test in Las Vegas,
13:47where he counted cards at the blackjack tables in the 1970s.
13:51He is one of the first to recognize the potential of automated financial trading.
13:56In 1985, he founds his own company in Chicago, the center of U.S. derivative trading.
14:09Heim Bodek joins Hull's team in September of 1997.
14:15Quantitative analysts are also in high demand in Europe, the best emerge from the universities in Paris.
14:24Mathematicians are welcome additions to the market, as are physicists like Stéphane Teague.
14:38At the time, on Wall Street, all banks wanted to hire physicists.
14:45And so I tried to say that I was going to do the same thing in France.
14:50And the BNP, at the time, also wanted to hire physicists.
14:54And so I went to go.
15:01One of the things I learned and discovered,
15:04is that the financial mathematics live with kind of calculations,
15:13which are justified, but which are kind of calculations.
15:16And it's impossible to describe the reality perfectly.
15:20And there is no equation of the Bourse.
15:22There are only approximations of the Bourse.
15:25But it's still a world where BNP Paribas and many other banks who founded Derivet
15:33have built an activity that works, in fact, very well.
15:39So it's a world, it's an artifice, which is both faux, because it doesn't describe the reality,
15:46but that is sufficiently well approached for being able to increase
15:52the activities of several billion dollars per year.
15:57While we were doing the Roulette Project,
15:59our friend Rob Shaw came over and said,
16:02I just found out about this amazing thing of chaos.
16:06So we went up to the lab,
16:08and we looked at the simulation on the analog computer,
16:11and we were just amazed, because you could see,
16:13even though there were deterministic equations,
16:16was making this random looking set of patterns on the screen.
16:20And so we got completely fascinated by that,
16:24and we created algorithms for finding, identifying patterns in nature
16:32that are described by chaotic dynamics.
16:44Doyne Farmer devotes himself to studying patterns
16:47that allow predictions to be made.
16:50One of his areas of study is meteorology.
16:54Farmer and his team in Los Alamos are able to expand the timeframe
16:58for precise hurricane predictions from two days to five.
17:02This gives emergency services valuable time to evacuate people in danger.
17:12Every time I would talk about this, somebody would say,
17:14well, have you applied this to the stock market?
17:18And so I kind of got tired of saying no.
17:23Meanwhile, I heard about the efficient markets hypothesis,
17:26which says that stock markets are inherently unpredictable.
17:29It was a very popular theory at the time, and I didn't believe that.
17:36So, Norman and I, once again, left.
17:39He left his job at University of Illinois,
17:42and I left my job at Los Alamos,
17:44and we gathered with some other younger scientists
17:47and a business guy, and started a company to use computers
17:51to predict the stock market.
17:52This was in 1991.
17:55And we called ourselves Prediction Company.
18:10So, the Bourse of Paris, the Palais Broignard,
18:13was built in 1829,
18:15and the men, the Corbeille,
18:18was replaced by computers
18:19until the end of the 1980s.
18:22It was the continuous quotations,
18:24the famous CAC,
18:26which is the CAC-40 indicator.
18:29And now, there's nothing more,
18:31in any case, nothing more about the Bourse,
18:33in the Palais Broignard.
18:35Everything went to London,
18:36in the bank of London, in Basildone.
18:40The Paris Stock Exchange is not the only empty shell.
18:44Even Wall Street is little more than a movie set these days.
18:50This Interaction Data Center is located in an anonymous building
18:55on the outskirts of Vienna, Austria.
18:57The company has 13 such co-location centers throughout Europe.
19:08More than 100 internet service providers,
19:11five internet exchanges and important cloud services
19:15have computers stationed here.
19:18As does Vienna's Stock Exchange.
19:20The exact location is kept secret,
19:23and there is no official confirmation
19:25of the presence of Stock Exchange computers.
19:29What is important for the business model,
19:31that Interaction has been used to,
19:32is to place the account center
19:34on one of the key points
19:35of international data routes,
19:38to place this community of interest,
19:42as we call it,
19:43optimal to use it.
19:45That means,
19:45we combine here at our standard
19:47companies together,
19:49to optimize their jobs optimally.
19:53Time is of the essence.
19:56The speed plays an important role.
19:59The latency time that the customers expect
20:01is moving in a millisecond area,
20:03and that is for different business models
20:05from gaming to finance.
20:10For this reason,
20:12the Vienna Stock Exchange's most important server,
20:14over which deals are made,
20:17isn't located in Vienna at all.
20:19It is in Frankfurt.
20:25In the race to buy or sell,
20:27it's important to be as close as possible
20:30to the main server,
20:31of the main Stock Exchange.
20:34Proximity is time,
20:36and time is money.
20:38In the 1990s,
20:40US trading is divided between
20:42the New York Stock Exchange on Wall Street,
20:44with its traditional trading floor,
20:47and its traditional companies,
20:48such as IBM and General Motors,
20:51and the electronic stock exchange,
20:53NASDAQ.
20:57The European markets actually went electronic first,
20:59and many of us who were part of that transition
21:02from floor to electronic markets,
21:03we kind of were groomed in the European markets first,
21:07and then we came back to the States.
21:08As a matter of fact,
21:09Urex is quite interesting to me,
21:12looking back,
21:13because it was the first exchange
21:15where I found, you know,
21:19basically a cheat to exploit.
21:27I got a lot of credit and I got a big promotion
21:31for finding a cheat on the Urex exchange
21:33to get prices faster than we should have.
21:36And, you know, years later,
21:38when I hit the front page of the Wall Street Journal
21:40for my whistleblowing against Direct Edge,
21:45a former colleague from Hull contacted me by email and said,
21:49hey, what you turned on at Direct Edge is very similar
21:52to what we were doing on Urex.
21:54And I had been a whistleblower for a year,
21:57and I had not really acknowledged that I was guilty
22:01of the same type of thing that I had turned in.
22:04You know, I just hadn't equated the two,
22:06and it made me realize how difficult it is
22:09for people to really see what they're doing in our industry.
22:22When we founded Prediction Company,
22:24we didn't know anything about financial markets.
22:28Our method was to look for patterns in the data,
22:33and when we found those patterns,
22:35then we would design a trading system
22:36that would make bets based on the patterns.
22:41What do you think? Let's go.
22:45We learned some lessons from that.
22:48One lesson was that you've got to bet big.
22:53The stock market also, to make a good model,
22:56you don't have to predict very well.
22:58If you're right 55% of the time,
23:01if you make enough bets,
23:03then you can make a pretty good profit.
23:07And unlike roulette, where if you start to win,
23:09they can just throw you out or threaten to break your kneecaps,
23:14the stock market, they don't kick you out for winning.
23:19So we actually took us about five years before we really found something that genuinely worked in a consistent way.
23:26But once we did that, it worked very well.
23:29And so Prediction Company traded subsequently for about 25 years.
23:37So I left in 1999.
23:42My goal had originally been to do it for five years and make some money and then go back to
23:46doing science research.
23:48It took me eight years.
23:51So we ended up selling it to UBS in about 2006 for $100 million.
23:58Speed is not only essential when operating a stock exchange.
24:02Investors and traders also have to react quickly.
24:05Those who are first to recognize price differences between stock exchanges can use this advantage to make a lot of
24:13money.
24:13At the turn of the millennium, the large banks realize that they are falling behind their competition.
24:19In 1999, Goldman Sachs buys the Hull Trading Company for half a billion dollars.
24:26Hull was supposed to become like the backbone of Goldman and it really didn't.
24:32And it really was a very, very different culture at Goldman than at Hull.
24:40And many of the people I related to didn't feel at home.
24:47Dissatisfied with the investment bank's business practices,
24:50a number of former Hull Trading Company employees leave Goldman Sachs.
24:54In 2003, Bodek joins UBS, which is planning to open a new trading floor in Stanford.
25:00He is tasked with creating the infrastructure from the ground up.
25:04I remember the first time I heard about high-frequency traders.
25:09I was in the elevator with a person who is very, very senior in the business now.
25:15And he was complaining about how they would interfere with the trading profits of the desk he was running.
25:25But that they didn't make that much money.
25:28And that was just so interesting, is that this strategy class made a huge amount of volume, but not much
25:36money at all.
25:37And we were like, why would people do that? You know, why would you do all that work and make
25:44such a small profit?
25:49Many were surprised by the rise of high-frequency trading.
25:52The financial journalist Jennifer Neal kept a close eye on developments from the very beginning.
26:05Well, of course, at the time when we were in a financial crisis, all the focus of the media was
26:10on what was happening.
26:11The American immobilist who disappeared, all the banks fell after the other.
26:15It was completely gone through silence by the press generalist.
26:21It was also a very delicate time, because all the financial system was going to be threatened.
26:26But during this time, paradoxically, in 2008, there was a company called Getco, who is a trader out of frequency,
26:33who has gained historically its best benefit this year.
26:37Because these sectors love the volatility on the markets.
26:40And at this time, we were at a peak of volatility, because the people sold and sold out of an
26:47efferent manner,
26:48so they were afraid of what was happening on the markets.
26:52There is a confusion very often between the automatic trading trading and the finance.
26:59These two worlds are close, but they have nothing to do with.
27:02The trading automatic, it's just a program, a computer that facilitates the transactions,
27:12which make them less cheap, more sure, more fast.
27:17So it's something that is essentially good.
27:20When there is a bourse, and the public can go and sell it in a bourse,
27:25it is assured that the difference between the purchase and the purchase of the purchase is small.
27:30It is assured that the transactions are low and that everything is good.
27:36While the explosion of high-frequency trading in the early 2000s continues,
27:42Haim Bodek is developing new trading strategies for his employer UBS.
27:48He calls it size gain.
27:51His algorithm follows a simple logic.
27:54Because large orders are given preferential treatment by the stock markets,
27:59the money bot dynamically increases its order to remain at the forefront and push out competitors.
28:05However, there is a catch.
28:07He has to ensure that the artificially inflated order never actually takes place.
28:13If you put huge orders out there of that size, and they all trade, you will go pretty much instantaneously
28:19belly up.
28:22Okay, I won't go into details, but it happened to other firms.
28:27It's a game of chicken in a way.
28:29We had to write algorithms that said, if you traded 30,000 contracts, you didn't want to trade those 30
28:36,000 contracts.
28:37So try to dump them immediately on another exchange where some other market maker is still quoting 30,000 contracts.
28:48So we created all this algorithm and this kind of war in terms of getting the largest what we call
28:54participation rate.
28:56It turned into this horrible game of chicken and hot potato where you quoted more than you really should in
29:03any responsible manner.
29:05And then if you actually got hit, which happened, you needed to dump it on other people before they realized
29:18that the hot potato was coming.
29:21It was as stupid as you can get. It was just stupid.
29:28So the algorithm that I designed was powerful enough that everybody else had to match it.
29:35And that actually brought down the market system, the market data system of the US options market to the brink
29:42of like collapse, basically.
29:45And what was so cool for me, I was like, I just like literally just love this.
29:50I could see how many months it took for another firm to realize what I was doing and try to
29:57introduce the same algorithm.
29:58And it took three years for seven firms to figure it out.
30:03Heim Bodek's size game turns him into a financial rock star.
30:08However, he knows that he can't play the game forever.
30:12It is time to strike out on his own.
30:17I walked out with the team and I set up trading machines.
30:24It was 2007 and it was designed to be, you know, a Ferrari, but an independent Ferrari.
30:35To me, you know, for me it was really a dream is to be independent.
30:41I didn't realize it was going to turn into an odyssey.
30:46Bodek leaves UBS at a time when the stock markets in the USA and Europe are changing dramatically.
30:52Part of this is due to the high frequency traders.
30:55The large investment funds are angered by the fact that the small, flexible high speed traders are constantly snapping at
31:03their heels.
31:05The banks begin to set up internal exchanges which allow trades to take place in secrecy.
31:15More than 40 of these so-called dark pools are established as alternatives to the 13 regular stock exchanges in
31:24the USA alone.
31:25The movement of investors into the dark pools forces the stock markets to cater to the high frequency traders and
31:32they soon make up the majority of trades.
31:35Some estimates put the number as high as 70%.
31:38Before long, the banks themselves cannot resist the temptation of the high frequency money and open up the dark pools
31:46to high frequency trading.
31:48Speed is all important.
31:52We are now approaching the speed of light.
31:54We are now at about 10 microsecondes above the speed of light.
32:02We are now at about 10 microsecondes above the speed of light.
32:09We have launched right before the financial crisis. We have launched basically into it. We traded through it. We scaled
32:16out.
32:17All of our algorithms and memberships on major exchanges pretty much hit what I call full footprint.
32:24And we went to half percent U.S. market share pretty much overnight.
32:31And things were going very well until May 2009 when things mysteriously stopped working.
32:46We built a system that would basically measure that the trades that we intended to do were in fact profitable
32:56and that were the trades that the machine should be doing.
32:59Over the period of 2009, I started seeing changes.
33:10The stock trading that we did changed quite a bit over that period of time.
33:15And what happened was I, you know, kind of bought into what I think of now as a myth where
33:23people say, oh, that exchange is toxic.
33:27And, you know, without, you know, that's just kind of a blanket term, right?
33:30You know, I'm getting screwed on that exchange. That's what being toxic means.
33:33So what I, what was happening is I would keep redirecting our trading towards the exchanges that were less toxic.
33:45And then there were none.
33:47So the last exchange to go toxic was direct edge.
33:55In May of 2009, I was out of the, I was, you know, away from the office and for a
34:01week and I came back and there was this dramatic change that was so severe that it basically took the
34:08business from modestly profitable to flat.
34:12In order to confront the problem, Heim Bodek immediately buys faster connections and price feeds.
34:19He optimizes his codes, but the problems increase.
34:23He watches powerlessly as the blood drains out of his company.
34:28Facing bankruptcy, Heim Bodek encounters a drunk direct edge salesman at a party.
34:41He just told me and he outlined three major abuses.
34:47He, I got interested and I remember taking on a napkin.
34:51I don't know where we found the pen or something, but I wish I had this napkin still.
34:55I'm, you know, I'm like, what? This, do you mean it works like this?
34:57And he explained to me these very, very sophisticated abuses.
35:02And, um, he said to me that they'd built all this stuff to attract high-frequency traders, but they hadn't
35:11done a good enough job, so now they were going to make new high-frequency traders.
35:16And I was invited in the club to become a new abuser.
35:21It was 20 minutes of work to fix these problems that I had been chasing for half a year, and
35:29this guy could have just told me half a year ago.
35:32And, you know, if he had told me much earlier, I probably wouldn't have crossed my mind how wrong it
35:41was.
35:41But I was so angry.
35:43You know, it was very clear to me that you either used the cheats or you were a victim.
35:48It was designed. There was no middle ground. You could not be innocent.
35:53Once Bodek discovers the reason for his company's failure, he embarks on a last-ditch attempt to save it.
36:00However, he is too late.
36:02Trading Machines enters bankruptcy during the turbulence of 2010 and 2011.
36:10For the first time, Haim Bodek is forced to confront the realities of high-frequency trading.
36:18Do you think the high-frequency trading industry wants to say,
36:22well, we're using undocumented special order types and an obscure exemption and regulation NMS to get to the top of
36:31the queue?
36:32No. They're going to tell you they're faster than you.
36:36And that's the false flag, let's say. I don't know what the right term is. But, you know, the red
36:45herring. Speed is a red herring for a lot of tricks that my industry has developed to get ahead of
36:58other people.
36:59The activities of the high-frequency traders remain hidden from view for a long time.
37:04In 2010, something happens that suddenly lifts the veil.
37:09At 2.32pm on the 6th of May 2010, the US financial planning company Waddell & Reed places a huge
37:18buying order for forward contracts valued at $4.1 billion, using an automated execution algorithm.
37:26It is a huge order and extremely promising for high-frequency traders.
37:31The money bots immediately begin to buy and sell. At lightning speeds and without letting up, share prices immediately begin
37:40to collapse.
37:41Approximately 3.5 million shares are affected and more than 200 companies lose their entire share value in the matter
37:49of minutes. Nobody knows what is happening.
37:5113 minutes later, the Chicago Mercantile Exchange is forced to stop trading. So much liquidity has been sucked out of
38:00the market that it's in danger of total collapse.
38:03After a break in trading of 5 seconds, the market calms down again. The cause of the events of May
38:102010 is never fully resolved.
38:13Experts assume that the collapse began at precisely 2.42pm and 44 seconds, when the money bots had bought up
38:22the maximum permissible number of contracts and simultaneously had to change their position.
38:27A digital stampede which caused the market to implode. The stock exchanges immediately question the veracity of this explanation.
38:36Instead, 5 years later, a lone trader in the UK is convicted of having destabilised the stock market from his
38:45basement.
38:47This scapegoat allows the exchanges, banks and high-frequency traders to continue playing their game.
38:54In 2013, another crash is caused by a fake tweet. Syrian hackers successfully used the Associated Press account to fake
39:04a terrorist attack on Barack Obama.
39:07Just one fabricated report is enough to cause turbulence in the financial markets.
39:12Once again, the severity of the event is soon brushed aside. Part of the reason would appear to be that
39:19nobody really knows why the markets are increasingly acting strangely.
39:24The largest flash crash to date happens on the 5th of February 2018. Its impact is felt around the globe.
39:34You don't know it until today, but there is a doubt. A whistleblower called a whistleblower.
39:43And the banks had manipulated manipulation to the banks. He said that he was manipulated to carry out this flash
39:50crash.
39:55Doyne Farmer's journey from the casinos to Los Alamos and onto Wall Street eventually takes him to Oxford,
40:02where he now studies stability in complex systems like the financial markets.
40:14I think part of what happens in flash crashes, though we haven't gotten to the bottom of flash crashes as
40:19well as we should,
40:21there are algorithms running and they are just reflexively doing something and they go on right on doing it even
40:29though it would be better to stop.
40:32Because that's what they're programmed to do and they don't have time to think about it any more than that.
40:38Running a trading system that literally cares about microseconds, that means you have to write code that's extremely spare.
40:47Your signal hits your machine, your machine can at most run a few lines of code and then it has
40:53to make a trade.
40:54You have to do everything super quick, you can't think about it very much.
40:58When I say you, I mean the machine can't think about it very much.
41:06I do think we have to worry that as markets go faster and faster, we're opening up the possibility for
41:12an instability as a result of the collective interaction of high-frequency traders with other parts of the market.
41:19And we're doing that for no useful purpose.
41:24It would be one thing if we were taking these risks because it gave us something that was valuable.
41:30It doesn't give us anything valuable, it just makes a few people rich.
41:34And they get rich in a crazy way, they get rich by, you know, co-locating servers next to the
41:40stock exchange and writing algorithms that trade on the instantaneous glitches of the other orders hitting the markets.
41:48Critics of high-frequency trading have spent years in Brussels trying to achieve EU-wide regulation.
41:55The privatization of markets encouraged by the EU's 2004 MIFID-1 directive results in stock exchanges having to look for
42:04new clients, as similar regulations did in the USA.
42:09As a result, high-frequency trading gains an increasing foothold in Europe's markets.
42:14Regulations pertaining to high-frequency trading that were passed by the EU parliament are overturned at the last possible moment
42:22by the finance ministers in the Council of the European Union.
42:27In one reason, the government has gone on.
42:29In a reform process of the Regulierung of the finance markets.
42:32In a long time ago, 2004, 2007, there was already a blessing in the last generation that came to the
42:39pandemic.
42:39That was then to the crisis becoming more or less.
43:01The EU Parliament agrees on a holding period of 0.5 seconds and a cancellation fee designed
43:09to restrict high-speed trading.
43:39The EU Parliament agrees that they are not able to take back.
43:43Of course, the Vertreter of the Hochfrequenz-Handels were against this regulation.
43:50That is completely clear.
43:51And they have, this is my observation, a simple way to the governments of the government.
43:58When the Interests of the Lobbyists of the Hochfrequenz-Handels
44:04meet with the financial markets and plans of the government's countries,
44:13you can easily get the government's countries on the side.
44:18That is obviously happening in the background.
44:22And that has led to this part of the regulation of the MIFID.
44:32At national levels, the problems are very similar.
44:36In Germany, for example, an effective solution seems a long way off.
44:40It is a worrying sign that it is not even possible to say which German companies engage in high-frequency
44:47trading
44:48and how much money they earn.
44:50As far as Gerhard Schick is concerned, high-frequency trading is merely a symptom of a far greater problem.
44:58The financial services are actually a point of the financial services.
45:02But there is a relevant difference in the interest of the money.
45:07One is the financial services of the money.
45:09The financial services of the money is the direct support of the money.
45:12This function is currently in the financial services only a small part of the activities.
45:19The greater part of the money is actually a financial services.
45:26The financial services of the financial services have no more to the financial services.
45:34But on the other hand, from the financial services,
45:37we must say that the last economic crises
45:39have all been solved through financial developments.
45:43In this regard, a stable financial services is a part of the financial services of a stable financial development.
46:07In 2012, the same year that the EU starts working on the MIFID II Directive,
46:13Heim Bodek becomes a whistleblower in the USA.
46:17He is met with powerful opposition,
46:19and the bankruptcy of his company affects him deeply.
46:26To me, it's not about, you know, the possessions, right?
46:32And the status symbols to me, at first I rejected them,
46:35but then I realized at some point that you literally get paid,
46:41in that you get more opportunity, you get bigger bonuses, if you want a boat, right?
46:50They tune your compensation, what they expect your demands are.
46:54That's how this business works.
46:56It's very, very difficult to play the Wall Street game and be modest.
47:02You are not rewarded for modesty.
47:05Now, I will say, after a long period in the business, it did start to corrupt me.
47:11And the whistleblowing was an amazing experience because I didn't think I had been corrupted by the business.
47:17And I will tell you, one of the first times I realized that I was, was when I had to
47:23mow my own lawn.
47:25I live in a neighborhood where people don't mow their own lawns.
47:28So if you go out and mow your own lawn, it's noticed.
47:32And I will say that shedding that was liberating, but it took a lot longer than I thought.
47:42It's disturbing how much those symbols of status and wealth become, you know, essential to people feeling self-worth.
48:15Most of the retaliation was done through the industry quietly.
48:23And, you know, it was kind of a strange mix in that the people who were least supportive were the
48:35victims.
48:36It was completely confusing.
48:39I thought that what I did was providing information that would balance things and kind of repair the industry.
48:49But the victims were basically embarrassed.
48:53What investment bank wants to admit has been routing orders that have been abused for like five years by exchanges
48:58and high-frequency traders?
49:01What pension fund wants to admit that?
49:05Right, so I didn't get the love from that side of the fence.
49:11But what was very strange is I got a lot of attention from the high-frequency traders, especially the high
49:16-frequency trading executives, who kept trying to reel me back.
49:29I will say that the process made me respect the bad guys way more than the rest of the industry,
49:36because they were actually right about the risks.
49:39They were right about the regulatory response.
49:42They pushed things to a level where they could manage it, and they didn't go beyond it.
49:49There was one president of a large high-frequency trading firm that had multiple percent market share, 4% market
49:59share.
50:00And he told me, he says, even if you tell everybody what we're doing, the buy side is so lazy.
50:08They won't read anything, they won't understand anything, they just won't do the work.
50:13You could just put it all out in a map, and nothing will change.
50:18And I said to him, I said, we'll see about that.
50:20So a year later, I'm at a conference where I debated him.
50:25And after the conference, I said, you're totally right, I put it all out there.
50:29And it went whoosh over everyone's head.
50:31I was like, there's, I think, over 500 pages in the federal registrar of changes and documentation on features of
50:43exchanges.
50:44Its entire purpose was to level the playing field.
50:49And I think the only people who really read it were the high-frequency traders.
51:15So when I got into this business, I thought the whole point was to replace the Old Boys Club.
51:21To get rid of it.
51:23And what I learned after 20 years is what we ended up doing is that we automated the Old Boys
51:31Club.
51:32It's alive, it's thriving, it's in the Algos.
52:02And we started to return to our amigos.
52:02We used to remember the Old Boys Club.
52:02He used to remember the Old Boys Club.
52:03It's been the Old Boys Club.
52:11And we used to remember the Old Boys Club.
52:11And that's where we ended up being the old, the old man, the old man, the old man.
52:35Transcription by CastingWords