Saturday, November 19, 2016

Katsuyama and Flash Boys

The ebullient Brad Katsuyama recently spoke at University of Toronto. It was a wonderful talk, and here are some reflections on what I learned. (Any errors are of course due to my own misunderstanding.)




Why Stock Exchanges Weren't Fair
Brad Katsuyama became my hero when I read Michael Lewis' riveting book Flash Boys.The book describes how Katsuyama, a Canadian working for RBC in New York, discerned how stock exchanges are not unbiased referees in the markets. He would observe the price of a stock, place an order, and, consistently, the fill price would be higher. It was only a small increase, but still, it shouldn't have been there.

Katsuyama initiated a detective hunt to figure out why this was happening. He pieced together that the exchanges had huge conflicts of interest: their  profits at stake motivate them to tilt the odds in favour of their best customers, the high frequency traders.

Traders have many choices of where to places their trades, putting the exchanges in competition with each other. Exchanges entice brokers to place trades with them by giving the brokers rebates. These rebates stayed in the pockets of the traders not the customers on whose behalf they were trading. In order to force traders to choose the best price, not the exchange which offers the best rebate, new regulations were introduced. These regulations required a trader to take up the cheapest offer price when placing an order for shares.

The Law of Unintended Consequences
The regulation had a totally unintended consequence. Suppose you're a trader wanting to buy 100,000 shares of Microsoft.  Large buys like this get filled with purchases on many different exchanges. Before placing your order, you determine the price on your trading terminal - say it's X. But high frequency traders have laid a trap for you - on one exchange they  offer to sell 5,000 shares at, say, 'X minus a tiny little bit'. As a buyer, you buy those shares because they're the cheapest available and because regulations require you to buy at the cheapest price for your customer. But you still have most of your order left to fill - 95,000 shares.

The high frequency traders, knowing about this demand, rush around to other exchanges to buy shares at the quote price X. This demand pushes up the price of Microsoft. By the time you arrive to complete your purchase of 95,000 shares, the price has risen to 'X plus a little bit'. The high frequency trader has bought at X and sold at 'X plus a little bit', a tactic called front running. Thus he's made 95,000 times 'a little bit' in profit. For the original baiting trade, it only cost him 5,000 times 'a tiny little bit' to set this up.

We're Talking Billionths of a Second
You might ask how these high frequency traders beat the buyer to those other exchanges. Well, the laws of physics are at work. Distance travelled equals time. The closer you are to the exchange, the faster your messages get there. Flash Boys opens with a description of cable being laid in an obsessively straight line to minimize transmission time from Chicago to New York. It's a stealth operation so that the land rights sellers don't realize just how valuable their particular piece of land is. Builders don't want the cable to deviate even a foot from the optimum route.

Of course, the best place to be is right in the computer room of the stock exchange. So stock exchanges charge big bucks for you to co-located your computers within their computer room. It takes a billionth of a second to travel over 11.8 inches of cable. So the closer you get to the exchange's computer, the faster you can trade. In order to be able to sell space to many different players, the exchanges introduce the exact amount of coiled cable between a customer and their computer so that all the customers buying space in the room have exactly the same arrival time at the exchange's computers.

The business of selling computer room real estate is hugely profitable for the exchanges, and accounts for their conflict of interest. It behooves them to cater to the high frequency traders - the flash boys - who account for the majority of their profit. The exchanges also make a lot of money selling data, with the high frequency traders again being their best customers.

Feeling the Pain - First Step in Addressing A Problem
Katsuyama says he would never have tackled this problem if he himself hadn't experienced the pain of getting dinged on his own trades. When he first arrived in New York in 2006, he could get his whole order of 100,000 shares filled at the quote price X. Every year, the proportion of the order filled at the quoted price X kept dropping. By 2009, he observed that he could only get about 60,000 of his desired 100,000 filled at the original price X. (Note that I exaggerated the proportions in my example above to emphasize the point.)

Katsuyama didn't initially understand this. But he was a humble and indefatigable searcher for truth.  He was willing to admit his own ignorance and doggedly talked to many people who did understand these things until he figured out what was going on. He had taken the job as Manager of the Algorithmic Programming Team based on the assurance from RBC that he would have a free hand in hiring the best people.  He doesn't think he'd have had the persistence for his painstaking research if he hadn't felt the pain himself. He didn't say this, but I think he was also motivated by his innate sense of fairness which was offended by the way the market was working.

Thinking Differently
Once he understood what was going on, Katsuyama concluded that there was no way RBC could win in this rigged market. Because RBC was trading on behalf of customers, there were checks and balances in the system that would always make them slower than the flash boys. RBC took about 2 milliseconds, whereas the HFT guys were taking 476 microseconds: they'd always be four times faster than RBC.

He'd read Michael Lewis' previous book, Moneyball, which described Billy Bean's unorthodox approach to making the Oakland As a successful team despite the disparity in their budget. It all came down to thinking differently. He decided to apply this lesson to the financial markets.

Katsuyama concluded that the key for RBC was to get slower. Introduce delays in your messages to all the different exchanges so that your offers arrive at the same time and the flash boys can't front run you. In my innovation courses, I call this Letting Go Of What You Know. If the industry 'knows' that the way to succeed is to get ever faster, let's get slower! Selectively slower. Insert a delay into the orders to the closest exchanges so that all your orders arrive simultaneously at all exchanges. Suddenly, you're seeing 100% of your trades getting executed at the original offer price X. A diagram in Wikipedia explains it very simply:


Start trading this way, tell your customers about it, and in one year RBC moves in the trading rankings from #19 to #1.

So what do you do about it? IEX
Now that Katsuyama has untangled the secret of the market, he faces several fundamental choices: stay at RBC and help their customers get fair trades, get very rich by becoming a flash boy himself or reform the entire market by publicizing what he's learned.  

Katsuyama choose the third option and decided to create a new exchange. He has to leave RBC to do this, and the great people he's hired there follow him. They call it the Investors Exchange and it is designed to be fair, simple and transparent. No co-location. No fees for trading data. An even playing field for everybody.

IEX started as an 'alternative stock exchange' but soon reached volumes that required it to get approved by the SEC as a full exchange. Naturally there was lots of opposition from the incumbents, the exchanges and the high frequency traders; IEX was attacking their livelihood. 

Katsuyama described this as Diffuse Harm, Concentrated Benefit. For the exchanges and high frequency traders, there was a lot at stake and they were prepared to fight hard; for the ordinary traders, it was just about fractions of a cent, and there were more important issues for them to worry about. Nevertheless, mostly because of the Flash Boys book, the SEC received more input supporting IEX's application than in their total history before. In the end, the exchanges' insertion of cable in their own computer rooms (to equalizes things for all those who'd bought colocation) set a precedent for the equalization that IEX was imposing and so IEX won the day.


Did Being Canadian Make A Difference?
At the presentation, I asked a question of Katsuyama. I started by saying that, when reading Flash Boys, I'd been incredibly proud that he was Canadian. The audience - a completely packed room - burst into spontaneous, exuberant applause. (The Economist had just put Canada on the front cover as an example to the world. It was US election day and the extent of Trumptastrophe was not yet known.)

Katsuyama, with typical modesty, said he didn't spend much time thinking about himself, but Michael Lewis did feel that only a Canadian would have reacted the way he did - seeking to make the world a better place by creating a level playing field, rather than seeking personal profit. You could almost feel a maple leaf tattoo mystically appearing on everyone's forehead!

Flash Boys the book
I would put this book on your must-read list. It reads like a suspense novel, with high frequency traders as sinister predators and the boyish, charismatic, unassuming Katsuyama as the hero. Lewis explains the technical and financial details clearly. He brings the characters to life. It's just an enjoyable read in every way.

For my other book reviews, click here.



No comments: