A loyal MR reader on high-frequency trading

He/she writes to me:

Since there is a lot of confusion in the media regarding HFT, I wanted to clear some of it up.  Disclaimer:  this is my personal view, I am not representing any firm or group and I wish to remain anonymous.

Colocation is a practice, whereby any market participant can pay the exchange a fee which allows them to locate their trading computers in the same building as the exchange itself ( the matching engine ).

Colocation is actually good for investors.  Why?  Suppose a mutual fund from Kansas City wants to execute orders to buy stocks on an exchange which physically located in New Jersey.  If this mutual fund is looking at the market data feed directly from Kansas City, then it is at a disadvantage relative to investors who just happen to accidentally be in New Jersey.  So, what are the options?  Well, the fund can either rent space in a New Jersey data center or execute through a bank/broker which is doing exactly that.  However, all New Jersey data centers would still have to somehow connect to the exchange.  And their location within New Jersey would matter – some data centers would be better to locate at relative to some other ones.

With co-location, all investors are given the opportunity to trade from the right data center – the same one that houses the exchange.

If exchanges operated on some sort of discrete auctions rather than continuous matching, some of the issues above would be mitigated. But other issues would arise, for example, lack of synchronization of auctions between different exchanges, as well as increased incentives to trade more prices within a given amount of time, thus generating increased volatility

Furthermore, exchanges, as private for-profit businesses, are able to grow their profit and bottom line by selling co-location services to speed sensitive traders, thus mitigating the need to grow the revenues in other ways, such as raising commissions and trading fees on all investors

Direct Data Feeds and Accusations of Insider Trading

Exchanges sell access to direct data feeds to all investors.  When high frequency traders subscribe to a real time direct feed in the colocated facility and they observe the order book as well as trades, they have no idea who is trading – a customer, a big bank or another HFT firm.  They see the same exact trades in this feed as all other market participants.   Many, if not most,  HFT firms do not deal in any way with customers whatsoever.  The ones that do are supposed to have a clear separation (a Chinese Wall) between customers and proprietary trading, so no customer information can flow through to the prop desk – the same thing is true of big banks and other broker/dealers.

Consider, for example, the trade that is described in Flash Boys. HFT places 100 shares on the offer at 100.01 on BATS and when it trades, it goes and buys the same offer price of 100.01 at Nasdaq, thereby running up the price to 100.02 and selling back the liquidity at 100.02 to the non-HFT customer

This doesn’t actually work.  Why?  because in order to sell successfully at 100.02, HFT algorithm had to have priority in the order book queue and the only way to do that is to continuously quote that price, not knowing at all whether or not anyone is going to buy at 100.01 on BATS, and facing market risk of market running though 100.02

Secondly, just because the offer price on BATS traded, that doesn’t mean that the market cannot go down right after that and the HFT has no way of knowing who bought the price they bought and why they bought it.  It could have easily been a fund or a retail trader who have no alpha (no predictive power ).

Contrast this with the main protagonist of Flash Boys Brad Katsuyama, who was getting paid multiple millions at RBC for executing customer order flow.

Suppose a customer called Brad and asked to buy 3 million shares of IBM right now and 2 million potentially later in the day

The current market on public ( “lit” ) exchanges is 188.00 bid 188.01 offered

Brad says, ok, I will do the deal for 188.15 ( 14 ticks above the current market ) and then proceeds to work the 3 million shares in the market in order to buy back what he sold at a price lower than 188.15

Now, Brad is trading in the market, having material non-public information about his customer’s order flow.  This trading has to be immensely profitable for the bank to pay Brad millions a year.  So to the extent that there is any insider trading going on, and not market making, it is being done by Brad and not by the HFT algorithm because Brad really is using truly non public information and HFT is reading the direct data feed available to anyone who cares to sign up for it

Bank Equity Trading Revenues

http://www.economist.com/news/special-report/21577187-trading-equities-barely-profitable-these-days-many-banks-are-carrying

Indeed, as a result of HFT which quotes much tighter spreads than the banks, bank equity trading revenues have gone down dramatically, by much more than the profits generated by HFTs in equities.  Where did the difference go?  it accrued to investors in the form of lower trading fees

Ok. Is There Anything Wrong With US Equity Markets?

Yes.  Dark pools are destructive to all investors, including HFTs.  An HFT firm that derives its edge from mining statistical patterns in the data cannot do this very well for dark pool data because it is simply not available in a clean format.  But neither can any other investors

In order to promote transparency and reduce conflicts of interest between broker/dealers and their customers, our regulatory agencies should force all equity trading to happen on lit exchanges

By the way, this is a problem not just with equity dark pools.  Consider trading in off-the-run treasuries ( off the run meaning not the latest issue ) or interest rate swaps, or many other securities that Wall Street banks trade for their customers

Since none of these are on the exchanges, and there is no transparent data, effective spreads paid by customers are wide, lining the bank pockets.  If regulators were able to force trading in these instruments to happen on exchanges, it would reduce fragility of the financial system and create pricing transparency

What about Reg NMS

Reg NMS was designed to make sure that if a better price is available on any exchange, that price has to be filled before taking out worse prices on other exchanges

In practice, it has created a lot of complexity in the market and forced market participants to all implement their own software solutions to comply and monitor with it

All this compliance should be fully shifted to exchanges and other trading venues themselves, where it’s much easier for regulatory bodies to verify compliance

Comments

Comments for this post are closed