The debate over high frequency trading (HFT) and whether it will lead to a real life Terminator scenario with computers taking over the world, heated up again with the Knight Capital Group glitch. However, the doomsday scenario seems less likely as details of the Knight case have come out that show it had more to do with a lack of pretty ordinary due diligence than computers run amok.
What failed Knight was a faulty algorithm and poor testing. And, more generally, perhaps the need to be precise in slicing out a tiny edge in the market and then leveraging it up in order to make that tiny edge profitable. Despite all the warnings regarding HFT that involve spooky computer programs, what these traders do is not very different from what the old floor traders have always done. Attempt to find an edge and exploit it. There is this mass of order flow out there from people who are using the market to hedge risk and speculate on the long-term performance of the underlying instruments. Short-term traders have always been content to provide liquidity by constantly making a market and finding small edges. The faster they were, the better access to information and order flow they had, the more successful they could be.
The problem with HFT is that so many people are doing it. It is similar to certain arbitrage trades that exploit small inefficiencies in the market. As more people discover an edge they compete over the same inefficiency until it goes away. But if more volume in the market is coming from folks trying to exploit an inefficiency than the more classical hedgers and speculators, you have a problem.
The question with HFT is if it is a problem that needs to be dealt with through regulation or if it can moderate itself. Are they distorting the market to the detriment of real price discovery or are they simply over extending a slight edge and will go the way of other unsustainable strategies?
In the old floor trading days there were traders who would come in with huge size on the bid or offer. Some suspected these large players where trying to scare the market. They would come in with 1, 2 even 3,000 contracts on the bid or offer. This at times had the effect of scaring folks already on that bid or offer to simply give up the edge. There are those that speculated that was the intention and if they were say 110-18 bid for 2,000 bond contracts, they were secretly offering 200 at 19. This would happen often enough in the 30-year bond pit that some of the big boy locals like Tom Baldwin or Charlie DiFrancesca would occasionally try and keep these folks honest.
Sometimes the only thing worse than slippage is a lack of slippage. If you are 18 bid for 1,000 with the hope of actually selling 19s, it can be disconcerting to get hit on your 18s and have a local offer you 1,000 more.
While electronic trading has helped the markets grow that is perhaps the one thing missing. A large player keeping things honest; hitting that odd bid or lifting that odd offer that appears to be nothing but a loss leader. With all the iceberg orders and algos out there someone to let you know that if you place in order it may be hit or lifted.
HFT is not monolithic and its effects are different depending on the market structure. It is a completely different thing in securities markets, which allow payment for order flow, internalization and maker taker rules and the futures markets with pure price time priority. Perhaps that is why it is more prevalent in securities markets. But the rule is that each order and each trade is exposed to risk.
Shortly before Knight’s faulty algorithm cost them $400 million, the Futures Industry Association’s European Principal Traders Association put out a paper titled: Market Integrity Framework: Best Practices to Preserve Market Integrity.
In it they stated that “Market manipulation can be characterized by any of the following: 1) Transactions or orders that give a false or misleading signal or secure the price of a financial instrument at an artificial level; 2) Transactions or orders that employ fictitious devices and/or other forms of deception or contrivance; and 3) Distribution of information likely to give false or misleading signals.
This seems like a logical framework for regulators to work with. Every order must be exposed to risk. When bandwith was a concern some exchanges floated the idea of charging a fee per order into the market. Perhpas that could be reexamined? Perhaps market structures that advantage whether you are a market maker or taker needs to be looked at, particular in liquid markets that execute, hundreds, even thousands, of transactions per second. In such a market there is little need to provide incentive to place resting orders.
When payment for order flow became prevalent, there were traders that created programs that would be successful simply based on that. One thing we know about traders is that they will breakdown any market structure, any matching algorithm, to find an edge. That is why the simplest most straight forward, most transparent rules are best.