The Gateway to Algorithmic and Automated Trading

The fragmentation factor

Published in Automated Trader Magazine Issue 31 Q4 2013

Many, though certainly not all, of the recent higher-profile trading glitches have occurred in the fragmented securities market rather than the more liquidity-centric derivatives venues. Jim Overdahl, representing the Futures Industry Association's Principal Traders Group, talks about why responding to the technological challenges can be different for the derivatives industry than for securities venues.

Jim Overdahl

Jim Overdahl

Describe the main difference between the situation for securities and for derivatives:

The main difference (for derivatives venues) is you don't have a linkage system like you have on the securities side within the National Market System in the United States.

On the derivatives side, liquidity tends to migrate to a single trading venue and stay there. For example, Eurodollar futures are traded almost exclusively at the Chicago Mercantile Exchange whereas if you have a product on the securities side, such as IBM stock or some type of ETF, it might be traded at multiple venues, perhaps as many as 50 venues, including exchanges and alternative trading systems, broker networks, dark pools.

On the securities side, the number of venues trading a single product adds to the complexity of trading.

In the derivatives world, related products are also linked. For example, the S&P 500 futures contract is going to be very closely tied to the stocks within the S&P 500 or to an S&P-linked ETF that's traded on a securities exchange, but by and large, on the futures side, you see trading volume and liquidity migrate to the single trading venue.

The main thing is that regulation has promoted a fragmented market structure on the securities side through deliberate policy choices to encourage competition between trading venues. I think, in some ways, if you looked at securities markets like options markets pre-NMS, they behaved very much the same way as derivatives trading venues behave today. And really what you're seeing is the network effect of concentrated trading, that is: a crowd attracts crowd.

Trading on the derivatives side resembles a natural monopoly in that trading costs become lower when more people participate in trading on a single trading venue. There have been instances through time where you saw contracts migrate from one trading venue to another. One example that people point to is the German Bund futures contracts that went from LIFFE to Deutsche Börse. However, these instances are so rare that they are noteworthy when they occur. The economics of moving liquidity requires that trading volume increase to a tipping point in order to attract additional trading volume.

That is why new product development is so important in the derivatives world - to try and be the first to attract liquidity and then have the network effects work to the exchange's advantage when launching a new contract.

On the question of how the dichotomy between the more complex, fragmented securities side and the more liquidity-centric derivatives side affects exchange performance:

A lot of exchanges, depending on their bandwidth, will define how many trades you have to do per number of orders submitted. It may be a fixed number like, say, 100 orders per trade. And they do that, first, to protect their bandwidth. Second, they're encouraging good quoting behaviour, that is, quoting behaviour that adds to market quality, while discouraging quoting behaviour that potentially detracts from market quality. On the futures side, where a single venue hosts the trading of a product, the exchange can say, 'Okay, here's our policy: 100 quotes per trade'. And other exchanges can do the same. In the securities world, establishing such a rule is harder because it has to be coordinated across different trading venues. A single exchange probably couldn't unilaterally come out with an order-to-trade ratio that would really mean anything, because of it being a National Market System.

As an aside to this, we see competition between the venues in setting rules. The Chicago Mercantile Exchange, for example, sets limits on orders-to-trade ratios and then it will revise those ratios depending on how their bandwidth changes over time. The InterContinental Exchange has taken a different approach. It tries to reward good quoting behaviour and discourage behaviour that they don't think is adding to market quality. That's one example where this competition plays out, which is really related to technology (in terms of bandwidth) and promoting market quality.

Another example where you'll see differences between the securities side and the futures side, and this played out after the Flash Crash, is with error trade policies.

On the futures side, an exchange can set its policy and that's that. On the securities side, when individual venues have done this in the past they've had their own policies, but we saw during the Flash Crash that this led to a lot of confusion because traders weren't sure of which rules applied across different trading venues and whether their trades were going to stand or not stand. And if you're a market maker, carefully trying to manage your capital and market making risk, it's really an unacceptable risk to not know whether one side of your hedge is going to be cancelled later and leave you exposed.

One of the things that the SEC did as a regulatory response to the Flash Crash was that it tried to coordinate error trade policies across trading venues to add certainty, so that market participants would not have any doubt as to which trades would stand and which might be cancelled.

On the way forward, in terms of technology:

In general, the FIA-PTG is very supportive of efforts to try to improve the resiliency of trading infrastructure. Firms recognise that a lot of these improvements are going to have to come from the trading venue side, and the firms have been very supportive of these efforts. It's clearly in the interest of all trading firms to have a market infrastructure that they can depend on.

Firms have been working with regulators to promote risk-mitigating ideas. Some of these ideas we saw implemented after the Flash Crash. And new ideas have been advanced more recently. In fact, the FIA recently released a guideline document on drop copy functionality as a best practice control that may help firms reconcile their trades in real time.

The FIA-PTG also put out a white paper on software change management that identifies best practices on how to test trading software at the firm level as a way to help regulators and exchanges create testing environments and best practices for software upgrades. Software testing is an important issue as - at least in one instance - it was a contributor to trading glitch at the exchange level.