Cutting dark pools down to size

The headline issue for the European Commission's review of MiFID is dark pools. Introducing a threshold for order sizes on dark pools could threaten the models of some but equally could improve the buy-side's ability to trade large blocks.
By None

The headline issue for the European Commission's review of MiFID is dark pools. A substantial number of non-displayed venues, particularly crossing networks operated by brokers, have attracted attention because they do not easily fit any of the existing categories set out by MiFID for trading venues and are therefore perceived as having an unfair competitive advantage, both by regulated markets and multilateral trading facilities.

Dark pools operate under MiFID using waivers granted by the Committee of European Securities Regulators that free them from the obligation to publish pre-trade prices according to factors such as large trade sizes or matching trades at the mid-point of European best bid offer, compiled using reliable trade data. Some broker-operated dark pools match at the broker's discretion within the spread, as an automated version of the pre-existing practice crossing clients' flow. Others cross their own proprietary flow with that of clients trading in their dark pools.

In principle the crossing network allows buy-side firms to trade large orders while minimising price impact. But in practice the fragmentation of liquidity in the European market has left buy-side firms splitting orders up across multiple trading venues lit and otherwise, to minimise risk of an order not completing.

The slicing of orders has left the order sizes being placed on lit and unlit orders at very similar sizes. In its ”Let there be light' monthly liquidity report on European dark trading venues, boutique US agency broker Rosenblatt Securities noted that the average execution size in the largest European dark pools for October ranged from 1,831 shares / €16,576 (Nomura) at the top end, falling to 1,024 / €9,270 (Chi-X Europe's Chi-Delta) down to 593 shares / €5,010 (Citi Match) with the majority of venues trading closer to the latter than the former.

Liquidnet Europe, which operates under a large-in-scale waiver, has had an average trade size across 2010 of €1.049 million.

Comparatively, in October the average execution size for FTSE 100 index shares on the London Stock Exchange was €10,340, and for CAC 40 index shares traded on NYSE Euronext it was €9,775, according to agency broker CA Cheuvreux.

This suggests that one of the perceived advantages for buy-side firms of using dark pools, that of trading large orders, is in fact negated by the widespread use of dark pools for processing small order types. If trader lacks the confidence that a pool has sufficient liquidity on it, large orders will not be placed. If they place a large trade in a pool and only have a small portion completed, they risk giving information to the market that a larger order is being traded.

This adds up to a situation in which a critical mass is hard to build up without taking potentially costly risks. In its consultation on MiFID II, expected to be launched by the 8 December 2010, the EC has said that it will create a new regime of regulatory oversight to ensure that broker crossing networks (BCNs) are regulated and asked for feedback on the idea of placing thresholds on BCNs to have them categorised as multilateral trading facilities, systematic internalisers or organised trading facilities.

Depending on the size of the threshold imposed on venues then dark pools businesses could cease to function as they currently do. Although the size of threshold has not yet been proposed, the gap between execution size on lit and dark venues is such that if the average was based on the average trading size for blue chip stocks on major European venues for October, €12,397, only Nomura's venue would be eligible. In that instance, it may be that liquidity would aggregate onto fewer dark pools, thereby offering the chance for larger orders to be traded safely.

Click here to vote in this month's poll.