The Best of Both Worlds

Introduction

On 20240520, the Xetra Retail Execution service went live. It is a very elegant solution to a multi-faceted problem.

#1: Suppose you are a market maker. I would happily quote a retail investor a better price than a broker or a prop firm. Not out of the goodness of my heart but because I can be pretty certain that this is "low-information" flow. You are not trying to latency-arb me, you are not planning to trade much more volume in the same direction over the next few hours. Unfortunately, on a lit central limit order book I cannot pick and choose my counterparty. My prices hence include the anticipated adverse selection bias.

#2: The bulk of German cash equity retail volume is traded off the primary and any of the large MTFs. Most is traded on single-market maker venues which effectively resembles the US market structure where market makers pay brokers for their flow (PFOF - payment for order flow). However, these venues are much less liquid than e.g. the primary and their prices follow Xetra with some latency.

Xetra Execution Service in a Nutshell

The new model on Xetra addresses both. I will describe it colloquially - please consult the official documentation. The scheme works by introducing RLP (retail liquidity providers) orders. These are normal limit orders, which are flagged as being submitted by an RLP. They are treated differently from other orders in two important aspects:

• They only interact with RMO (retail member organization) orders, i.e., orders originating from retail end clients (via brokers).
• They are not displayed in the book (see footnote).

Otherwise, they behave like a normal limit order and obey price-time priority. Note that the first point means that RLP orders could be in cross with the "normal" book (or even with other RLP orders on the opposite side) without a trade occurring. This necessitates the second point as crossed books (and volume not accessible to the whole market) would cause issues in most trading platforms.

Everybody can register as RLP or RMO - provided, obviously, that certain requirements are met.

Mechanism

For incoming "normal" orders, the presence of RLP orders has no effect. Only incoming RMO orders can "see" RLP orders. Hence, when there are RLP orders at a better price than the "normal" touch, the retail investors (and only them) will benefit from price improvement. Otherwise, they benefit from the full liquidity at the touch including that from any RLP orders at this price. They get the best of both worlds - price improvement and full liquidity if required.

Crucially, this does not remove the retail flow from the book. It does not fragment the market as much as trading them on a separate venue. Retail flow remains accessible on the primary market and ensures a healthy mixture of flow. Plus, non-RLP orders can still interact with retail flow on the touch where RLPs have no special privilege (price/time priority is obeyed).

The Full Book

Fig. 1 below shows the order book for 1&1 (an SDAX constituent) in the afternoon of 20240610. The left (right) half shows buy (sell) orders at the respective price levels. As it is an SDAX constituent, the book is comparatively thin (only 2661 updates between the open!). RLP orders are shown in light blue. These would normally not be seen in the level-3 EOBI market data. This snapshot shows RLP liquidity at a better price than the "normal" touch on both sides. The spread is one tick for retail orders vs. three ticks for non-retail order.

For the sake of completeness, a full explanation of the various elements of Fig. 1 (not relevant for the purpose of this article):

The central dark blue bar shows the prices. The narrower grey bands on either side would display the traded volume at the respective level over the previous 10 seconds. But there was none at the time.

The two horizontal red and green bars represent the aggregate volume on each level. Individual orders are separated by thin lines. Those orders which will eventually be traded are highlighted by a narrow, dark blue bar. The narrow red and green sections denote orders which were inserted within the first millisecond of the level's lifetime. GTC (good-till-cancel orders) from the previous day would be shown in a more solid color but none are present in the chosen snapshot.

The numbers in the outer grey bands are the aggregate volume on the level and, in brackets, the number of orders on this level. RLP orders are not included.

The bottom panel, finally, shows the price and volume chart up to this point. The same script is also used for Xetra - hence the 24h range.

Some Early Stats

Over the first three weeks, RLP orders quoted a tighter spread 12% of the time for DAX constituents, 30% of the time for MDAX constituents, and almost 50% of the time for SDAX constituents. RLP orders traded in over 300 distinct instruments on average every day. Half of their traded notional has provided price improvement to retail investors.

Footnotes

There exists a separate real-time data feed for the "RLP book". In addition, the RMO-to-RLP trades are published in the EOBI feed as TradeReport updates.