Episode 9: Implicit Trading Costs Matter Part 2 - Liquidity Pool Mechanics
Intraday market volumes are fragmented across five main liquidity pool types, each with distinct cost characteristics, creating a complex picture of relative trading performance.
The European market comprises 24 country primary exchanges and MTF lit order books, more than 30 dark pools, periodic auction venues, and integrated market mechanisms, and hundreds of bilateral inter-participant connections. We combine this mosaic into five main categories, and this episode examines each type of liquidity pool, on and off the electronic order books using the mark-out methodology described in Episode 8 of Liquidity Matters.
Our benchmark, shown in Figure 1, is the observed, absolute, weighted average change in the mid price at a series of mark-outs before and after a trade in a given venue type. For lit continuous order book trades (blue line) this is 2.93 bps at one millisecond and 14.78 bps at five minutes.
Measured over a five-minute window, off-order-book bilateral Negotiated Trades (in yellow) show the lowest implicit costs. For the most liquid UK names (UK100), this pool records 4.72 bps versus 14.78 bps on lit order books, a 68% cost advantage of nearly 10 bps.
This reflects the market-neutral nature of balanced block trades: price impact only emerges when one counterparty is substantially larger and continues seeking liquidity elsewhere.
Figure 1: Implicit trading cost measured by changes in market price - ‘mark-out’. The change in price is calculated as the difference between the Primary BBO mid at t=0 (time of trade report) and at t=n (a series of offsets in milliseconds) throughout a ten minute window from five minutes before to five minutes after the trade report for the given venue type. Longer mark-out series, including multi day are also available.
Periodic auctions (in purple) carry the lowest market impact among on-order book venues with immediate mark-outs of 0.3 bps, rising to 13.58 bps at five minutes, 8% less than lit continuous order books. This may be contributing to the finding in Episode 7, that periodic auctions have become the preferred alternative for intraday trading.
The frequent batch mechanism repeats auctions at short intervals throughout the day, attracting aggressive orders that capture spread and passive orders that bypass the lit queue, without requiring ultra-low-latency infrastructure. Execution probability has historically been lower, but rapid market share growth may now be creating a virtuous circle.
Dark pool trades (in pink) are pegged to the mid of the most liquid reference market BBO, incurring a half spread cost of around 3.5 bps for UK100 for passive orders, or saving half a spread for aggressive orders. In isolation, the originating order creates no impact on the market mid-price because it does not interact with orders on any lit order book. However, the trade report signals intent to the market. Regular or larger dark pool trades amplify this signal, particularly when combined with parallel lit order book trading.
The mark-out data reveals the consequence. At one millisecond mark-out, dark pool trades show 1.38 bps of cost - 47% of the equivalent for lit continuous trades. But, unlike the other liquidity pool types, this advantage does not persist. After one minute, the position flips, and by five minutes, dark pools reach 16.01 bps compared with the lit order books 14.78 bps - 8.3% larger. This reversal reflects the larger average trade sizes in dark pools, which are further skewed upward by conditional venues where order sizes must be above €1m.
Liquidity pools are closely linked. Small sized dark pool trades may execute with minimal price impact, much as on lit order books. A large buyer or seller, however, may exhaust dark pool liquidity and must continue seeking supply in lit venues, thereby transferring market impact to other venues. Market impact is contagious.
Systematic Internalisers (SI) (in green) are the fifth liquidity pool represented. Approximately one third of SI trades are automated, executed at trade sizes comparable to lit orders and informally pegged to the EBBO with almost no initial mark-out. This shows the nature of off order book SI electronic trading pools, which report their trades almost instantly. They don’t move the market because they do not interact with the Lit order book and are normally reported at or inside the EBBO.
Larger SI trades that are filled from facilitation and central risk books, often with some manual intervention, are subject to a one minute reporting window. The mid-price drift at the 60 second mark-out reaches 4.87 bps compared to 7.76 bps for the Lit order books (saving 37% in implicit costs) and 8.71 bps for the Dark order books. For a breakdown of off order book trade size analysis, refer to Episode 2 of Liquidity Matters.
Figure 1 also shows the most liquid Madrid listings, in the ES35 index, where mark-out results differ significantly across all liquidity pool types and time horizons. This difference reflects network and infrastructure latency. The primary market matching engine is based in Madrid, while most MTFs are physically located in London - a distance of over 1,260 km. As a result, the primary market lit order book price updates approximately 9 milliseconds later than the UK100 equivalent in response to a trade on an MTF dark pool. The UK100 and ES35 data are not directly comparable as the names have different liquidity profiles, but the systematic delay shows that lit order book market prices that are further from the MTF matching engines incorporate prices from other liquidity venues more slowly. Even though it is tiny, it is long enough to make a difference to price formation.
Conclusion
Post-trade mark-outs reveal implicit trading cost differences across types of liquidity pool and individual trading venues. Every stock and venue has distinct characteristics. Latency varies across geographies. Market conditions change. The mark-out dataset is complex to create but powerful in helping to measure execution performance by investigating the contribution of every fill to overall benchmark slippage and identifying elements that lead to the best and worst performing trades.
Regular and frequent measurement of fill level performance provides feedback into strategy selection and routing tactics (such as venue prioritisation, limit selection, sizing and order book join speed). Pre-trade applications help optimise market impact estimates and trading speed.
Follow us on LinkedIn to stay up to date with the latest series of Liquidity Matters, news and platform enhancements.