Episode 13: Volume Curves and the Fragmented Market
Across the 20 trading days of April 2026, orders in the top 40 French names confined to the central limit order book were exposed to just 44.5% of available intraday liquidity (excludes the closing auction). The remaining 55.5% was spread across systematic internalisers (19.5%), dark pools (14.2%), periodic auctions (11.0%), off-book trading (9.7%), the opening auction (0.6%) and combined intraday auctions (0.4%). This aggregated view belies extraordinary complexity beneath. As our focus is on the intraday trading period, Closing Auctions are excluded.
Volume curves provide essential intelligence for managing market impact, and therefore trading cost, by offering a roadmap for spreading larger trades over several hours. That roadmap is not just a schedule but a guide for navigating fragmented liquidity in a market with up to five lit continuous order books and associated periodic auctions, over ten dark venues of various flavours, and dozens of bilateral electronic and manual connections with varying levels of addressability - for every stock. To optimally capture liquidity, an order needs to be in all places at once, and in the right size.
Try exploring the dataset for individual days. Hover over the chart to see the values for each bin.
Figure 1. Our sample shows the 20 day ADVT in €, and the distribution of trading across the eight types of liquidity pool included (excludes closing auctions). The top panel shows the typical shape and distribution of liquidity across the day. These properties are remarkably predictable but with some impressive and random exceptions. The unknown is the level of the curves on a day to day basis, which as we have seen in recent weeks varies according to volatility and newsflow with liquidity events such as block trades causing random deviations in the roadmap.
The dual chart captures two distinct stories. The upper panel shows the 20 day average value traded across the session in five minute bins, confirming the familiar U-shaped profile; elevated at the open, compressed through mid-morning, and building into the close. The lower panel shows the average composition of liquidity at each five-minute interval, built from a total of over 18,000 datapoints. The shape of the distribution, and the relative proportions between the categories, holds with notable stability throughout the session and with statistical significance.
The opening auction captures on average 21% of liquidity during the first five minutes of the day, with unpredictable delayed auctions sometimes extending this event, and with the auction share of the first five minutes ranging from 12% to 30%. This illustrates that the market open (in broader terms the first 15 minutes of trading) is dominated by liquidity pools other than the auction itself. This is the moment for the biggest systematic contribution to liquidity by conditional dark pools for the day. These venues capture large electronically matched block trades with on average 20% of flows in the first five minutes - matching the size of the opening auction itself. The first five minutes of the day, often maligned for its volatility, is on average the most liquid interval of the intraday session, excluding the closing auction.
It is observed that the average value of dark trading through the day is 14.2% - well above the 7% limit imposed by the regulators - the ‘volume cap’. There are two reasons; the denominator used for the cap includes the closing auction and ‘after hours’ prints - both excluded here. Secondly, the dark trading category contains trades resulting from ‘large in scale’ block orders, that are exempt from the volume cap. We will look at a classic example of this in a moment.
While the open may be the ‘quietest’ time of day for the systematic internalisers they largely track the lit order books with striking consistency, rising from around 8% at the open to approximately 20% by 09:25 and holding that level for the remainder of the day. This reflects the electronically driven nature of internalisation where pricing and capacity is directly conditioned on lit market activity, sometimes with significant boosts from block trade reports. An important caveat for any off order book trading is that the mechanics of the trade reporting rules can give the impression of amplified liquidity due to late trade reports and double counting.
As noted in other episodes of Liquidity Matters and our recent review of Q1 2026 microstructure, periodic auctions hold a steady and substantial share throughout the day, at 11% of intraday flows, which since September rivals the dark pools, and with presence in every time bucket.
While the 20 day average is typically used as a standard for algorithms, it masks substantial day to day variations that underline the scale of the challenge in constructing and maintaining these curves (a daily and intraday exercise).
Figure 2 shows the equivalent data for a single day, April 15th in the same French names when some impressive bursts of trading showed up in the dark trading venues with a beat approximately every hour - commencing with two of almost perfectly matched sizes (€180M each) and indications of connections to related SI blocks visible at 11.45. To put this in context, a 10x spike in dark trades can be expected once a day for this dataset so for four events to turn up in a single day is unusual.
This points to several large French names concentrated into the dark venues at several points over the day reflecting a substantial liquidity event. Hermes International led the way with €1.45bn in total for the full day (2.3x its average for the period) and €176m in the dark pools, clearly contributing significantly to all liquidity pools.
This episode underlines the randomness when the market departs from the average curves and underlines the need for dynamism on an intraday basis - both to identify inflection points in the curve and to identify when to ignore large blocks while monitoring liquidity distribution. Comparing the average to the daily in real time is an important signal for liquidity availability - particularly relevant for implementation shortfall strategies seeking to be opportunistic during favourable bursts of liquidity.
While the statistics show that the general shape of the curves and liquidity pool distribution is on average reliable, the daily picture on a volatile day such as this also shows how dynamic an execution strategy needs to be to stay on track. Variation increases during volatile events, and - crucially - reliability reduces as shares become less liquid. The data is typically fat tailed, reflecting the large and random events such as Hermes International on April 15th. But as we have shown, liquidity tends to revert to the shape of the standard curve quite quickly.
Conclusion
Volume curves matter because trading schedules need to be predictably accurate. They encode both the schedule and the distribution of liquidity across its multiple dimensions simultaneously. The eight aggregate categories illustrated here are only the beginning of the picture. Every stock has its own profile, distributed among dozens of electronic venues, each with their own profile. Engineering and maintenance of these curves, consuming billions of data points per day to ensure precision forms the foundation on which superior execution is built.
Follow us on LinkedIn to stay up to date with the latest series of Liquidity Matters, news and platform enhancements.