MiFID II Article 27 imposes a best execution obligation on investment firms: when executing orders on behalf of clients, firms must take "all sufficient steps" to obtain the best possible result. The obligation is well understood for discretionary trading and single-ticket order flow. For automated portfolio rebalancing, it introduces a complication that regulators and platforms alike have been working through since MiFID II came into force in January 2018.
The complication is this: the best execution for a single trade and the best execution for a portfolio rebalancing sequence are not always the same thing. Optimising each trade in isolation can produce a rebalancing outcome that is, in aggregate, worse for the client than a sequence that sacrifices marginal execution quality on individual trades in favour of a better portfolio-level result.
This piece works through the technical framework, the regulatory obligations, and the practical implications for platforms running automated rebalancing.
What Article 27 requires
MiFID II Article 27(1) requires investment firms to "take all sufficient steps to obtain, when executing orders, the best possible result for their clients taking into account price, costs, speed, likelihood of execution and settlement, size, nature or any other consideration relevant to the execution of the order."
The best execution obligation is not a best-price obligation. Price is the most important factor for most retail equity orders, but the regulation explicitly enumerates other factors — costs, speed, likelihood of execution — that firms must weigh. For a portfolio rebalancing sequence that spans multiple exchanges and involves both sells and buys, these other factors become material.
Firms must establish, maintain, and implement execution policies that specify how they weigh execution factors for different order types and client categories. For retail clients, price is typically given overriding importance — but the policy must address the rebalancing context specifically, because the relevant "price" for a rebalancing event is not the individual trade price; it is the net portfolio cost of the entire sequence.
It is worth noting what Article 27 does not require: it does not require firms to achieve the lowest possible price on every individual order. The "best possible result" is assessed against the firm's own execution policy and the client's best interests in aggregate. A firm that consistently achieves better-than-market execution on 80% of orders and marginally worse on 20% is not in breach of Article 27, provided its policy is documented, its venue selection logic is defensible, and its ongoing monitoring demonstrates the policy is working.
RTS 28: annual reporting obligations
Regulatory Technical Standard 28 (Commission Delegated Regulation 2017/576) requires firms to publish, annually, the top five execution venues for each class of financial instrument, along with a summary of the quality of execution obtained. This is a disclosure requirement: firms must be able to demonstrate that their execution venue selection was consistent with their execution policy and produced good results for clients.
For platforms running automated rebalancing at scale, RTS 28 reporting requires the infrastructure to record, per order, which venue was selected and why — the routing rationale at the time of execution, not a reconstructed post-hoc justification. This requires structured audit logging built into the execution path, not appended after the fact.
RTS 28 also covers the summarisation of execution quality outcomes. For a platform routing rebalancing orders across XETR, XPAR, XLON, and XSTO, the annual publication must address each exchange class separately and describe whether execution quality metrics (typically spread cost versus benchmark mid, or arrival price) were consistent with the policy's stated objectives. A platform that has only portfolio-level records — "rebalance X achieved target allocation" — without per-order venue and cost records cannot produce an RTS 28 summary that satisfies supervisory expectations.
The single-trade versus portfolio-sequence tension
Consider a simplified rebalancing scenario. A portfolio requires selling 3,000 EUR of a cross-listed ETF (available on both Xetra and Euronext Amsterdam with similar liquidity) and buying 3,000 EUR of a Nasdaq Nordic-listed Swedish equity (XSTO only). The rebalancing engine has two choices for the sell leg:
- Option A: Execute the sell on Xetra at a marginally tighter spread (0.5bps better price), then execute the buy on XSTO. Total slippage: 8bps.
- Option B: Execute the sell on Euronext Amsterdam at a marginally wider spread (0.5bps worse price), but sequence the sell and buy to execute within the same 4-minute window before XSTO approaches its close auction. Total slippage: 6bps because the buy benefits from better depth during continuous trading.
Option A is better on the sell. Option B is better at the portfolio level. A trade-by-trade best execution assessment would select Option A. A portfolio-level best execution assessment would select Option B.
This is not a theoretical edge case. European equity ETFs are routinely cross-listed on two or more primary venues with spreads that differ by less than a basis point. The execution sequence — which trade goes first, which exchange is used for each leg — has a larger impact on the portfolio-level outcome than the per-trade spread optimisation on any individual order.
How to construct a portfolio-level best execution policy
The FCA's guidance on best execution (FG12/15 and subsequent consultation papers) and ESMA's Q&A on MiFID II best execution both acknowledge that portfolio management involves aggregating orders across clients and instruments, and that the "best result" must be assessed in the context of the investment management objective, not merely the individual trade.
For automated rebalancing, a defensible best execution policy needs to address four specific questions:
1. What is the relevant execution unit?
Is best execution assessed per order, per instrument, or per portfolio rebalancing event? For retail automated rebalancing, the most defensible answer is per rebalancing event. The objective of a rebalance is to return the portfolio to its target allocation at minimum cost — and minimum cost is a portfolio-level measure, not a per-trade measure.
This framing should be explicit in the execution policy. If your policy states that best execution is assessed per order, you will struggle to justify routing decisions that sacrifice marginal per-trade execution quality for portfolio-level efficiency.
2. How are execution factors weighted for each order type?
For retail equity, MiFID II presumes that price is the dominant factor. But for a rebalancing event, the weighting of execution factors is more nuanced:
- Price: Important, but assessed at the plan level (total net cost of the rebalancing event), not per trade
- Cost: Includes FX conversion, exchange fees, and any broker commission — assessed on the full sequence
- Speed / likelihood of execution: Higher weight than in discretionary trading because the rebalance has a time-cost; a stale plan (triggered by drift that has since partially reversed) may be better not executed if it cannot be completed within the market session
- Settlement: Particularly relevant for multi-exchange rebalances where settlement dates differ; the engine must ensure that sells scheduled for T+2 settlement don't inadvertently place the portfolio in a failed-settlement position
3. How is the execution quality monitored and reported?
Best execution is not a one-time policy declaration — it requires ongoing monitoring. For automated rebalancing, this means running transaction cost analysis (TCA) at the plan level: comparing the estimated cost of each rebalance plan at creation time to the realised cost at settlement. Systematic deviations — the realised cost is consistently higher than estimated — signal that venue selection or sequencing logic is underperforming the policy's stated objectives.
RTS 28 reporting requires firms to document execution quality outcomes. For platforms running thousands of rebalancing events monthly, this means maintaining a queryable record of every venue selection decision, the rationale for it, and the realised execution quality. Without structured audit logging at the order level, RTS 28 compliance becomes a manual reconciliation exercise.
4. How are conflicts of interest managed?
For vertically integrated platforms where the platform also operates the execution venue or has revenue-sharing arrangements with specific brokers, the best execution obligation includes a requirement to demonstrate that the venue selection is not influenced by the platform's own commercial interests. This is a governance question as much as a technical one, but it has technical implications: the venue selection logic must be auditable, and the audit trail must demonstrate that commercial considerations did not override execution quality factors.
The rebalancing-specific reconciliation framework
A practical framework for demonstrating best execution compliance on automated rebalancing rests on three pillars: plan-level cost measurement, venue selection documentation, and post-execution review.
Plan-level cost measurement
At plan creation, the rebalancing engine should compute an estimated total cost for the rebalancing event, expressed in basis points of portfolio value. This estimate includes: estimated bid-ask spread costs for each order based on current market depth, FX conversion costs for any cross-currency trades, exchange fees, and an estimate of market impact for larger orders.
At settlement, the engine should compute the realised total cost using actual fill prices and settlement costs. The difference between estimated and realised cost — execution slippage at the plan level — is the primary TCA metric.
A well-calibrated rebalancing engine should produce positive slippage (better-than-estimated) roughly as often as negative slippage. Persistent negative slippage (realised cost consistently exceeds estimate) indicates either that the cost model is poorly calibrated or that execution quality is deteriorating.
Venue selection documentation
For each order in a rebalancing plan, the audit record should document:
- The set of eligible venues for the ISIN (based on the venue connectivity available)
- The rationale for the selected venue (typically: highest liquidity, tightest spread, base-currency match, or lowest FX conversion cost)
- The estimated cost of the order at the selected venue versus the best alternative venue
- The timestamp of the venue selection decision
This record does not need to be human-readable in real time — it is a structured log that supports ex-post review and RTS 28 reporting. The important property is that it is generated at execution time, not reconstructed afterwards.
Post-execution review
Best execution policies under MiFID II are not static documents. They require regular review: at least annually, and whenever there is a material change to the markets or the platform's execution arrangements. For automated rebalancing, this review should include an analysis of plan-level TCA trends over the review period: are execution costs stable, improving, or worsening? Are there patterns by exchange, time of day, or instrument type that suggest the venue selection logic needs recalibrating?
The review should also assess whether any new execution venues should be added to the eligible set, or whether any existing venues have deteriorated in execution quality to the point where they should be deprioritised.
Supervisory expectations: how the FCA and BaFin approach this
The FCA and BaFin have both issued guidance and supervisory letters on best execution quality since MiFID II's 2018 implementation. A consistent theme in both regulators' supervisory focus is the quality of post-trade monitoring: firms that have written execution policies but no systematic analysis of whether those policies are producing good outcomes are more likely to receive supervisory attention than firms with genuinely functional TCA infrastructure.
The FCA's multi-firm review on best execution (2019) found that many firms were still assessing best execution on a per-trade basis without any portfolio or client-level aggregation. For portfolio management services specifically, the FCA's view was that the best execution assessment should reflect the portfolio management objective — which for a rebalancing service means tracking how well the rebalancing sequence achieves the target allocation at minimum aggregate cost, not how each individual order ranked against its contemporaneous market mid.
BaFin's focus, as communicated through its supervisory priorities for investment services, has been on the transparency of venue selection rationale and the adequacy of execution quality monitoring systems for retail-facing platforms. German platforms routing retail rebalancing orders through to XETR and other domestic venues need to demonstrate that venue selection is driven by execution quality analysis rather than default routing convenience.
Neither the FCA nor BaFin requires platforms to use any specific TCA methodology. The standard is whether the methodology in use is appropriate for the platform's order types, client profile, and execution arrangements — and whether it is actually producing data that management reviews.
Common failure modes in rebalancing best execution
In practice, the most common compliance gaps in automated rebalancing best execution fall into a few recurring patterns.
Static venue mapping: Routing each ISIN to a fixed default exchange without dynamically considering spread, depth, or FX cost. This produces defensible average outcomes but misses easy optimisation opportunities and creates audit trail records that don't reflect any actual best-execution analysis.
Per-trade TCA without plan-level aggregation: Measuring best execution trade-by-trade but not measuring the aggregate portfolio cost of the full rebalancing sequence. This misses the sequencing cost described earlier and makes it impossible to detect systematic deterioration in plan-level outcomes.
Audit logging at the portfolio level without order-level detail: Recording the rebalancing event outcome (portfolio moved from X allocation to Y allocation) without capturing the per-order routing rationale. Sufficient for internal performance monitoring, insufficient for RTS 28 demonstration.
Best execution policy that doesn't address rebalancing: A generic best execution policy designed for discretionary trading, applied unchanged to automated rebalancing. The policy needs to explicitly address the rebalancing context — particularly the treatment of portfolio-level cost versus per-trade cost — to be defensible in a supervisory review.
What platforms should ask their rebalancing infrastructure provider
If you are procuring rebalancing infrastructure rather than building it, the best execution question has implications for what you should require from your provider. Specifically:
- Does the engine support dynamic venue selection based on real-time spread data, or does it route to static venue assignments?
- What cost data is captured at plan creation time, and how is it compared to realised execution cost at settlement?
- What does the per-order audit record contain, and can it be exported in a format suitable for RTS 28 reporting?
- How does the engine handle cross-currency trades — is the FX cost included in the best execution assessment, or is it handled outside the engine?
- Does the provider offer plan-level TCA reporting, or only per-order fill quality?
We are not saying that every platform needs bespoke MiFID II best execution infrastructure built from scratch. What we are saying is that a platform relying entirely on its broker or custodian's best execution framework — without any plan-level TCA or venue selection audit capability of its own — is transferring a compliance obligation that regulators expect it to own.
MiFID II best execution for automated rebalancing is technically achievable and regulatorily required. The platforms that build it into the architecture — rather than retrofitting it as a compliance overlay — are the ones that will have defensible answers when the FCA or BaFin come asking.
Best execution at the portfolio level is not harder than best execution at the trade level. It is just a different optimisation target. The infrastructure has to know which one it is solving for.