The Impossible Fill Price

Client B is a booking error, not a market outcome. A price that never traded cannot be a legitimate allocation. The market trade itself is valid and executed at a real average price; only the internal split is wrong. So the market trade stands - cancelling or busting it would create a second problem and could prejudice the other three clients. What must be done is to re-book B at the true executed price, with the firm absorbing any difference, so B's cash and stock land exactly where they would have been.

Client A is a mandate/weight breach. The 5.4% position breaches a documented 5% cap. Before acting, establish causation: did the excess arise from the allocation (a firm error) or from market movement on a position that was inside the cap when bought? Here the manager bought inside an aggregated order without checking A's cap, so the firm caused it.

Before doing anything, operations should establish:

• The actual executed prices from the execution file, not from the allocation sheet.
• Whether the same keying error hit other clients on the sheet (one bad cell rarely travels alone).
• Whether contract notes, client statements, or ledger entries have already gone out with the 414.1p price - if so, a correction and communication process is needed, not just a re-key.
• Who authorised the allocation and when, since promptness of allocation is itself a control.

Average-price allocation is permitted, but it is a policy decision, not a convenience. It is defensible only where:

• It is written into the order execution/aggregation policy, with the method (pro rata by order size, typically) set out in advance.
• Aggregation is disclosed to clients and the policy acknowledges that aggregation may on occasion work to a client's disadvantage on a given day.
• Allocation is prompt - same-day where practicable. A completion at 16:50 for a 09:14 fill is at the edge; a fill that sits unallocated overnight is not.
• No client is systematically favoured. Rotating who gets the favourable side of a partial fill is a breach even if it averages out.

The manager's argument is therefore not automatically wrong, but it is an argument about policy compliance, not about convenience. If the policy is silent on aggregation and average pricing, the correct answer is that the firm has a documentation gap to remediate - and on this occasion it should allocate at the real price achieved, deal by deal, and record why.

The firm pays the direct cost of the correction - the dealing charges and any spread cost of selling down the 0.4% excess - because the breach was caused by the firm's failure to apply an instruction the firm had on file. The objective is to put A back in the position she would have occupied absent the error.

What would change the answer:

• If the cap was advisory rather than mandatory and A had previously said she was relaxed about small overshoots, the firm should confirm with her rather than unilaterally sell, and document the confirmation.
• If the breach arose purely from market movement on a compliant initial position, the client bears the market movement but the firm still bears the cost of the avoidable delay in monitoring and rebalancing.
• If selling down crystallises a loss, the firm absorbs it; firms should not normally claw back a gain made in error, and the policy should say so, because the absence of that rule is itself a conduct risk.
• If A had given a specific instruction to hold the position despite the breach, the firm would follow the instruction but must record the instruction and the warning given.