T+1 Post-Trade Control Tower Playbook (for Small Quant Teams)

Date: 2026-03-10 (KST)

TL;DR

If your strategy can place orders automatically but your team still reconciles allocations, fees, and settlement status manually at night, you are one incident away from a preventable PnL leak.

Under T+1 settlement, the practical goal is simple:

• Close economic truth on T day
• Wake up on T+1 with zero surprise breaks

A lightweight control tower is enough if it enforces:

1. one canonical trade ledger,
2. explicit cutoff checkpoints,
3. owner+SLA for every break,
4. immutable daily evidence snapshots.

1) Why T+1 changes daily operating posture

T+2 tolerated “fix tomorrow morning.” T+1 punishes it.

The problem is not just settlement failure. The real risk is a chain reaction:

• late mismatch discovery,
• rushed manual edits,
• stale position/PnL base,
• bad risk decisions next session.

Treat post-trade as a same-day reliability domain, not back-office cleanup.

2) Control tower scope: what it should own

A practical control tower for a lean desk should own four streams:

1. Execution truth: fills/cancels/corrections from OMS + broker + venue/drop copy
2. Economic truth: gross/net notional, commissions/fees/taxes, FX conversion basis
3. Allocation truth: strategy/account/book mapping + average price consistency
4. Settlement readiness: affirm/confirm status, SSI validity, expected settle date

If these four are not joined in one view, the team is blind by construction.

3) Timeline model (T day → T+1)

Phase A — Intraday (execution window)

Objectives:

• ingest all execution events with low lag,
• reconcile deterministic keys in near real time,
• open break tickets immediately when deterministic match fails.

Hard checks:

• duplicate exec ID with different economics,
• side/symbol/account mismatch,
• correction chain gap.

Phase B — Pre-close hardening (last 60–90 min)

Objectives:

• freeze reference mappings used for EOD (symbols, books, commission tables),
• reduce unresolved economic breaks below desk threshold,
• detect concentration (single broker/venue/source failures).

Phase C — EOD hard close

Objectives:

• produce immutable “economic truth snapshot” for the trade date,
• explicitly classify every unresolved item as accepted risk or escalated issue,
• record sign-off actor/time/reason.

Phase D — T+1 pre-open readiness

Objectives:

• verify affirmation/confirmation completeness,
• verify settlement instruction validity (SSI, account, custodian routing),
• verify no overnight drift between EOD snapshot and morning broker confirms.

4) Minimal data contract (don’t skip this)

Use one canonical row model for “execution economic event.”

Required keys (minimum):

• tradeDate, symbol, side, qty, price
• execId, orderId, parentOrderId
• account, book, strategyId
• broker, venue, currency
• commission, fees, tax, grossNotional, netNotional
• eventTsExchange, eventTsBroker, eventTsIngestUtc
• correctionSeq, sourceSystem, rawPayloadHash

Design rule:

• append-only events + derived views
• never “fix in place” by mutating historical rows.

5) Break taxonomy that actually routes work

A break taxonomy is useful only if it routes to the right owner instantly.

A. Economic breaks (Trading Ops owner)

• qty/price mismatch
• fee/tax mismatch
• net-notional mismatch

B. Allocation breaks (Portfolio Ops owner)

• account/book mismatch
• strategy allocation sum != execution qty
• avg-price inconsistency across child allocations

C. Reference breaks (Data owner)

• symbol mapping drift
• stale instrument metadata
• wrong corporate-action factor usage

D. Settlement breaks (Middle-office owner)

• missing/invalid SSI
• failed affirmation/confirmation
• settle-date derivation mismatch

Each break must have:

• severity, owner, openedAt, slaDeadline, status, resolutionCode.

No unowned breaks. Ever.

6) SLOs for post-trade reliability (starter set)

Use small, auditable SLOs:

1. EOD unresolved economic break ratio < 0.10%
2. Median break resolution time < 20 minutes (intraday) / < 45 minutes (EOD window)
3. T+1 pre-open unresolved settlement breaks = 0
4. Snapshot reproducibility: re-run with same inputs must hash-identical output

If you cannot measure these, your process is opinion-driven.

7) Control dashboard (single-page, no vanity charts)

Top panel:

• unresolved breaks by severity
• unresolved notional by category
• oldest open break age

Middle panel:

• top concentrated sources (broker/venue/symbol/strategy)
• ingestion lag and parser error rates

Bottom panel:

• checklist status (EOD snapshot, sign-off, affirmation completeness)
• T+1 readiness traffic light (green/yellow/red)

This page should answer: “Can we sleep safely tonight?” in 10 seconds.

8) Practical runbook when break volume spikes

1. Stabilize input quality first
   • check ingestion lag, sequence gaps, parser/schema rollout
2. Cluster before firefighting
   • find highest-concentration bucket (same broker/venue/source)
3. Freeze nonessential changes
   • no new parameter rollouts during break storm
4. Apply reversible mitigations only
   • e.g., widen timing tolerance, but never relax side/symbol hard checks
5. Publish concise status
   • blast radius, current risk, next checkpoint ETA
6. Capture incident artifacts
   • payload samples, decision logs, final root-cause tag

If you skip step 6, the same incident repeats next month.

9) Weekly hygiene loop (keeps debt from compounding)

Every week:

• review top 10 recurring break patterns,
• delete dead matching rules,
• tighten one tolerance that caused false negatives,
• add one regression test from real incident payload,
• review one near-miss with both trading + ops in same room.

This is where operational alpha compounds.

10) 30-day implementation plan (lean team)

Week 1 — Canonical truth

• finalize event schema + normalization contracts
• deterministic matching only (broker+execId first)
• break table with owner/SLA fields

Week 2 — Visibility + discipline

• one-page control dashboard
• EOD immutable snapshot export + signed metadata
• on-call notification for SLA breaches

Week 3 — Settlement readiness

• affirmation/SSI status ingestion
• T+1 pre-open checklist automation
• incident template + postmortem format

Week 4 — Hardening

• replay tests from incident payloads
• reproducibility hash checks on snapshots
• game-day: forced feed lag + correction chain disorder

If game-day fails, you are not production-ready under T+1.

11) Common anti-patterns

• “We’ll reconcile tomorrow morning.”
• Spreadsheet-only break ownership.
• Silent tolerance changes without versioning.
• One giant “other” break category.
• Mutable historical ledgers.
• Treating settlement status as external-only responsibility.

All six eventually surface as PnL noise or compliance heat.

Bottom line

A small quant desk does not need a giant post-trade platform. It needs a boring, strict control loop that closes truth daily:

• canonical ledger,
• deterministic reconciliation,
• explicit break ownership,
• immutable EOD evidence,
• T+1 pre-open readiness gate.

That is how you convert post-trade from operational drag into decision confidence.