The bond consolidated tape reaches production before the equities one, which means fixed-income desks are the first to have to actually consume a European tape rather than read about it. The plumbing is due before the policy debate has fully settled.
Most of what is written about the EU consolidated tape treats it as a market-structure story: who won the mandate, whether the incumbents are unhappy, what it does to best execution. That is the policy layer. Underneath it is an engineering job that lands on a specific timetable, and for bonds that timetable is ahead of everything else. ESMA selected Ediphy, trading as fairCT, as the first consolidated tape provider for bonds on 3 July 2025, and only selected EuroCTP for shares and ETFs on 19 December 2025, with the derivatives selection opening in January 2026. Bonds are the pathfinder. If your firm trades or holds fixed income, you are in the first cohort that has to build for a live tape.
Why the bond tape is the harder one to go first
Equities are, from a data engineering point of view, the easy case. Instruments are standardised, liquid names trade continuously, and a print is a print. Bonds are the opposite. A single issuer has dozens of lines, many barely trade, identifiers proliferate, and reference data — coupon, maturity, seniority, issuer hierarchy — carries as much weight as the trade itself. A post-trade record without the reference context around it tells you very little. So the market that is least standardised is the one being consolidated first, and the firms that treat the bond tape as a lightweight feed to bolt on will find it is neither light nor a simple bolt-on.
Free · 4 minutes
If your most senior engineer left tomorrow, would anyone still understand the system?
Fourteen questions on documentation, dependencies, and the gap between how the architecture works and how many people know it. Banded finding on screen, full sheet by email.
The tape for bonds is a post-trade product. It consolidates transaction reports — price, volume, venue, timestamp, instrument — from trading venues and approved publication arrangements into a single stream, alongside the reference data needed to make each print interpretable. It is not a pre-trade order book. What you are building for is a consolidated record of what traded, when, and on what terms, assembled from many contributors under one clock.
The consumer side: three feeds, not one
Firms tend to think of “connecting to the tape” as a single integration. In practice the consumer side is three streams that have to be reconciled against each other.
- The post-trade transaction feed. The core product: individual prints as they are published, subject to deferral. This is the stream your valuation, TCA and surveillance systems ultimately want.
- The reference data feed. The instrument master that lets you attach each print to the right line, issuer and classification. Stale or missing reference data silently corrupts everything downstream, and bonds churn their reference data far faster than equities.
- Corrections and cancellations. Prints get amended and pulled. A pipeline that ingests the happy path but ignores the amendment stream will drift out of line with the official record within days.
The architectural decision that matters most here is whether you treat the tape as a system of record or a system of reference. For most firms it is the latter: an authoritative external view you reconcile your own trade and position data against, not a replacement for either. Build it as a reconciliation target with a clear lineage back to the contributor, and you keep the tape honest without letting it quietly overwrite your books. This is the same data-plumbing question the EuroCTP selection raised for the equities side, arriving earlier and against a messier instrument universe.
Deferral flags are the part that breaks naive code
The single thing most likely to produce wrong numbers is mishandling deferrals. Under the MiFIR Review — Regulation (EU) 2024/791, amending the original MiFIR — bond post-trade publication moves to a revised regime built around near-real-time publication with a structured set of deferrals. The default is prompt publication, on the order of fifteen minutes for the relevant trades, with longer deferrals applied by category according to instrument liquidity and trade size. Crucially, price and volume are deferred separately: a large trade can have its price published while the true size is withheld and released later, in stages, up to a matter of weeks.
That decoupling is where naive consumers go wrong. A print arrives with a price and a placeholder or capped volume; a system that reads it as a completed trade will feed a distorted size into TCA, liquidity metrics or a risk model, and then never revisit it when the full detail is published later. The feed handler has to know that a record can be provisional, has to carry the expected completion time, and has to reconcile the later full print against the earlier partial one rather than treating them as two trades. The publication-side logic for deferrals under the MiFIR Review is the mirror image of this problem; consumers inherit it whether or not they publish.
A minimal handler makes the provisional state explicit rather than collapsing it:
from dataclasses import dataclass
from datetime import datetime, timedelta
# Deferral durations are fixed by the RTS under the MiFIR Review
# (Regulation (EU) 2024/791). The bands below are illustrative
# placeholders keyed by category - replace them with the categories
# and durations in the final RTS before you go live, and treat the
# category itself as coming from the tape's published flag set.
PRICE_DEFERRAL = {
1: timedelta(minutes=15),
2: timedelta(hours=8), # end of trading day, resolved by session
3: timedelta(days=1),
4: timedelta(days=14),
5: timedelta(days=28),
}
@dataclass
class TapeRecord:
isin: str
price: float | None # None while a price deferral is in force
quantity: float | None # None while a volume deferral is in force
traded_at: datetime
deferral_category: int
def is_complete(rec):
"""Complete once neither price nor volume is being withheld."""
return rec.price is not None and rec.quantity is not None
def expected_completion(rec):
"""When the full trade detail should reach the tape."""
return rec.traded_at + PRICE_DEFERRAL.get(rec.deferral_category, timedelta(0))
def handle(rec, book):
"""Fold a record into a local view, never treating a deferred
print as final."""
if is_complete(rec):
book.upsert(rec.isin, rec.price, rec.quantity, provisional=False)
return
# Partial print: publish what is known, but mark it provisional so
# valuation and TCA do not consume a deferred trade as a settled one.
book.upsert(
rec.isin,
price=rec.price,
quantity=rec.quantity,
provisional=True,
resolves_at=expected_completion(rec),
)
The point is not the exact durations, which you must take from the final technical standards rather than from this table. The point is the shape: a record has a completion state, a provisional print is not a trade, and the later full detail has to be joined back to the earlier one. Get that structure right and the numbers follow when the RTS bands are confirmed.
The contributor side is a data-quality obligation
If your firm operates a venue or an approved publication arrangement, you are on the other side of this: a contributor to the tape, held to input standards on completeness, timeliness and format. The consolidated tape is only as good as what feeds it, and the regime puts the burden of clean, timely, correctly flagged input on the contributors rather than on the tape provider to clean up after them. That means your outbound publication has to apply the right deferral flags, carry accurate reference identifiers, and meet latency expectations measured against a common clock — not your own.
There is a classification discipline buried in this that firms underestimate. Deciding which deferral applies to a given trade, and flagging it correctly, is a data-classification problem before it is a transmission problem — the same category of work as deciding what may and may not be shared under the Data Act’s trade-secret carve-out: the value is in getting the classification right at source, because everything downstream trusts the flag. A wrongly flagged trade either leaks size that should have been protected or withholds data that should have been public, and both are supervisory problems.
Treat the bond build as the rehearsal it is
The firms that will struggle are the ones that defer. The bond tape and the equities tape are close enough in shape that the ingestion framework, the reconciliation model, the deferral-aware handling and the reference-data joins are largely reusable. Build them properly for bonds, against the harder instrument universe and the earlier deadline, and the equities tape becomes a configuration exercise. Defer the bond work and you buy yourself two rushed builds instead of one deliberate one — the second under more time pressure than the first.
The tape is arriving as infrastructure, not as a consultation. The question a supervisor, or your own risk committee, will eventually ask is not whether you support the consolidated tape but whether the numbers you derive from it are right — and that is decided entirely by whether you treated a deferred print as provisional the first time it arrived.
Free interactive tool
Website compliance checklist
What your site has to do, based on what it actually does
Answer as much or as little as you like — the list builds as you go. Nothing is stored against your name and no email is required.
Everything that applies
Ordered by what to do first: legal requirements you can close quickly, then larger pieces of work, then what is expected rather than required. Not exhaustive, and not a legal audit.
Dated PDF, yours to keep or circulate.
Can you trust the architecture you have?
Architecture diagrams rarely show the reality of how systems actually operate. An independent review establishes what is really there.