Data quality · Exchange-traded funds
ETF quotes, held against the tape.
Listed funds get the same test as US stocks: every quote is compared with the national best bid and offer, one sample per second.
XLE against the NBBO
One sample per second across the run.
Headline measurements
Measured, not asserted.
14k matched samples across 12 names, one per second, regular session.
Within 1 bp of the NBBO midpoint
91.5%
Median deviation 0.13 bp; within 2 bp in 99.2% of samples.
How is this measured?
- Definition
- |TickerLayer mid − reference mid| / reference mid × 10,000 ≤ 1, share of samples per symbol, then averaged across symbols.
- Window
- 20 minutes, sampled every 1 s
- Universe
- etfs-core-12 v1, 12 of 12 symbols
- Samples
- 14k matched samples
- Method
- etfs-quality-v1
Quote ranges overlap
90.1%
TickerLayer bid to ask meets the reference bid to ask. Midpoint inside the NBBO in 73.8%; penny-wide books flip several times a second.
How is this measured?
- Overlap
- TickerLayer bid ≤ reference ask and TickerLayer ask ≥ reference bid, per one-second sample.
- Inside
- reference bid ≤ TickerLayer mid ≤ reference ask. A one-second sample against a book that moves many times a second lands outside more often on penny-spread names.
- Window
- 20 minutes, sampled every 1 s
- Universe
- etfs-core-12 v1, 12 of 12 symbols
- Samples
- 14k matched samples
- Method
- etfs-quality-v1
Source to platform p95
363ms
Reference event to the matching TickerLayer state; p50 67 ms.
How is this measured?
- Definition
- When the reference quote changes, the first TickerLayer frame carrying the same quote; the difference of the two timestamps.
- Window
- 20 minutes, sampled every 1 s
- Universe
- etfs-core-12 v1, 12 of 12 symbols
- Samples
- 14k matched samples
- Method
- etfs-quality-v1
REST and WebSocket agree
99.2%
Same quote on both interfaces across 119 checks.
How is this measured?
- Definition
- A REST quote compared with WebSocket frames received within ±1.5 s of the request; exact when bid and ask are identical to one of them.
- Cadence
- One check every ten seconds, symbols rotating through the universe.
- Window
- 20 minutes, sampled every 1 s
- Universe
- etfs-core-12 v1, 12 of 12 symbols
- Samples
- 14k matched samples
- Method
- etfs-quality-v1
Distribution
Where the deviation actually lands.
Every matched sample by absolute deviation. Threshold shares are symbol balanced.
Distribution of absolute deviation
Share of all matched samples in basis points.
- under 0.5 bp82.3%
- 0.5 to 1 bp9.2%
- 1 to 2 bp7.6%
- 2 to 5 bp0.8%
- over 5 bp0.1%
Threshold alignment
Symbol balanced share of samples within each band.
- Within 1 bp91.5%
- Within 2 bp99.2%
- Within 5 bp100.0%
- p99 absolute deviation1.91 bp
State for state
Our feed against the NBBO.
Not the same feed, and this close to it.
Inside the NBBO
73.8%
TickerLayer midpoint within the sampled NBBO spread. A quote comparison, not execution quality.
Quote ranges overlap
90.1%
TickerLayer bid to ask range meets the reference range.
Within 2 bp
99.2%
Midpoint within two basis points of the reference, symbol balanced.
Source-to-platform freshness on exact matches: p50 67 ms, p95 363 ms, p99 4.11 s over 323 reference changes. Inside the NBBO measures how often the TickerLayer midpoint falls within the sampled NBBO spread. This is a quote comparison, not a measure of execution quality.
Universe detail
Every symbol in the universe.
One row per name in the benchmark universe, with the figures a customer would compute from the same feed.
| Symbol | Samples | Median dev bp | p95 bp | Within 1 bp | Within 2 bp | Inside NBBO | Spread bp |
|---|---|---|---|---|---|---|---|
| SPY | 1,185 | 0.13 | 0.73 | 98.3% | 100.0% | 48.5% | 0.26 |
| QQQ | 1,185 | 0.28 | 1.34 | 90.2% | 98.9% | 43.8% | 0.42 |
| IWM | 1,185 | 0.17 | 1.21 | 90.3% | 99.8% | 58.7% | 0.35 |
| DIA | 1,185 | 0.19 | 0.86 | 97.1% | 99.9% | 80.0% | 0.96 |
| VTI | 1,185 | 0.13 | 0.80 | 97.9% | 99.9% | 78.6% | 0.80 |
| VOO | 1,185 | 0.14 | 0.72 | 97.9% | 99.9% | 69.1% | 0.57 |
| GLD | 1,185 | 0.25 | 1.50 | 87.5% | 98.1% | 67.9% | 1.00 |
| TLT | 1,185 | 0.00 | 0.00 | 95.6% | 100.0% | 95.6% | 1.24 |
| XLF | 1,185 | 0.00 | 1.76 | 90.5% | 99.8% | 90.4% | 1.76 |
| XLE | 1,185 | 0.00 | 3.07 | 75.4% | 94.8% | 74.8% | 1.54 |
Freshness and consistency
Delivery, agreement, reproducibility.
Timing is kept apart from price fidelity on purpose.
Update cadence
64ms
Median gap between consecutive frames per symbol on the public stream.
REST vs WebSocket exact
99.2%
119 checks, midpoint gap p95 0.00 bp.
Settled bars reproducible
100.0%
68 one-minute bars re-fetched later; close unchanged in 100.0%.
Reference model
Consolidated US market reference
Each fund is compared with the national best bid and offer from a consolidated US market reference, using the same inside-NBBO methodology as US stocks.
Universe
Twelve US listed funds: broad equity, bonds, gold, sectors.
Reference
National best bid and offer from a consolidated US market reference.
Session
Session stated with every run; pre-market is never mixed with regular hours.
Benchmark run
One run behind every figure.
We keep measuring and publish each run as it lands.
- Benchmark run
- TLQ-ETFS-20260910-1345
- Methodology
- etfs-quality-v1
- Universe
- etfs-core-12 v1, 12 of 12 measured
- Observation window
- Sep 10, 2026, 13:45 to 14:06 UTC, regular session
- Window length
- 20 min, sampled every 1 s
- Matched samples
- 14k
- Published
- Sep 10, 2026, 14:06 UTC
Benchmark results describe observed behaviour under the stated methodology during the published window. They are not a guarantee of future performance and not a service level agreement unless separately agreed in writing. These comparisons evaluate TickerLayer quotes against selected market references during the stated observation window. They do not imply exchange affiliation, endorsement, or certification.
Frequently asked questions
Is an ETF quote compared with its index?
No. A listed fund is compared with its own consolidated quote. The gap between a fund and its index is a premium or discount, not a data quality signal.
Why is within 1 bp the headline rather than inside-NBBO?
Broad funds trade with one-cent spreads that flip many times a second; a one-second sample can land a cent outside while the midpoint stays within a basis point.
Which funds are in the universe?
Twelve US listed funds across broad equity, fixed income, gold and sectors, fixed and versioned.
Does the session matter?
Yes. Pre-market quotes are wider and sparser; each run states its session and the two are never mixed.
How fresh is an ETF quote on the platform?
Source-to-platform lag is measured on exact matches and published as p50 and p95 for the run.
Are the numbers a guarantee?
No. They describe observed behaviour during the published run under the stated methodology.