 |
|
PULSE NEWSLETTER
|
|
KX Pulse
|
|
Temporal precision, always-on markets, the KX Trader Agent Blueprint, and the KX NYSE TAQ Benchmark.
|
|
| |
|
Hi ala,
In this issue, of KX Pulse we return to a question that sits at the foundation of many production-grade capital markets AI systems: Can the system tell the time?
In his latest blog our CEO, Ashok Reddy, likens it to the longitude problem.
For centuries, sailors could calculate latitude from the stars but had no reliable way to fix their longitude without a clock accurate enough to keep time for months at sea. A one-minute drift put a ship miles off course. In this piece, Ashok makes the case that temporal precision (knowing what happened, when it happened, and what information was available at the decision point) is now the foundational infrastructure requirement for capital markets AI.
We go on to cover what always-on markets are doing to the infrastructure layer. Conal Doyle from KX and Alexander Unterrainer of DefconQ discuss how extended trading hours, rising odd-lot volumes, and 24/7 data flows are challenging architectures built for defined sessions and overnight recovery
windows.
Laura Kerr then introduces the KX Trader Agent Blueprint, a reference architecture for AI-powered financial research. Using NVIDIA NeMo orchestration and KDB-X, it gives AI agents access to point-in-time data across price history, filings, news, and macro feeds.
We also share new work from our Head of Benchmarking, Ferenc Bodon Ph.D.: The KX NYSE TAQ Benchmark and benchmark.kx.com, an interactive dashboard for exploring those results across engines, hardware configurations, and query types.
Grab a coffee and enjoy! ☕
|
|
|
|
|
Solving the Longitude Problem in Capital Markets: Why AI Needs Temporal Precision
|
|
|
|
Our CEO Ashok Reddy explains why temporal precision is foundational for AI in capital markets, and lays out the five-layer temporal trust stack that makes it possible: capture, sequence, replay, reason, and audit.
| ▪ |
Sequence Defines Meaning: A price, trade, or filing only makes sense in its correct temporal context. If source sequence and arrival time are not preserved at ingestion, the original event order may be impossible to reconstruct reliably later. |
| ▪ |
Bi-Temporal Tracking: Capturing when a fact was true in the market and when it entered the system makes historical market state reproducible from the records available to the system, rather than inferred from the latest corrected dataset. |
| ▪ |
Auditability Completes the Stack: An AI decision that can't be traced to the data state that informed it can't be defended in production or with regulators. |
|
| Read the full blog → |
|
|
|
|
When Markets Never Sleep: Extended Hours, Odd Lots, and the Infrastructure Implications of Always-On Markets
|
|
|
|
Markets are moving closer to continuous operation, and traditional architectures weren't built for it. Conal Doyle from KX and Alexander Unterrainer of DefconQ look at what extended sessions, near-continuous weekday operation and growing overnight data flows mean for infrastructure teams, and what
needs to change.
| ▪ |
Session Boundaries Are Disappearing: Architectures built for defined trading windows, overnight recovery, and batch cycles face growing pressure as markets move toward continuous operation. |
| ▪ |
Odd Lots Compound the Challenge: Rising odd-lot activity and sub-penny pricing increase message volumes and add to the temporal complexity of the data. |
| ▪ |
Fragmented Stacks Get Harder to Scale: As real-time analytics, historical data, and AI workloads converge, separate systems become bottlenecks. |
|
| Watch on demand → |
|
|
|
|
The KX Trader Agent Blueprint: AI Research Powered by KDB-X and NVIDIA
|
|
|
|
Financial research is fragmented across price data, filings, news, fundamentals, and macro feeds, and that fragmentation degrades AI research quality. Laura Kerr introduces the KX Trader Agent Blueprint, a reference architecture using NVIDIA NeMo and KDB-X that gives AI agents access to traceable, point-in-time financial data across all of them.
| ▪ |
Multi-Agent Architecture Mirrors Analyst Workflows: Specialized agents handle different data types in parallel, each grounded in the same temporal data layer. |
| ▪ |
Point-in-Time Joins Prevent Look-Ahead: KDB-X as-of joins ensure agents retrieve data as it was known at the decision moment, not a later corrected version. |
| ▪ |
Every Output Retains Links to Source Data: Responses link to the data that supported them, which is what makes AI research defensible in regulated environments. |
|
| Read the full blog → |
|
|
|
|
The KX NYSE TAQ Benchmark
|
|
|
|
Our Head of Benchmarking, Ferenc Bodon Ph.D., introduces an open-source suite of 84 queries on real NYSE trade-and-quote data — including the as-of join that no other benchmark covers — alongside benchmark.kx.com, an interactive dashboard for exploring results across engines, hardware configurations, thread counts,
and query types.
| ▪ |
Real Data, Production Scale: Over two billion quote rows from a single trading day, sized to match what capital markets teams handle in production. |
| ▪ |
Every Result Is Independently Reproducible: The benchmark is open-source and runs on named hardware configurations. Every number can be verified — no controlled environment, no withheld methodology, no vendor-run tests. |
| ▪ |
Results You Can Interrogate: benchmark.kx.com lets you filter by engine, hardware, threads, and query type, with every comparison shareable as a plain URL. |
|
Read the benchmark blog →
Read the dashboard blog →
Explore benchmark.kx.com → |
| |
|
|
Latest from KX
|
|
Product Updates
Tick-X vs. the Scalable Ingestion Blueprint: Two Ways to Scale kdb+ Tick Ingestion
Two approaches to scaling kdb+ Tick ingestion through workload separation, sharding, and real-time data scaling.
The KX Trader Agent Blueprint: AI Research Powered by KDB-X and NVIDIA
Reference architecture for multi-agent AI financial research.
KDB.AI — Inside KDB-X: Vector Search Where Your Data Already Lives
Vector search and time-series analytics in one engine.
Create KX Dashboards Using Natural Language
KX Dashboards 2.22 introduces AI Builder: Create and edit dashboards from plain-language prompts.
|
|
Webinars
US Book Depth: Market Microstructure Analysis with OneTick Cloud
Level 3 order book analysis in practice, from individual order messages to consolidated book depth across US equity venues.
Rewiring the Front Office: Cloud, Data, and AI in Action
Hosted by Waters Technology. How cloud infrastructure, data architecture, and AI are changing front-office technology in capital markets.
Adapting to Market Data Changes: US SIP, Odd Lots, and 24/7 Trading
What the latest SIP reforms mean for market data teams, and how OneTick Cloud has adapted.
Upcoming: OneTick Live Session
Register for our next live OneTick session.
|
|
ebooks
The Case for Unified Analytics with KDB-X
Why fragmented analytics stacks are costing capital markets firms their edge.
8 Reasons Quants Choose KDB-X
Faster research, fewer rewrites, one environment for time-series, vector, and AI workflows.
Stop the Data Tax with Managed Market Data
How OneTick Market Data reduces the data preparation overhead before research can begin.
|
|
Blogs
What is the Best Time-Series Platform for Quantitative Back-Testing?
OneTick vs. cloud data warehouses vs. legacy databases, ranked on ingest speed, query latency, symbology handling, and Python workflows.
Rewiring the Front Office: 5 Key Takeaways on Cloud, AI & Data Quality
Five observations from the buy-side and sell-side on cloud migration, AI data quality, and front-office modernization.
The Market Data Evolution: Odd Lots, Fractional Shares & the Future of the US SIP
A breakdown of the US SIP reform timeline and the data volume explosion that followed.
See You at XLoD Global
OneTick's trade surveillance capabilities, with a look at what the team brought to XLoD Global in London.
|
| |
|
| |
|
Join Our Community
Applications are still open for our Developer Advocacy Program, 'Community KXperts'! This program is ideal for anyone passionate about sharing their knowledge on KX through blogs, articles, or other content. To apply, contact evangelism@kx.com.
|
| |
|
|
|
| KDB-X | Why KX? | Resources | Developers |
|
|
This email was sent to ala@niepodam.pl by: KX, 45 Broadway, New York, NY 10006, USA
Manage Preferences or Unsubscribe | Privacy Notice |
|
