Research Library
Twenty articles on portfolio research, volatility, session structure, data integrity, AI agents and knowledge engineering from NOVAQ.
Featured
- 01
How Many Bets Does a Portfolio Really Hold?
A portfolio of 100 strategies sounds safer than a portfolio of 10. It is safer only if those strategies do different things. If twenty of them win and lose on the same days, they are one position with twenty names, and…
- 02
Forecasting the Size of the Day
Most market forecasting asks where price will go. It is the hardest question in finance and the one with the least to show for it. NOVAQ's research suggests a more answerable question with direct practical value: how…
- 03
What 16 Years of Session Data Show
A session framework describes how a trading day unfolds: where the opening range forms, whether price accepts or rejects it, and what that implies for the rest of the day. Traders have used such frameworks for decades.…
- 04
An Eight-Role AI Audit Swarm
An AI agent writing code is a persuasive reviewer of its own work. It can explain why a change is correct with great fluency, and a fluent explanation is not evidence. As AI agents take on more engineering and research,…
- 05
Making a Live System Six Times Faster Without Changing a Decision
Performance work on a trading system has an unusual rule. A faster system that decides differently is a different system, and all the research behind it no longer applies. So the goal in October 2026 was narrow: reduce…
Technology
- 06
One Code, the Same Trades
Every quantitative result sits on a stack of data decisions that readers never see. When a result changes, the first question is whether the code changed or the data did. Without an answer, research turns into argument.…
- 07
Migrating a Running Trading System
By late June 2026 the first generation of NOVAQ had reached its ceiling. One process and one analytical database served everything, and a heavy calculation could stop the whole application for over a minute.
- 08
From a C# Core to Study Parity
Some strategies live on a timescale that a polling engine cannot serve. NOVAQ's answer was to run them natively inside the trading platform, from a C# core working on one-second data.
AI
- 09
Checking What an AI Analyst Is Told
A well-written market brief can be wrong in a way that is hard to see. The prose reads well, the reasoning sounds sound, and one input is quietly incorrect. For an AI system that reports on markets, the usual quality…
- 10
Building a Runtime Expert From a Knowledge Base
Most AI assistants know what their model happened to learn, and nobody can say exactly what that is. For a system that interprets markets, this is a real problem: when its view changes, you want to know whether the…
- 11
Testing an Autonomous AI Trader
Autonomous AI trading is easy to demonstrate and hard to evaluate. NOVAQ designed a test that could fail.
Knowledge
- 12
From Library Claim to Pre-Registered Test
A knowledge base earns its keep when it changes what gets tested. Many research teams keep literature in one place and experiments in another. In NOVAQ they are connected, and a note can become a test with its…
- 13
From 198 Notes to 145,000
Trading knowledge is abundant and badly organised. Books, courses, videos and papers repeat, contradict and quietly copy each other, and almost none of it can be cited at the level of a single claim. NOVAQ set out to…
Research case studies
- 14
Reproducing the Market's Best-Known Numbers
Reproduction is one of the most valuable things a research group can do, and NOVAQ has done a lot of it.
- 15
How a Portfolio Held Up Under a Point-in-Time Review
Look-ahead is the classic quantitative error: a value calculated at the end of a session gets copied onto its earlier bars, and a model learns from a future it will never see.
- 16
What Changed in the Market in 2026
Between 4 August and 14 September 2026 the live strategy book lost 246.5 signal R, or −89.9 R as independent bets. When that happens, a team needs to know whether the market changed or the system broke. NOVAQ's…
- 17
How Much, Not Where
Dealer-positioning data is often read as a map of where price will go. NOVAQ and Vašek tested that from several sides, on a year of positioning data, commercial level products and hundreds of gigabytes of joint research…
- 18
What the First Hour Can Tell You
How much can you know about a day shortly after the open? For a discretionary trader this is the daily question. Open types, gaps and early structure are central to auction market theory, and they have rarely been…
- 19
A Chart 34 Ticks Too Kind
Trading charts carry a subtle bias: markers are drawn after the fact, at prices that were not available when the signal formed. An order-flow indicator looked excellent on the chart, and a trader's eye saw a hit rate of…
- 20
Reproducing a 25,000-Configuration Study
Public backtests have become much more ambitious since AI tools made large parameter sweeps cheap. Reproducing them is a good way to learn what they really show. In August 2026 the channel AI Pathways published Claude…