HackStack

Project Info

Winner

Elastic: Best use of the Elastic Agent Builder on a Serverless instance

MarketMind

Devpost

Inspiration

Modern (novice) traders are flooded with fragmented, fast-moving information (price action, fundamentals, news/sentiment, and options/volatility) and lack a single, interpretable surface that converts these streams into trustworthy, real-time signals. Making fast, data-driven trading decisions becomes guesswork rather than insight. We wanted to bring order to that chaos by giving traders a clear and interpretable view of market energy in one place.

What it does

MarketMind turns raw market data into simple, real-time visual signals. Six agents, each built with Elastic Agent Builder, analyze a different layer of the market: price movement, company fundamentals, news sentiment, volatility, etc. Their combined outputs appear as glowing pulses around each stock, showing whether momentum is building, risk is rising, or sentiment is shifting.

How we built it

We built 6 agents connected to Elasticsearch, each reading from multiple indices and returning standardized JSON results. Every agent computes its own metrics, scoring, and reasoning before sending its output back to the MarketMind client. The frontend was built with React 19 and Next.js 16, using TypeScript, SVG physics-based visuals, and Chart.js for live candlestick charts. Yahoo Finance provided real-time data, while the app’s animation engine created the smooth pulsing and orbiting interactions that make the experience feel alive.

Challenges we ran into

ESQL was powerful but strict, and small syntax differences broke queries. Aligning timestamps across datasets and keeping data fresh without overwhelming performance required multiple iterations. The hardest part was keeping every agent’s logic explainable while still running fast enough for real-time feedback.

Accomplishments we're proud of

We successfully integrated real trading analysis techniques inside a fully explainable multi-agent system. The Oracle Network computes Technical Analysis indicators such as RSI, Bollinger Bands, and MACD to assess short-term price momentum. The Arbitrage Hunter applies Fundamental Analysis metrics including P/E Ratio and Beta to detect valuation imbalances. The Volatility Prophet uses Portfolio Optimization with Risk-Adjusted Allocation concepts like Modern Portfolio Theory and the Black-Litterman Model to contextualize volatility regimes. Bringing all of these together in one live system that is both interpretable and reactive is a major achievement.

What we learned

We learned that modularity and explainability are essential for both trust and performance. By designing each agent as an independent analytical node with clear responsibilities, we reduced complexity while improving interpretability. We also discovered how visual design can make complex market models not only accessible but intuitive, helping users “see” patterns they might otherwise miss.

What's next

Our next goal is to use realtime data sources to injest into Elasticsearch. This will provide better data to the agents which will result in better responses. We also aim to implement a user chat interface. This user agent will communicate with the Elastic agents using the A2A protocol to provide a response.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • Next.jsIn code
  • ReactIn code
  • TypeScriptIn code

4 of 4 appear in the indexed code.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

237 KB

Source files

42

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars