top of page

Background

VelocityTrade Capital is a mid-size algorithmic trading firm running systematic strategies across 240 actively monitored symbols on NYSE and NASDAQ, with automated execution driving a $1.8 billion trading book. The firm's trading signals depended on market data refreshed on a 15 to 45 minute batch cycle, a latency gap that was quietly eroding returns and leaving the firm unable to reconstruct its market data state for regulatory review. This case study documents the design and delivery of a governed, real-time market data pipeline built on Microsoft Fabric's Real-Time Intelligence workload, using Eventstream for continuous tick ingestion, a KQL Database for live storage and querying and a Real-Time Intelligence dashboard for monitoring.

Problem

  • Market data refreshed on a 15 to 45 minute batch cycle, causing trading signals to act on prices that no longer reflected the live market.

  • Estimated $2.1M to $3.4M in quarterly return leakage from entry and exit timing errors across roughly 4,200 monthly automated executions.

  • No event-level audit trail — each batch refresh overwrote the prior state, leaving no way to reconstruct market data at the exact moment a trade decision was made.

  • Intra-period price volatility discarded entirely, so sharp moves within a batch window went undetected by risk models designed to catch them.

  • No streaming enrichment layer, pushing derived-field computation into the query layer against stale snapshots instead of live data.

  • No event-driven alerting — price breaches relied on manual dashboard monitoring, most likely to fail during the high-activity periods that mattered most

Solution

Eventstream Pipeline

Set up an Eventstream pipeline that pulls in live stock tick data (date, open, high, low, close, volume, ticker), renames key fields for consistency, and streams it straight into the Eventhouse for storage and analysis.

KQL Queries

Wrote the first KQL query (Q1) to find the most recent closing price for each stock ticker, giving the trading desk a quick, always-current price reference to work from.

Wrote the second query (Q2) to calculate each ticker's average intra-day price range (High minus Low) and ranked them highest to lowest, giving the desk a quick view of which stocks are moving the most within a single day.

Wrote the third query (Q3) to calculate each ticker's average daily return (Close minus Open) and ranked them from highest to lowest, showing which stocks are trending upward and which are trending downward.

Wrote the fourth query (Q4) to compare each ticker's average trading volume against its highest recorded volume, flagging tickers where volume spiked nearly double the norm as possible signs of unusual activity worth a closer look.

Wrote the fifth query (Q5) to measure how far each ticker's latest closing price has fallen from its peak price, ranking the biggest drawdowns first to spot stocks that may be in a pullback or reversal.

Wrote the sixth query (Q6) to catch bad data before it reaches the trading desk, flagging any record with a negative or zero volume, negative prices, or a High price that's lower than the Low price.

Wrote the seventh query (Q7) to check how long it's been since each ticker last received a new data point, helping confirm the pipeline is actively streaming live data for every monitored stock rather than silently going stale.

Real-Time Intelligence Dashboard

Built a live Real-Time Intelligence dashboard bringing together three tiles — Latest Close, Intra-Day Range, and Volume Surge — so the trading desk can monitor current prices, volatility, and unusual activity across all tickers at a glance.

Outcome

  • Replaced a 15 to 45 minute batch refresh cycle with a live streaming pipeline, giving trading signals access to current, not stale, market prices.

  • Directly addresses the $2.1M to $3.4M in estimated quarterly return leakage caused by timing errors from outdated data

  • Adds a governed data quality layer (Query 6) that filters out negative or zero prices, bad volume figures, and inconsistent High/Low values before they reach any trading signal.

  • Introduces a pipeline freshness check (Query 7) so the desk can immediately spot if a ticker's data feed has gone stale or stopped updating.

  • Gives the trading desk a single live dashboard covering current prices, intra-day volatility, and volume surges, replacing scattered manual checks with one governed source of truth.

bottom of page