Skip to main content

Nowcasting

Leverage data to capture a better perspective of reality.

Be process-driven, targeted and differentiated in your research while embracing intellectual honesty and curiosity. Mastering the intersection of multiple disciplines and filtering signal from noise is key to generating unique, high-conviction insights.

Model​

Nowcasting is a reusable pattern: build real-time gauges from primary data, then read the present before the official print confirms it. The principle is that behaviour leaves traces now; you do not have to wait for the lagging report. Apply it whenever a slow indicator hides a decision you must make today — when to use it is exactly the moment narrative and reality might diverge.

Systems​

  • Source data directly from online retailers across various geographies to track prices in near real-time, ahead of official inflation prints.
  • Monitor daily electricity consumption as a gauge of real-time economic activity. Spikes in electricity usage can indicate something is happening in the economy even if the narrative suggests otherwise.
  • Develop automated nowcasting tools to track as many relevant time series as possible in real time, to pinpoint anomalies and strange moves in markets as they happen.

Discipline​

  • Admit when you are wrong to maintain transparency and trust with your customers / community.
  • Focus research on areas outside the US, such as Europe, China, Japan, to find tradable trends that are less-covered.
  • Develop a structured research process, but be willing to explore new topics by regularly meeting with experts outside your domain.
  • Keep a running list of research ideas and observations, connecting dots to generate insights. Writing is a powerful tool to structure thinking.
  • Simplify the expression of big macro views into single, high conviction trades rather than trading everything.
  • Read widely outside of finance, in areas like technology, science, history, to draw insights and develop a latticework of mental models.
  • Maintain a carefully curated Twitter feed to track sentiment, narratives and potential gaps in your own views.
  • Ruthlessly prioritize what data, assets and markets to monitor - only follow what you intend to trade based on your research process and edge.

Failure Modes​

Nowcasting fails in specific ways:

  • Manufacturing confidence — mistaking more signals for more certainty. A good nowcast narrows uncertainty; the anti-pattern is a dashboard that looks decisive without earning it.
  • Tracking noise — following a series with no causal link to what you trade. Every gauge you cannot act on is a distraction and a risk.
  • Updating the conclusion, not the model — being wrong and patching the answer while the flawed process survives to fail again.

Changes my mind: If real-time proxies systematically diverge from what official prints later confirm, the nowcast is measuring noise and the edge is illusory.

Next question: Which single signal in your current nowcast is most likely capturing noise rather than reality?

Context​

  • Pipeline Nowcast PRD — The platform implementation: composite variance + prediction validation
  • Matrix Thinking — The structural tool for identifying which signals to track
  • Belief System — Nowcasting is how Belief stays grounded in reality rather than narrative
  • Macroeconomics — The macro signals that feed the nowcast

Questions​

If your nowcast contradicts the narrative — whose job is it to explain the gap?

  • Which of your current signals is most likely to be capturing noise rather than reality?
  • When you admit you were wrong, what process do you use to update the model rather than the conclusion alone?
  • What's the difference between a nowcast that narrows uncertainty and one that manufactures confidence?