Introducing Deep Learning for Spatio-Temporal Modeling: Dynamic Traffic Flows and High Frequency Trading for after-work quants

Traffic jams and order books both change in space and time. Can one deep-learning habit help you think about both—without pretending markets are highways?

Deep Learning for Spatio-Temporal Modeling: Dynamic Traffic Flows and High Frequency Trading (Dixon, Polson, Sokolov, 2018) presents deep models for spatio-temporal data and discusses applications that include dynamic traffic flows and high-frequency trading.

Fit for learners: Open-PDF classic/working paper on microstructure — useful for after-work Python quant learners building intuition before coding.

This Hub note is a plain-language map. Not investment advice.

Core ideas in plain language

• Spatio-temporal data: what happens here depends on what happened nearby and recently.
• Deep nets can stack local spatial filters with temporal memory—useful when raw tables are huge and nonlinear.
• The paper’s dual setting (traffic + HFT) is pedagogical: same modeling language, different physical stories.
• For market microstructure, treat HFT sections as intuition builders, not as a plug-and-play alpha recipe.

You are not asked to train a production HFT net tonight. You are asked to recognise why “space × time” shows up in modern microstructure ML.

Why after-work Python learners should care

If you only ever plot one ticker’s midprice, you miss cross-sectional structure (related names, venues, levels). This paper is an open-PDF door into that vocabulary—while still tying to skills you already practice in evening notebooks: tensors, sequences, and careful evaluation.

Healthy mini-project: take a tiny public LOB or trade print CSV, build lagged features across a few levels or symbols, and stop at plots—no live orders.

One-evening reading checklist

1. Read the motivation: why traffic and trading share a spatio-temporal framing.
2. Skim the deep architecture section for what “spatial” vs “temporal” means in their setup.
3. Note evaluation choices (what they optimise / report).
4. Write three caveats you would paste above any personal HFT experiment.

Download the PDF

Download PDF

Landing: arXiv:1705.09851

Important disclaimer

Educational only. Not investment advice. Research models of high-frequency data are stylised. They do not tell you what to trade.

What this is not

• Not CEF content
• Not Quant/Macro course CTA
• Not an HFT vendor pitch

Next step

Need cleaner Python before deep models feel friendly? Use Introduction to Python Programming only—the single course link on this page.