Quantitative trading workstation with live data feeds

Machine learning model training for financial signals

Workshop participants studying algorithmic trading strategies

About Turbaselon

Where quantitative thinking becomes a practised skill

Turbaselon was built in 2023 around a straightforward frustration: most education on machine learning in trading either stays theoretical or skips to code without explaining the financial logic. Neither approach sticks.

We run structured workshops that connect statistical methods, Python tooling, and real market data — so participants build strategies they can actually interrogate and refine.

8+ Workshop modules

4 provinces Participants reached

Live data Used in every session

How we work

Each workshop is built around a specific decision a trader actually faces

Signal generation, portfolio weighting, drawdown control — participants work through each problem with real equity and futures data, not toy datasets. The goal is not familiarity; it is functional understanding.

Sessions run online, are capped deliberately, and include asynchronous review so participants from Calgary to Halifax can engage on their own schedule without losing the collaborative element.

Practical construction

Every assignment produces a working artefact — a backtest, a feature pipeline, a risk report. Participants leave with files they wrote, not slides they photographed.

Honest scope

We do not promise trading profits. We teach the mechanics of building and evaluating quantitative models — what they can detect, where they fail, and how to tell the difference.

Iterative feedback

Instructors review submitted work and return written notes. Participants revise. That cycle — build, review, rebuild — is the core of how skills actually form.

Ingrid Voss

Lead Instructor & Curriculum Designer

"Most participants arrive knowing either finance or Python — almost never both fluently. The workshop exists in that gap. We spend the first two sessions just on why a signal that looks good on a chart can still be statistically useless."

14 Hours per cohort Across live instruction, async review, and peer critique — structured tightly enough that the time commitment is predictable.

6 Max participants Small groups mean submitted work gets genuine attention. Feedback is specific to the participant's own model, not generic notes.

ML Core discipline Gradient boosting, LSTM architectures, walk-forward validation — applied to equity signals, not academic benchmarks.

Collaborative session debugging a backtesting pipeline