Statistics & ValidationStatistics and mathematics students

Statistics & Validation

This stream is for students who like breaking results. The job is to find out whether a signal still looks real after costs, new samples, different regimes and less flattering assumptions.

Apply when intake opens See open roles

Example work

A few examples of what this stream could pick up inside SSI Labs.

01
Rerun a strategy with wider costs, delayed execution and stricter fill assumptions.
02
Split results by volatility, trend, liquidity or funding regime and explain what changes.
03
Check whether a tuned parameter works because of the market or because it was fitted to one sample.

Work

What this stream owns

Separate in-sample and out-of-sample results.
Design the checks a backtest has to pass before anyone trusts it.
Look for look-ahead bias, survivorship bias, data leakage and overfitting.
Summarise drawdown, turnover, exposure, volatility and transaction-cost sensitivity.
Call out when a result breaks instead of dressing it up.

Preparation

What to build toward

Probability, inference, regression, hypothesis testing and time-series basics.
Python for analysis, especially pandas, NumPy, plotting and reproducible notebooks.
Understanding of backtest traps: look-ahead, survivorship, multiple testing and regime dependence.
Clear communication: what changed, why it matters and whether the original claim survives.

Fit

Who should look here

You like finding the assumption that makes a result fall apart.
You would rather be accurate than impressive.
You like turning messy output into a clear yes, no or not-yet answer.

Reading