The Test Bench

Where market ideas get tested.

Markets are full of rules and claims. The Test Bench turns them into questions you can measure.

We write the question, the rules and the comparison first, run the test on past data, and publish the result either way.

Some studies test trading rules. Others look at market behaviour, risk, costs or how a strategy actually behaves. The right comparison depends on the question — sometimes Buy & Hold, sometimes something else.

What every Test Bench study includes

• A clear question. What exactly are we trying to find out, stated before the test is run.
• A stated method. The data, timeframe, definitions, assumptions and method are shown so the test can be understood and challenged.
• A relevant comparison. Results are measured against a baseline appropriate to the question.
• The wider evidence. Show the broader set of results rather than selecting only favourable examples.
• Risk, costs and exposure where relevant. Use the measures that matter to the question rather than assuming return is always sufficient.
• Limitations. State what the historical test can and cannot establish.
• A conclusion that matches the evidence. Some results are clear, some mixed, and some leave the question open.

Studies

Start with the summary

Each study opens with a short summary of the question and what we found. The full research follows with the method, tables, assumptions and limitations.

If the study uses a concept covered in GU Analyser’s learning library, we link to the relevant lesson.