AQIF

Both, but weighted toward application. Expect definitional and computational items on statistics and instruments, and longer scenario items where you must judge whether a model, backtest or risk estimate is sound. Memorising formulas without understanding assumptions is usually not enough to pass.

It is a mid-career credential, so it assumes comfort with probability, regression and basic calculus, plus familiarity with equities, fixed income and derivatives. Candidates strong in only one area find the cross-domain questions hardest. If your maths is rusty, rebuild it before starting the machine-learning material.

Those figures are set and periodically revised by the awarding body, so check its official candidate handbook for the current question count and appointment length rather than relying on second-hand reports. Your booking confirmation will state the format you have purchased.

Fees, rescheduling terms and the passing standard are published by the awarding body and change over time. Consult its official website or handbook for the current figures before you budget or book. No third-party summary, including this one, should be treated as authoritative on price or cut score.

Retake rules - permitted attempts, waiting periods and any additional fees - are defined in the awarding body's policy documents. Check those before booking, because the practical consequence of a failed attempt differs depending on whether a waiting period applies to you.

Start with quantitative methods, since the data-science and backtesting material assumes it. Then markets and instruments, so your models have realistic inputs. Build portfolio and risk topics on top. Leave ethics until late but do not skip it: it is a small share of the paper yet easy marks for candidates who read the scenarios carefully.

Most candidates report several months of part-time study, longer if the programming or the finance side is new. Judge yourself by coverage of the syllabus and by whether you can explain each method's failure modes, not by hours logged. Practice questions are a diagnostic, not a substitute for understanding.

Three recurring reasons: treating backtested performance as evidence rather than hypothesis, weak grasp of statistical assumptions such as stationarity and multiple testing, and ignoring transaction costs, survivorship bias and look-ahead leakage. A fourth is neglect of ethics and model-governance material as 'free' content.

The syllabus emphasises concepts and interpretation, but fluency in a language such as Python or R makes the data-science and backtesting domains far easier to internalise. Permitted aids at the test centre are specified by the awarding body; verify before exam day rather than assuming a policy calculator is allowed.

Recognition varies by firm and region, and no credential substitutes for demonstrable project work. It is most useful as structured evidence of breadth across quant methods, data science and risk for candidates whose current role covers only part of that span. Ask target employers directly what they value.