ECON4916 Topics in Applied Financial Econometrics
The Complete Exam Bible · S2 2026ECON4916 Overview
- Semester 2 2026
- University of Sydney
- Level 4 honours
- Closed-book final
- Economics
ECON4916 Topics in Applied Financial Econometrics is a University of Sydney unit on statistical and econometric methods for financial time series. The teaching materials connect mathematical reasoning with MATLAB implementation, but the central skill is not software syntax.
- Assessment focus The supervised final exam carries 50% and the in-semester test carries 25%.
- Core method State the stochastic object and assumptions before calculating or interpreting.
- Forecast discipline Compare models on aligned dates, horizons and decision-relevant loss.
- Notes strategy Place conditions beside every formula on the handwritten A4 sheet.
What you get in the full ECON4916 Exam Bible
- 6 chapters, 38 pages — every examinable topic, week by week
- 18 original practice questions, each with a model answer in the Final exam Multiple-choice and short-answer questions format
- 6 worked examples, solved the way ECON4916 marks
- 8 computed figures, redrawn from the course’s own numbers
- Exam-morning plan for the Final exam Multiple-choice and short-answer questions (50%) — what to revise last, where marks are lost
Free on this page: assessment table, chapter map, one worked example, key terms, FAQ. The rest is in the full exam bible.
How ECON4916 is assessed
| Component | Weight | Format |
|---|---|---|
| Final exam Multiple-choice and short-answer questions | 50% | Written exam in the formal examination period, in person and supervised, with a published duration of 2 hours |
| In-class quizzes multiple-choice or fill-in-black questions delivered via Canvas quiz | 5% | In-person quiz activity across multiple teaching weeks, with the best results counted under the published unit arrangement |
| Assignment 1 Short/long-answer questions submitted via Canvas | 10% | Group written work due in Week 7 under the current unit-outline record |
| In-semester test Paper-based test | 25% | Supervised paper-based test in Week 8 with a published duration of 1 hour |
| Assignment 2 Short/long-answer questions submitted via Canvas | 10% | Group written work due in Week 13 under the current unit-outline record |
Published numeric weights total 100%. The current official record does not label a component-specific hurdle. The final exam is described as closed book with one two-sided A4 handwritten notes sheet allowed; confirm operational instructions in the formal timetable and current Canvas site.
What other students are asking about ECON4916
- Where can I find lecture notes or study guides for ECON4916 at USyd?
- The ECON4916 study guide in AskSia Library covers Topics in Applied Financial Econometrics in 6 chapters, with the assessment table, a chapter-by-chapter topic map, key terms and a worked example, free to read on this page.
- Is there a final exam in ECON4916, and how much is it worth?
- Yes. In the offering this AskSia Library guide is built from, Final exam Multiple-choice and short-answer questions carries 50% of the ECON4916 mark; the full split is Final exam Multiple-choice and short-answer questions 50%, In-class quizzes multiple-choice or fill-in-black questions delivered via Canvas quiz 5%, Assignment 1 Short/long-answer questions submitted via Canvas 10%, In-s.
- What does ECON4916 cover on stationarity, dependence and forecasts?
- In AskSia Library's ECON4916 guide, the Stationarity, Dependence and Forecasts chapter explains sample paths, weak stationarity, autocorrelation and conditional means.
- What does ECON4916 cover on regression inference and linear restrictions?
- In AskSia Library's ECON4916 guide, the Regression Inference and Linear Restrictions chapter explains oLS consistency, asymptotic inference and Wald restrictions.
- What does ECON4916 cover on autoregressive and moving-average dynamics?
- In AskSia Library's ECON4916 guide, the Autoregressive and Moving-Average Dynamics chapter explains stationary moments, shock propagation and multi-step forecasts.
Current ECON4916 dates
| Date | Item | Control |
|---|---|---|
| 20 September 2026 at 11:59 pm | Assignment 1 | 10% group written task |
| 24 September 2026 at 6:30 pm | In-semester test | 25% supervised one-hour paper |
| 8 November 2026 at 11:59 pm | Assignment 2 | 10% group written task |
| 16 to 28 November 2026 | Formal examination period | The 50% final exam is scheduled within this period |
Dates are as published in the University of Sydney 2026 S2 unit overview and unit-outline record. Confirm exact deadlines and submission settings in the live LMS.
What ECON4916 covers
The unit builds a forecasting workflow from probability and conditional expectations through regression inference, ARMA dynamics, likelihood and honest out-of-sample model comparison.
Probability Tools for Financial Data
Random variables, distributions, moments and dependenceFinancial econometrics begins by defining what is random and how possible values receive probability.
- Sample spaces and random variables
- Discrete probability mass functions
- Continuous densities and cumulative distributions
- Expected values and raw moments
- Variance and covariance
- Independence and nonlinear dependence
Stationarity, Dependence and Forecasts
Sample paths, weak stationarity, autocorrelation and conditional meansA time series supplies one realised path through an ordered collection of random variables. Inference becomes possible only by placing defensible structure across dates.
- Time series and observed sample paths
- Strict and weak stationarity
- Ergodicity and time averages
- Autocovariance and autocorrelation
- Conditional expectations
- Mean-squared-error forecasts
Regression Inference and Linear Restrictions
OLS consistency, asymptotic inference and Wald restrictionsRegression analysis combines a fitting criterion with assumptions that support estimation and inference.
- Linear regression and OLS
- Fitted values and residuals
- Consistency and orthogonality
- Asymptotic normality
- Linear restriction matrices
- Wald test interpretation
Autoregressive and Moving-Average Dynamics
Stationary moments, shock propagation and multi-step forecastsAR and MA models encode different forms of financial time dependence. An autoregression places lagged outcomes in the conditional mean, so one innovation travels forward through recursive powers of the AR coefficient.
- AR recursion and innovation propagation
- Stationary AR moments
- AR conditional forecasts
- MA covariance cutoffs
- Stationarity and invertibility
- Residual dependence diagnostics
Likelihood and ARMA Estimation
Generated shocks, objective functions and estimator diagnosticsARMA estimation is difficult because lagged innovations are not directly observed. For every candidate parameter vector, the model must recover a new residual path before a likelihood or conditional sum of squares can be evaluated.
- OLS and moment estimators
- Likelihood and log-likelihood
- Scores and curvature
- Recursive shock recovery
- Initial-condition choices
- Optimisation and estimator diagnostics
Selecting and Validating Forecast Models
Penalised fit, holdout loss, shrinkage and leakage controlModel selection asks whether extra flexibility captures stable signal or merely follows noise in the estimation sample.
- Fit and complexity trade-off
- Penalised in-sample criteria
- Time-ordered holdout evaluation
- Rolling and recursive windows
- Ridge and lasso shrinkage
- Leakage and fair model comparison
Students must identify the stochastic object, state the information set, derive the relevant moment or forecast, estimate a model under explicit conditions, and decide whether the fitted result remains useful outside the sample that produced it. The teaching sequence begins with random variables, probability distributions, expectations, variance, covariance and dependence.
It then treats a time series as one observed sample path and develops weak stationarity, ergodicity, autocorrelation and conditional-mean forecasting. Regression material covers ordinary least squares, consistency, asymptotic normality and tests of linear restrictions. AR and MA models show how shocks travel through financial data, while likelihood and recursive shock recovery support estimation of mixed ARMA specifications.
Model selection closes the grounded sequence by balancing fit against complexity and comparing forecasts on time-ordered holdout observations. The Semester 2 2026 assessment has five published components: in-class quizzes, two group assignments, an in-semester test and a final exam. The final is worth 50%, runs for two hours and uses multiple-choice and short-answer questions.
The unit overview describes it as an in-person, supervised, closed-book paper with one two-sided A4 page of handwritten notes allowed. The official record does not publish a final question count or section split, so preparation should centre on methods and visible working rather than a guessed paper template.
Later official topics include volatility, risk and high-frequency data, but the detailed materials available here stop at model selection; this guide stays within those taught methods.
ECON4916 cheatsheet for the final exam
AskSia publishes a free ECON4916 cheatsheet for the final exam, a Cheatsheet that follows the official permitted-materials rule for this assessment and condenses the topics in this guide. Open the ECON4916 cheatsheet.
Worked example · free
Stationary AR(1) moments and forecast
- +1Check |0.6| < 1, so the stationary AR(1) moments exist.
- +1Solve mu = 1.2 + 0.6mu to obtain mu = 3.
- +1Compute variance as 4 divided by 1 minus 0.6 squared, giving 6.25.
- +1Use rho_2 = 0.6 squared to obtain 0.36.
- +1Forecast 1.2 + 0.6 times 4 = 3.6 and note that it moves toward the long-run mean.
Key terms
- Random variable
- A mapping from outcomes in a sample space to real numbers, allowing uncertainty to be described with a distribution.
- Probability mass function
- A function assigning nonnegative probabilities to the support points of a discrete random variable, with total mass equal to one.
- Probability density
- A function whose integral over an interval gives probability for a continuous random variable; its point height is not itself a probability.
- Conditional expectation
- The mean of a random variable given a stated information set, and the optimal forecast under conditional squared-error loss.
- Weak stationarity
- A time-series property requiring a finite second moment, constant mean and covariance determined by lag rather than calendar time.
- Ergodicity
- A dependence restriction that lets time averages learn population moments as the observed sample grows.
- Autocorrelation function
- The covariance between observations at a given lag divided by the stationary variance, producing a unit-free dependence measure.
- Orthogonality condition
- The zero population moment between regressors and the regression error that supports OLS consistency under the other regularity conditions.
- Wald test
- A quadratic test comparing estimated linear restrictions with their null values after scaling by estimated covariance.
- Autoregressive model
- A dynamic model whose conditional mean depends on lagged values of the modelled variable.
- Moving-average model
- A dynamic model expressing the current value through current and finitely many lagged innovations.
- Maximum likelihood
- An estimation method choosing the parameter vector that gives the observed sample the highest likelihood inside a maintained model family.
- Out-of-sample loss
- A forecast-error measure calculated on observations not used to fit the candidate model, with time order preserved.
ECON4916 FAQ
How is the unit assessed?
The 2026 S2 structure assigns 5% to in-class quizzes, 10% to Assignment 1, 25% to the in-semester test, 10% to Assignment 2 and 50% to the final exam. The numeric total is 100%. The two assignments are group written work, while the tests are secured supervised assessments.
What is the final exam format?
The current unit-outline record describes multiple-choice and short-answer questions in a two-hour written exam during the formal examination period. The unit overview calls it in person, supervised and closed book, with one two-sided A4 handwritten notes sheet allowed. It does not publish the final question count or section split.
How difficult is the mathematical work?
The challenge is cumulative rather than based on one advanced trick. Probability, conditional expectation, regression inference, ARMA recursion, likelihood and validation depend on earlier definitions and assumptions. Students who show the information set, conditions and intermediate steps can diagnose errors; memorising isolated formulae makes changed signs or horizons much harder to handle.
Is there a final-exam hurdle?
The official 2026 unit record does not label a component-specific hurdle. It does state that failure to take the final results in an absent-failure outcome regardless of other results. Treat attendance and special-consideration instructions as operational requirements, and confirm the current wording in the unit outline and Canvas before the sitting.
Which offering does this guide follow?
This guide follows the Semester 2 2026 overview. A reused Canvas home heading still says 2025 S1, but its current announcements carry 2026 dates and the dedicated overview identifies 2026 S2. Current assessment dates and the official unit-outline availability align with the Semester 2 offering.
What should go on the handwritten notes sheet?
Group formulae by task and put their conditions beside them. Include support and moment checks, weak-stationarity conditions, the information set for conditional forecasts, the OLS decomposition, restriction notation, ARMA shock paths, likelihood direction and aligned holdout-loss rules. A formula without its model and assumptions is too easy to misuse.
Are past papers included with these materials?
No full past final or practice final appears in the materials available here. Two Tutorial 3 items are identified as 2023 midterm questions, which provides evidence about conditional-expectation and moment-reasoning question types. This guide uses entirely new questions and numbers rather than reproducing or lightly rewriting those proprietary stems.
Questions students ask in each ECON4916 chapter
- Why can a density be greater than one?
- A density height is not a probability. Probability is the integrated area over an interval, so a narrow distribution may need a height above one while its total area remains exactly one. For a continuous variable, any exact point still has probability zero because that point has zero width. Chapter 01 · Probability Tools for Financial Data
- Does a flat-looking plot prove stationarity?
- No. A chart may suggest a stable mean while hiding changes in variance, dependence or higher moments. Weak stationarity is a population claim about finite second moments, constant mean and lag-based covariance. Use plots for diagnosis, then examine moments, breaks and model assumptions rather than treating appearance as proof. Chapter 02 · Stationarity, Dependence and Forecasts
- Why does OLS consistency need orthogonality?
- The estimator error contains the sample average of x_t times epsilon_t. A law of large numbers takes that term toward its population expectation. If the expectation is zero, the component vanishes; if it is not, more observations do not remove the systematic relation between regressors and the error. Chapter 03 · Regression Inference and Linear Restrictions
- Why does an AR shock persist indefinitely?
- Recursive substitution carries the innovation into every future outcome through powers of the AR coefficient. Under stationarity those powers shrink toward zero, so the response becomes small but does not reach a structural cutoff. A negative coefficient also alternates the response sign across successive horizons. Chapter 04 · Autoregressive and Moving-Average Dynamics
- Why can AR coefficients be estimated by OLS?
- Lagged outcomes are observed regressors, and the current innovation is uncorrelated with those past values under the model's assumptions. That makes a direct regression available. The conclusion still needs stationarity, rank and error conditions appropriate to the desired consistency and inference results. Chapter 05 · Likelihood and ARMA Estimation
- Why does training error favour larger models?
- Additional regressors or lags let a fitted model reproduce more features of the estimation sample. In ordinary least squares, residual sum of squares cannot rise when regressors are added. Without a complexity penalty or later evaluation, training fit rewards flexibility even when it captures noise that will not persist. Chapter 06 · Selecting and Validating Forecast Models
How to study for the exam
Revise the unit as one forecasting pipeline. Begin each study session by writing the stochastic object and its support or information set. Rebuild probability moments and conditional expectations without software, then use MATLAB output to verify the same quantities. For stationarity, separate time invariance from the dependence restriction that supports a law of large numbers.
For regression, keep coefficient consistency separate from the covariance estimator used for inference. For ARMA, trace one shock across horizons and distinguish stationary recursion from invertible shock recovery. For likelihood, write the objective and whether it is maximised before interpreting an optimiser result.
For model selection, preserve time order, align candidate forecasts on the same dates and horizon, and state why the chosen loss matches the financial decision. Build the handwritten A4 page gradually rather than copying formulae at the end. Each entry should include the condition that authorises it and one quick direction check.
Practise short answers with fresh numbers, show working, and close every result with a conclusion no broader than the maintained assumptions or tested null. Use the current Canvas site for operational exam instructions because the published information does not provide a final question count or complete permitted-materials list.
Your AI Economics tutor for ECON4916
Stuck on a hard ECON4916 question? Sia is AskSia’s AI Economics tutor — ask any ECON4916 Topics in Applied Financial Econometrics question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.