Algorithmic Trading

Use conditional probability and Bayes' rule to update a trading hypothesis after new evidence.

Calculate and interpret mean, variance, covariance, and correlation from return data.

Compare normal, Student-t, and empirical return distributions using quantiles, asymmetry, and tail behavior.

Apply the laws of large numbers and central limit theorem while stating when market dependence weakens their conclusions.

Fit a linear regression and quantify coefficient uncertainty using heteroskedasticity-robust standard errors.

Construct an empirical bootstrap confidence interval for a strategy statistic under independent sampling.

Perform a hypothesis test on mean returns and distinguish statistical significance from economic significance.

Control the false discovery rate when evaluating a collection of candidate signals.

Choose a data frequency and liquid-equity or futures universe that match a strategy's intended holding period.

Adjust equity data for splits and dividends without introducing future information.

Construct a continuous futures series while making the roll rule and price adjustment explicit.

Align features, prices, trading calendars, and execution timestamps without look-ahead bias.

Handle missing observations and outliers using rules justified by market mechanics.

Detect survivorship, selection, publication, and revision biases in a research dataset.

Create features and labels whose availability times respect the intended trading decision time.

Build a reproducible, versioned research dataset with automated schema and range checks.

Compute simple and log returns from price data and determine when levels, differences, or returns are appropriate modeling targets.

Use a unit-root test to assess stationarity and diagnose the risk of spurious price regressions.

Measure and interpret autocorrelation in returns, squared returns, and candidate signals.

Fit an autoregressive model and evaluate its one-step-ahead forecast against a naive baseline.

Test two price series for cointegration and estimate a stationary spread when supported by the data.

Estimate conditional volatility with an exponentially weighted model and compare it with a basic GARCH model.

Detect a structural break or regime change in a strategy's historical performance.

Generate dependence-preserving null data with a block bootstrap for a strategy robustness check.

Convert signals into positions and calculate returns, turnover, exposure, and cash usage.

Implement a vectorized backtest for a single-asset or cross-sectional strategy.

Model commissions, bid-ask spread, slippage, borrow costs, and market impact at an appropriate level of realism.

Split time-series data into development, validation, and untouched test periods.

Run a walk-forward evaluation with all model choices confined to past data.

Calculate and interpret CAGR, volatility, Sharpe ratio, drawdown, hit rate, tail-risk, and benchmark-relative metrics.

Evaluate strategy stability across parameter ranges using sensitivity surfaces and local perturbations.

Stress-test net performance against escalating transaction costs, slippage, and execution latency.

Construct a time-series momentum signal from past futures returns without overlapping the decision timestamp.

Construct a cross-sectional momentum signal that ranks liquid equities on a common date.

Build a short-horizon mean-reversion signal using standardized deviations from a rolling reference level.

Estimate a hedge ratio and convert a cointegrated spread into entry and exit signals.

Calculate a futures carry signal from contract prices and time to expiration.

Transform point-in-time equity characteristics into neutralized cross-sectional factor scores.

Normalize heterogeneous signals by volatility and combine them into a composite forecast.

Conduct subperiod and universe sensitivity tests to assess robustness across market regimes and asset choices.

Formulate a trading prediction task by specifying the observation unit, target, horizon, and decision timestamp.

Construct a leakage-safe preprocessing pipeline that fits every transformation only on past training data.

Configure purged and embargoed cross-validation when labels overlap in time.

Train a regularized linear model and interpret its coefficients as a predictive baseline.

Train a gradient-boosted tree model for nonlinear return prediction or trade classification.

Tune model hyperparameters using time-aware validation rather than random cross-validation.

Evaluate predictions with both statistical loss measures and strategy-level net performance.

Test whether a machine-learning model adds stable value over a simple rule-based benchmark.

Estimate a return covariance matrix and improve its stability with shrinkage.

Decompose portfolio risk into asset, strategy, and common-factor contributions.

Compute a constrained mean-variance portfolio and explain its sensitivity to expected-return estimates.

Construct an equal-risk-contribution portfolio from strategy or asset return series.

Scale positions to a target volatility while imposing leverage and concentration limits.

Neutralize specified market, sector, or style-factor exposures in a cross-sectional portfolio.

Choose rebalancing trades by balancing forecast changes against transaction costs.

Calculate value at risk and expected shortfall, then test their conclusions against historical and hypothetical stress scenarios.

Reconstruct best bid, best ask, spread, and displayed imbalance from limit-order-book updates.

Distinguish quoted spread, effective spread, realized spread, and implementation shortfall.

Measure intraday volume, volatility, and spread seasonality from high-frequency observations.

Estimate how order size and participation rate affect temporary market impact.

Choose an aggressive or passive market, limit, stop, or pegged order for a specified urgency and adverse-selection risk.

Compare TWAP, VWAP, and participation-based execution schedules on cost and completion risk.

Simulate fills for a simple event-driven strategy using latency, spread, and conservative queue assumptions.

Evaluate an intraday signal after execution costs with trade-level attribution.

Design interfaces between market data, feature computation, forecasting, portfolio, execution, and accounting components.

Implement idempotent handling of duplicated, delayed, or out-of-order market-data events.

Package research code and configuration so that a historical result can be reproduced.

Reconcile internal orders, fills, positions, and cash against broker records.

Specify pre-trade controls for stale data, position limits, price collars, order size, and duplicate orders.

Specify a kill switch and a safe recovery procedure for a partially executed portfolio.

Deploy a strategy to a broker sandbox or paper-trading environment with secrets kept outside source code.

Design an operational dashboard covering system health, exposure, P&L, execution quality, and active alerts.

Construct rolling estimates of exposure, turnover, cost, drawdown, and forecast efficacy.

Compare live results with backtest expectations using prediction intervals rather than point estimates.

Separate live underperformance into signal decay, execution slippage, risk scaling, and data-quality components.

Run a post-trade analysis that attributes implementation shortfall to delay, spread, impact, and price movement.

Detect feature or prediction drift without treating every distribution change as alpha decay.

Create statistically justified alerts for deviations in data, positions, execution, or performance.

Define evidence-based criteria for reducing risk, pausing, modifying, or retiring a strategy.

Design a quantitative research log that distinguishes precommitted tests from exploratory findings.

Write a precommitted research specification for one liquid-equity or futures strategy.

Audit the capstone dataset for timestamp integrity, survivorship, adjustments, and missing observations.

Implement one reproducible signal and map it to positions using only contemporaneously available information.

Calibrate a transaction-cost model from the strategy's turnover and the market's liquidity.

Run a locked walk-forward evaluation and report performance with uncertainty estimates.

Apply portfolio sizing, exposure limits, and scenario-based risk checks to the proposed strategy.

Configure a paper-trading trial with reconciliation, alerts, and explicit stop conditions.

Produce a decision memo that argues to deploy, revise, or reject the strategy based on the collected evidence.