Quantitative research and backtesting

Quantitative research turns a hypothesis about data into explicit, testable rules. Ranking securities by their price change over the previous 20 observed sessions specifies a calculation that another researcher can inspect. Record the inputs, window, decision time and assumptions. A reproducible rule does not establish future profitability.

A backtest applies those rules to history using simulated execution. Signal timing, quantity calculation, fill prices, fees and insufficient-cash handling can all change the result.

The KabuForge reference course uses the previous observed close to calculate signals and quantities, then executes at the next observed open. Historical Paper replays past observations; it is separate from forward simulation.

Consider a signal and quantity calculated after Monday’s close, with execution at the next observed open. Using Tuesday’s close for Monday’s decision would look ahead. A signal is a trading decision, an order is an instruction, and a fill is an executed or simulated transaction.

Price momentum

Momentum measures price change over a fixed past window. A move from 100 to 110 is a 10% simple price change. Fix the window, adjustment basis and signal timing.

Moving average (MA / SMA)

A simple moving average is the arithmetic mean of prices in its window. Prices 100, 102 and 104 have a three-observation SMA of 102. A crossover rule uses the relationship between short and long averages; the indicator and trading rule are separate.

Portfolio

A portfolio combines cash and holdings. Inspect position weights, concentration and cash as well as security counts.

Rebalancing

Rebalancing adjusts holdings toward a target composition at a specified frequency or trigger. Monthly rebalancing does not necessarily trade every day. Costs and execution constraints still apply.

Read the screenshot-led tutorial · All basic concepts · Research methodology