Quantitative trading is the systematic use of data, mathematical or statistical models, and tested rules to decide what to trade and how much risk to take. Its practical task is to turn a researched idea into positions that can actually be executed and monitored. A strong forecast can still produce a poor trade if position size, costs or fills are wrong.
The defining feature is a reproducible decision process. Researchers specify the inputs, selection rules, position sizing and conditions for changing or closing exposure. Human judgment remains central to choosing the hypothesis, checking data, setting limits and deciding whether the evidence still supports the strategy. Using a numerical screen while choosing every trade by discretion is a hybrid approach; a screen alone does not make the whole trading process systematic.
Follow the decision into the market
The work extends beyond finding a pattern in prices. A useful research and operating chain asks different questions at each stage:
| Stage | What must be specified? | What can go wrong? |
|---|---|---|
| Research | An economic hypothesis, eligible data and a rule that can be tested | A historical relationship disappears or was found by excessive searching |
| Validation | A backtest with information available at each decision time | Look-ahead bias or overfitting creates an attractive but misleading result |
| Portfolio construction | Desired holdings after risk, capital and transaction-cost constraints | Many individually appealing trades concentrate the same exposure |
| Execution | Orders, fill assumptions and actual position changes | The intended quantity does not fill, or trading consumes the expected gain |
| Monitoring | Reconciled holdings, cash, exposure, costs and system health | A data failure or changed market relationship goes undetected |
CFA Institute's overview of quantitative active strategies connects the investment thesis, data preparation, backtesting and evaluation to portfolio construction using risk and cost models. The operating loop adds a continuing question: does the implemented strategy still behave consistently with that evidence?
Original fictional USD example. Complete entry and exit fills and USD 15 additional cash charges are stipulated. The workflow feeds operating evidence back into research; the ledger illustrates accounting, not a predicted profit.
A signal is not a realised profit
Suppose a fictional model produces a long signal. After applying its sizing and risk rules, the strategy buys 100 shares at an actual fill price of USD 100 per share. It later sells all 100 at USD 102. Stipulate total additional cash charges of USD 15, with no dividends, financing charges or other cash flows beyond those already specified.
The completed round trip earns
The cash ledger independently gives the same answer: a USD 10,000 purchase outflow, a USD 10,200 sale inflow and USD 15 of charges leave a USD 185 gain and no remaining shares. Any spread or slippage already reflected in those fill prices must not be deducted again.
These are stipulated outcomes, not prices the model was guaranteed to obtain. Before the trade, the exit price, achievable quantity and costs were uncertain. If only part of the order filled, profit would have to be calculated from actual holdings. The Positions, Trades, Fills and P&L lesson develops that ledger, including partial execution; its full text requires membership and a public preview is available.
Quantitative does not mean fastest or fully unattended
Algorithmic trading describes the use of programmed procedures in trading, including execution. A broker can algorithmically split an order that a discretionary investor chose. Conversely, a quantitatively specified strategy can have a person submit its orders and supervise exceptions. The research process and degree of execution automation are related choices, not interchangeable labels.
High-frequency trading is a particular setting where speed and short horizons matter; quantitative rules can also operate at much slower decision frequencies. The SEC's 2020 staff report describes a broad range of algorithmic functions, including institutional order execution, rather than a single strategy type.
A quantitative process makes assumptions and decisions easier to test; it does not create an edge by definition. Losses can reflect ordinary uncertainty, implementation failure or model risk. Monitoring must distinguish those possibilities before a rule is revised or retired. For a fuller research workflow, the Quant Research Dossier assembles the hypothesis, data chronology, validation, net results and operating decision into one reproducible record. The full lesson requires membership; its public preview shows the scope.
Sources
- Rishi K. Narang, Inside the Black Box: A Simple Guide to Quantitative and High-Frequency Trading, 2nd edition, 2013. Chapter 2, pp. 14–18, on systematic decisions and trading-system functions; Chapter 10, pp. 193–195, on monitoring.
- CFA Institute, Active Equity Investing: Strategies, 2026 curriculum public overview: quantitative investment process and its pitfalls.
- U.S. Securities and Exchange Commission, Staff Report on Algorithmic Trading in U.S. Capital Markets, August 2020, sections I and III: algorithmic functions and institutional execution.