Quantitative trading means finding an edge in data and trading it by rule. Instead of reading a chart and forming a view, a quant trader tests an idea against history, measures whether it actually holds, and then lets a model decide what to trade and how much. The decisions come from numbers and statistics rather than opinion. It is the discipline behind most serious automated trading, and the source of a great deal of retail wishful thinking.
What it is
A quantitative approach treats a market as a stream of data and asks a testable question: does this pattern predict anything? Perhaps a pair tends to drift after a certain kind of session, or two instruments tend to snap back together when they pull apart. The idea is turned into a precise rule, measured across thousands of past cases, and kept only if the numbers say it has an edge worth trading.
Three things separate it from guessing:
- Data — prices, spreads, volumes, sometimes news or economic figures.
- A model — the rule or formula that turns data into a trade decision.
- Statistics — the tests that decide whether an edge is real or just luck.
The output is usually an algorithm. Quant is where the rules come from; the algorithm is how they are run.
How it works
The work happens before a single live trade. A quant trader gathers clean historical data, forms a hypothesis, and tests it on data the model has never seen, to check the edge survives outside the sample it was built on. If it does, the rule is coded and sized so that no single trade can do real damage. Position sizing is treated as maths, not nerve: a fixed fraction of the account per trade, adjusted for how volatile the instrument is.
An edge, tested out of sample, traded small and repeated often — that is the whole loop.
The individual trade barely matters. What matters is a small statistical advantage applied enough times for the average to show through.
A worked example
Suppose you test a rule: when two correlated pairs drift more than two standard deviations apart, bet they converge. Across ten years of data it triggers 300 times and wins 58% of them. Each win averages £120, each loss £110, an expectancy of about £23 a trade. Over 300 trades that is roughly £6,900 gross. Now subtract spread and commission at £8 a round turn, which is £2,400, and the edge falls to £4,500. Costs were half the battle, and a quant sees that before trading. A hopeful backtester never subtracts them.
The retail reality
Institutional quant desks have clean tick data, fast execution, teams of researchers and near-zero trading costs. A retail trader has a MetaTrader terminal, a retail broker's spread and a VPS. The maths is identical; the conditions are not. Plenty of edges that look real in a backtest are eaten alive by retail spreads, slippage and swap — costs a large desk barely notices. That does not make quant pointless for retail. It makes cost and honesty the whole game: a modest, well-tested edge that survives real costs beats a spectacular backtest that only works with zero commission.
Common misconceptions
- More maths is not more edge. A complex model fitted to old data has usually just memorised it.
- Backtest profit is not live profit. If the test ignored spread, slippage and swap, it was never real.
- A good model still loses, often. Edge shows up over hundreds of trades, not the next five. A run of losers is normal and proves nothing on its own.
- "Quant" on a sales page is often decoration. The word sells. Ask to see out-of-sample results and real trading costs before you believe it.
The Karnek note
Karnek shows what a quantitative strategy is doing once it is live — the equity curve, the drawdown and the trade-by-trade record from your own terminal. Backtests promise; the live account tells the truth, and Karnek reads that account read-only. It reports the numbers and can never place or alter a trade.
Written and reviewed by the Karnek Research team. Last updated August 2026.
Educational content only - not financial advice. Past performance does not predict future results. Trading carries significant risk.