Trading system
Two descriptions share an entry signal, but one exits at 1R while the other uses a 2R target or time-based rule. They are different system variants. In a simplified two-outcome model, 40% of results are +2R and 60% are −1R, giving gross expectancy of 0.4 × 2R − 0.6 × 1R = 0.2R. A constant cost of 0.15R on every trade reduces it to 0.05R. These are assumed model parameters, not an established edge in a real system.
A complete specification beyond the signal
Define the instrument universe, data, decision time, entries, exits, sizing, portfolio limits and handling of conflicts or failures. Exits can use targets, stops, time or another specified rule; a fixed take-profit is not mandatory. Discretionary guidance needs boundaries and a record of its application. Reproducibility means reconstructing decisions from information available then, rather than obtaining identical outcomes on every execution.
Testing includes execution conditions
If a rule depends on a completed candle’s close, it cannot use that value before it is known. A signal price need not be the fill price. Include Bid/Ask sides, commissions, financing, slippage and size restrictions. If one candle contains both SL and TP levels, OHLC alone may not establish their order. Assuming a favourable path in every ambiguous case can artificially improve performance.
Precision does not establish an edge
Explicit rules can be tested and compared, but a formal specification or code guarantees no positive expectancy. There is no universal hierarchy in which entry is always least important or exits and sizing independently create an edge. Assess interactions, sample coverage and out-of-sample stability. Version changes and record tested alternatives so the best result selected after many trials is not presented as independent confirmation.
Further learning and sources
8.1 Trading style versus trading system: when you are in the market and exactly what you do; 15.3 Strategy building — from hypothesis to testable rules; 15.4 Backtesting and forward testing — methods, evidence and pitfalls.
