Scorecard-first mechanical system construction

Match one pattern to the map

Piling extra indicators onto a still-forming chart is treated as the weakest path to real-time decisions. Construction of a mechanical-trading-system starts by matching one pattern to an ideal long-and-short map and then filtering surplus signals.

Those rules stay inside one coded procedure so entry, exit, and abstention can be tested together over the system holding period.

A two-average crossover is defined as long when the faster average crosses up through the slower average and short when it crosses down, with conventional first lengths of 9 and 18.

Filter the surplus crosses

When those lengths sit tight to price, range stretches produce repeated opposite crosses that miss the ideal map. That cluster is a whipsaw.

The same book then needs a no-trade filter, a channel substitute, or different inputs so surplus signals are not left standing in for the ideal long-and-short sequence.

Retune lengths when market speed changes

Keeping the same 9 and 18 lengths but switching to exponential averages removed several of those false crosses in the illustrated sample. Further input search is the system-optimization step that produced a speed-adaptive pair.

System-optimization searches the same rule inputs, average lengths, filters, targets, and stops, so the procedure better matches an ideal signal map without adding untestable chart clutter.

Re-estimating average lengths on a weekly or daily cycle is presented as one way to follow changing market character. That later-sample refresh is walk-forward-analysis, used after market character changes instead of locking lengths to one historical stretch or chasing last week's fit without a new test.

An alternative is a dynamic-length-pair that tightens in fast ranges near 9 and 11 and widens in congestion near 10 and 30, so the pair need not be re-estimated on a fixed calendar.

State every exit in the book

A complete mechanical book states when to enter, when to exit a working trade, and when to exit a failing trade.

The right-exit closes a working trade and can be coded as a cash target, one or two average-true-range or standard-deviation units from entry, or an oscillator extreme. The wrong-exit closes a failing trade and can be a fixed cash stop or the same volatility measures. Matching the two sizes creates a 1:1 payoff design.

A crossover engine can reverse as a reversal-system by trading enough size to flatten and invert, or flatten only with half that size.

Accept a book only on the scorecard

Borrowed or opaque procedures still require an independent test and a written log of symbol, timeframe, lookback, trade count, percent profitable, average trade, and win-to-loss ratio.

The cardinal-profitability-construct multiplies profit factor, percent profitable, and average-win-to-average-loss. Values below 1.2 were treated as not ready, and several large-profit sample books failed because that win-to-loss ratio was weak.

Hypothetical strategy reports are not live results and typically understate slippage and losers. A negative net or a drawdown larger than the account is a reject. About one-fifth of early effort is assigned to design and one-fifth to testing before execution is only following the script.