← Strategy Builder
Strategy specification guide

What parameters belong in a trading strategy?

Inspect the selection, sizing and exit parameters in a TradePolaris momentum template, then use the checklist before evaluating a backtest.

TradePolaris product reference · Reviewed

The specification

Make the rules explicit.

A trading strategy specifies its market, eligible universe, signal, entry conditions, ranking, position sizing and exits. A reproducible backtest also needs evaluation dates, data coverage, trading costs and a validation plan.

The example below is the classic momentum template shipped with TradePolaris. These are inspectable template settings, with no performance result or parameter recommendation attached.

A concrete example

Read the momentum template's parameters.

Selection and sizing rules from the downloadable classic momentum specification.
ParameterTemplate valueMeaning
Market and cadenceUS, quarterlyThe template selects a US equity basket on a quarterly schedule.
Minimum history300 trading daysEligibility requires enough history before a stock enters the candidate universe.
Minimum priceUSD 5The universe filter removes names below the stated price floor.
Average daily traded valueUSD 5,000,000 over 60 trading daysThis is a liquidity filter. It does not establish that every order can fill at the quoted price.
Momentum lookback231 trading daysThe return measurement spans this many trading days before the skipped interval.
Recent interval skipped21 trading daysThe signal ends before the most recent interval, as shown in the formula below.
Entry and selectionMomentum above 0, top 30Eligible names are ranked by the descending momentum score.
Position sizingEqual weighting, divisor top_nEach selected name uses the configured target count as its divisor. Unfilled slots can leave cash.
Maximum position weight10 percent of capitalThis constraint caps one position; it is separate from the equal-weight target.
Stop rule5 times ATR, 20-day windowThe stop is a strategy rule. Simulated gap fills and costs still affect the realized loss.
Capital and constraintsUSD 100,000, at most 30 positionsThe downloadable specification also states the maximum position weight and rotation exit rule.

The example capital is a simulation input. Its universe filters do not certify complete historical membership or point-in-time data coverage.

Signal definition

Separate the lookback from the skipped interval.

For decision time t, the momentum signal measures the return ending skip trading days earlier. It looks back another lookback trading days from that endpoint.

momentum(t) = close[t - skip] / close[t - skip - lookback] - 1

Here, lookback is 231 and skip is 21. The starting observation is 252 trading days before t. Trading-day counts approximate a calendar convention and should stay explicit in the research record.

This is a price-ratio signal. Record the input series and its corporate-action adjustment policy before interpreting it as a return; the formula alone does not establish a total-return series.

The template requires positive momentum, ranks eligible names and selects up to 30. Equal weighting uses that configured target count as the divisor, so fewer eligible names can leave part of the simulated capital in cash.

Before the evaluation

Keep the run settings beside the strategy.

Universe and dates

Retain historical eligibility, coverage dates, revisions and the treatment of delisted names.

Trading costs

Record fill timing, spread, impact and applicable borrow assumptions before reading returns.

Held-out evaluation

Separate fitting from evaluation and inspect the chronological walk-forward results.

Selection history

Retain attempted variants so overfitting checks account for repeated strategy selection.

The JSON describes strategy rules. It does not contain a completed backtest, a data version or proof that a candidate passes the validation gates.

Common questions

Questions to settle before running.

Is a lookback the same as a holding period?
No. The lookback measures the signal's past observations. The rebalance cadence and exit rules determine when the strategy reviews or closes positions.
Does changing one parameter require another validation run?
Treat the changed specification as a new candidate and retain the trial history. Repeated selection affects how evidence for the final strategy should be interpreted.
Does a valid strategy specification establish profitability?
No. Schema validation checks the document and its static limits. A performance claim needs a completed evaluation with data coverage, costs and held-out results.
Can an agent use this as a research input?
The specification and parameter meanings are public and readable without an account. Building or running the strategy still follows TradePolaris account access and the available product controls.
Inspect and continue

Keep the specification and its evidence together.

Download the complete strategy specification, including the position-weight constraint, entry filter and exit rules.

Inspect the specification fingerprint

SHA-256 of the compact JSON representation: e97742e24493ec34dbe5cbfd50777228de85e0682fbe0e05ceadafdb94d3d787

Read walk-forward testing vs backtesting, point-in-time data and survivorship bias, and Deflated Sharpe and backtest overfitting for the evaluation details.

Inspect the TradePolaris validation gates or use the Strategy Builder to define your own rules. Current pricing and account access apply to signed-in runs.

Backtested and hypothetical results do not predict future performance. TradePolaris provides research software and does not provide investment advice.