Signal Backtest vs. Portfolio Backtest

A signal backtest checks a rule. A portfolio backtest checks a simulated account.

Stingray provides signal replay by default. It helps define, replay, review, and monitor trading strategies. It does not simulate a complete portfolio by default, and it does not place trades.

The direct comparison

Question	Signal replay	Complete portfolio simulation
Main unit	A rule match or distinct trigger	A position, order, and account state
Starting capital	Not required	Required
Position sizing	Not assumed	Required for every trade
Cash tracking	Not included	Updated after orders, costs, and exits
Fees and spread	Disclosed as exclusions unless separately modeled	Applied to simulated fills
Slippage	Disclosed as an exclusion unless separately modeled	Estimated from stated assumptions
Concurrent signals	Can be reviewed as overlapping events	Must compete for capital and risk limits
Portfolio constraints	Outside standard replay	Applied across positions and orders
Execution assumptions	No simulated order is required	Entry, exit, fill, timing, and partial-fill rules are required
Result	Trigger evidence and follow-up outcomes	Simulated positions, account value, and portfolio statistics

The two tests answer different questions. Read each result within its stated scope.

What a Stingray signal replay shows

A Stingray replay starts with a defined trading rule. The rule names the market, source, conditions, thresholds, interval, date range, and waiting period between repeated alerts.

The historical replay can show:

• When the rule matched historical observations
• Which matches became distinct triggers
• How a waiting period changes the number of alerts
• What happened to the observed market after each eligible trigger at stated follow-up horizons
• Which data limits and assumptions affect the result

These outputs help answer whether the signal appeared often enough to inspect, whether it clustered in one episode, and whether the observed outcomes support more research.

Forward returns belong to the observed market after each signal. They are not a simulated account return. The replay has no default position size, cash balance, entry order, or exit order.

What a complete portfolio simulator must add

A portfolio changes after every simulated order. The simulator must carry that state from one event to the next.

Positions

The simulator records open positions, entry prices, quantities, direction, and exits. A new signal may add to an existing position, reduce it, reverse it, or be ignored. The rule for each case must be fixed.

Position sizing

Every accepted signal needs a size. The method may use a fixed amount, a percentage of capital, a volatility rule, or another stated calculation. Size affects both gains and losses.

Cash and collateral

The account needs a starting balance. Buys, sells, reserved capital, realized results, and costs update the available balance. A signal cannot create a position when the account rules say the required capital is unavailable.

Fees, spread, and slippage

The simulator needs a fee schedule and a method for estimating the difference between the observed price and a plausible fill. Spread and slippage can remove a small apparent advantage.

Concurrent signals

Several strategies or markets can signal at the same time. The simulator needs a priority rule and a way to allocate limited capital. Reviewing each signal independently does not answer that portfolio question.

Portfolio constraints

Constraints can cap total exposure, position count, position size, concentration, daily loss, or another account-level measure. The simulator must apply them before accepting a simulated order.

Execution assumptions

A simulated trade needs rules for order type, submission time, fill price, partial fills, missed fills, exits, and data granularity. Historical bars may not reveal the exact order of prices inside the interval, so the simulator must state how it handles that uncertainty.

Why the results can differ

Suppose a signal replay finds 30 eligible triggers. A portfolio simulation may accept fewer because:

• The account already has an open position
• Available cash is below the required amount
• A position or exposure limit blocks the trade
• Two signals compete for the same capital
• Spread, fees, or slippage remove the expected result
• The execution rule cannot obtain the assumed price

The signal evidence remains useful. It describes the rule’s historical events. The portfolio result answers a later question about how stated account and execution rules would have handled those events.

Choose the test that matches the decision

Use a signal replay when you need to decide whether a market condition is defined clearly, occurs often enough, survives basic data checks, and deserves monitoring.

Use a complete portfolio simulator when the decision depends on capital allocation, overlapping positions, trading costs, account constraints, and simulated order handling.

Many research processes use both in sequence. First reject weak signals. Then pass the surviving event set into a separate portfolio model with written assumptions. Keep the signal rule and portfolio rules versioned so a later reviewer can reproduce each result.

Read a Stingray replay within scope

When reviewing a Stingray result:

1. Confirm the tested rule and date range.
2. Check data coverage and trigger counting.
3. Treat forward returns as outcomes after signals.
4. Do not read those outcomes as portfolio P&L.
5. Record which costs and execution assumptions are absent.
6. Decide whether the signal should be rejected, revised, or monitored.

The backtesting methodology explains how Stingray counts triggers and measures follow-up outcomes. How to Validate a Trading Strategy covers holdouts, walk-forward checks, regimes, and rejection criteria. Use How to Read a Backtest Card for the field-level review.