How Strategy Review Works in Forex

What “strategy review” means in forex

Strategy review is a process for evaluating whether a forex trading strategy’s stated rules and assumptions match real-world outcomes. The focus is on the mechanism: how decisions were made, whether rules were followed, and what factors influenced results. In this sense, strategy review is not prediction. It is structured learning from past sessions under explicitly stated assumptions.

A useful way to define it is by separating three layers:

Plan: the strategy description (entry/exit logic, risk rules, time-of-day assumptions, and any execution constraints).
Execution: what was actually done (orders placed, fills, timing, slippage, and whether rules were followed).
Observed results: performance measures calculated from the execution record, with costs included.

Because forex markets and trading conditions change, the review should be written so a reader can verify the steps and recompute the key calculations from the same inputs.

The mechanism: inputs, process steps, and outputs

A strategy review usually follows a repeatable sequence.

1) Choose a review scope and assumptions

Before using any numbers, define what is being reviewed. Examples of scope choices are:

A specific strategy version (rules at a certain point in time).
A set of sessions or dates.
A subset of trades that meet predefined criteria (e.g., only trades placed during certain hours).

You also state assumptions used in later calculations. If you estimate costs or convert currencies, the method should be described. This matters because small differences in assumptions can change metrics.

2) Collect consistent inputs

Inputs are the raw material for comparison. Typical categories include:

Rule notes: what the strategy said to do in given situations.
Trade log: timestamps, order types, prices, and outcomes.
Execution details: fills versus intended entry/exit, slippage, and any partial fills.
Costs: spread or commission-like charges, plus any other stated transaction costs.
Context metadata (if used): e.g., session type, instrument identifier, or other constraints.

Stable mechanics require consistent labeling. If two data sources represent time zones differently, or if “planned entry” is logged differently across periods, comparisons become unreliable.

3) Map planned decisions to actual decisions

The core comparison step is to line up each decision in the plan with the corresponding actions taken in execution. This is where many reviews become ambiguous if not done carefully.

A reviewer typically checks:

Rule adherence: Was the stated rule followed?
Timing adherence: Did execution occur at the intended moment or after a delay?
Price quality: How much did fills differ from intended levels?
Eligibility: Were trades taken only when the strategy’s conditions were satisfied?

This step often produces a structured set of flags (e.g., “rule followed,” “rule violated,” “fill deviated,” “trade taken when condition not met”). The output is not a signal; it is a factual breakdown.

4) Compute outputs that explain the “why,” not just the “what”

After comparing plan versus execution, compute summary outputs that describe patterns in outcomes.

Common output types include:

Adherence metrics: proportions of trades where rules were followed.
Cost sensitivity: how much transaction costs influenced net results, compared to gross moves.
Deviation analysis: links between execution differences (timing/slippage) and outcomes.
Condition breakdowns: if the strategy had different assumptions under different regimes, summarize results by those regimes.

A well-formed review keeps calculations explicit. For example, if a metric requires converting values into a common unit, define the conversion rule and state whether results depend on that choice.

5) Formulate hypotheses for rule changes as testable statements

The review should end with hypotheses written so they can be tested later. Examples are phrased as “if we change X, then we expect Y under the same scope.”

This keeps the process informational. It does not imply guaranteed improvement; it simply defines what would be checked in a subsequent review.

Example of a worked comparison (with clear assumptions)

Consider a simplified strategy description:

The plan says: “When condition A is present, place a limit order and exit when condition B occurs.”
The plan also states: “Transaction costs are included in net profit calculations.”

A review for a given date range might proceed like this:

Collect trade logs for all instruments where condition A was believed to be present, using the strategy’s recorded timestamps.
Record intended versus filled prices for each order. The assumption here is that the log contains both the planned level and the fill level.
Compute gross movement as exit fill price minus entry fill price for long positions, and the reverse for short positions.
Compute net movement by subtracting explicit costs (such as commissions) and representing spread impact using the logged spread or an equivalent documented method.

The review output could include:

The share of trades where the entry rule was actually eligible and where the limit order filled as intended.
The average deviation between planned and filled entry price.
A summary of net results after costs.

Material limitations arise if any assumption fails. For instance, if “condition A was present” is inferred rather than recorded, or if spread is not consistently logged, the computed comparisons may be misleading.

Limitations and common failure modes

Strategy review can be valuable, but several limitations often undermine it.

Market and context variability

Historical relationships do not ensure future similarity. Even if execution quality is consistent, market microstructure and volatility regimes can shift, changing how strategy rules behave. Strategy review should therefore describe the period reviewed and avoid treating past outcomes as stable promises.

Data and measurement problems

Failure modes include:

Incomplete trade logs (missing fills, missing partial fill details, inconsistent timestamps).
Inconsistent cost representation (spread treated differently across periods).
Ambiguous rule definitions (rules that cannot be applied consistently to past data).

If a reviewer cannot reproduce the same outputs from the same inputs, the review is not verifiable.

Hindsight bias

A reviewer might unintentionally interpret outcomes in a way that fits expectations. This can happen when the “reason” for a deviation is defined after seeing results. To reduce this, record decision rules and deviation categories in advance, or at least document them before computing metrics.

Cost and execution constraints

Many strategies appear profitable before accounting for real transaction costs and execution differences. If the review excludes costs or assumes ideal fills, it may overestimate net performance. Conversely, if costs are double-counted or measured inaccurately, it may underestimate results.