Compare trades

Compare answers where does performance change across the selected trades, and under which conditions? It recalculates statistics independently for each group so you can move from a broad observation to a testable journal question.

Compare is available with Pro. For a product-level introduction, see Trade Comparisons.

Start with a focused selection

All normal trade filters are applied before groups are constructed. The fixed All selected trades baseline is calculated directly from every distinct matching trade and remains visible above the comparison results.

The baseline is not the sum of visible rows. It is the reference population against which each group can be understood. Sorting, searching, collapsing children, or changing the chart does not alter it.

Begin with a question such as “Does my breakout strategy behave differently by market?” rather than selecting many dimensions and metrics without a hypothesis.

Comparison dimensions

Personal comparisons support:

• Account
• Strategy
• Direction
• Symbol
• Market
• Tag
• Overall self-rating
• Hour of day
• Weekday
• Session
• Time by day, ISO week, month, quarter, or year

Trading-group comparisons can also use User, subject to membership and sharing permissions.

Missing values remain explicit groups such as No Strategy, No Tags, Unrated, or Not set. Legacy invalid Overall ratings are not rounded into a valid star group.

Open and close time

Hour, Weekday, and Time can attribute trades by the Open or Close timestamp. Close is the normal default because it associates a result with when it was realized. Open groups eventual outcomes into cohorts based on when trades were entered.

Hour and Weekday use the requested timezone. Sessions use the product's fixed UTC Sydney, Tokyo, London, and New York windows.

Add a second dimension

A split creates intersections between a primary and secondary dimension. Examples include:

• Strategy split by Market
• Symbol split by Session
• Tag split by Direction
• Weekday split by Strategy
• Overall rating split by Account

The primary parent is calculated from its distinct contributing trades. It is never derived by summing child values.

This matters because one trade can have multiple tags and can overlap multiple session windows. Those children overlap, so summing them would double-count trades. Ratios, averages, compounded percentages, and scores are also non-additive even when groups are mutually exclusive.

Independent percentage compounding

Every comparison group geometrically links its own saved trade percentage returns:

group compounded return =
    (product(1 + group trade return / 100) - 1) * 100

Group percentage returns must not be added or presented as contributions to the baseline. For example, two independently compounded strategy groups do not necessarily sum to the overall selected-trade percentage.

Profit and R net returns are additive within a group, but parent and child construction rules still prevent blindly summing overlapping groups.

Choose table metrics and a chart metric

The selected table columns and the active chart metric are separate:

• Table columns define the ordered statistics shown for every group.
• One active metric drives the chart and primary result sorting.
• Activating a metric not already in the table adds it.
• Removing the active metric clears the chart selection while leaving other columns visible.

Quick comparisons normally begin with Trades, Net R, win rate, trade expectancy, and profit factor. Together these provide sample size, total production, outcome frequency, per-trade edge, and payoff efficiency.

Always keep a sample-size or coverage metric visible when comparing sophisticated statistics. A striking ratio from five trades should not look equivalent to one supported by hundreds.

Available metric families

Compare includes most closed-trade statistics from the other Stats pages:

• Counts and net return
• Win rate, trade expectancy, profit factor, and average win/loss ratio
• Break-even win rate and break-even reward:risk
• Averages, extreme trades, and standard deviation
• Gross result, costs, and named adjustments
• Durations
• Maximum and average drawdown and run-up
• Recovery factor
• Ratings and rating coverage
• Planned R, Actual R, and plan deviation
• Execution efficiency and excursions
• Outcome streaks
• Loss containment
• Sharpe, Sortino, Calmar, volatility, and downside deviation
• Return smoothness and Monthly Return Reliability
• Trading Mastery Score and its components

Planning coverage and direction-prediction accuracy are intentionally excluded. They require invalid or other non-executed planned opportunities, while Compare has a closed executed-trade population. Including them would silently change the denominator for those rows.

Availability belongs to each cell

Every metric variant in every group retains its own availability and evidence. A row is not simply “high evidence” or “low evidence.” In one strategy row:

• Net R may be available.
• Profit factor may be unavailable without negative outcomes.
• Sharpe may be limited by the number of completed months.
• Trading Mastery Score may be unavailable because one required component is missing.

Low-trade groups remain visible. Review the marker or information for the affected cell rather than interpreting an unavailable value as zero.

Monthly and path-dependent statistics are recalculated for each distinct group. Explicit Close-date boundaries are common to the groups. When a window must be inferred, each group uses the span where its selected trades closed. This avoids penalizing a strategy for time before it existed, but groups with materially different spans are less directly comparable. Their effective dates and evidence remain visible for that reason.

Mixed currencies and shared groups

Monetary inputs are converted into the resolved reporting currency before the baseline and groups are calculated. Conversion is all-or-nothing for the response, so a monetary metric never mixes converted and unconverted values.

Trading groups apply membership, visibility, and each contributor's sharing permissions before calculation. Hidden accounts can be combined under Account not shared, and unshared tags are not exposed. Profit, percentage, review, or concurrent-risk inputs may be removed when the corresponding permission is disabled.

Visualizations

Bars and grouped bars

Bars are effective for one categorical dimension. Grouped bars are useful when a modest number of split combinations needs direct comparison. Avoid treating overlapping Tag or Session bars as parts of a whole.

Timeline

Timeline is the natural view for chronological primary groups or a manageable chronological split. Choose Open or Close attribution based on whether your question concerns entry cohorts or realized results.

Values grid and heatmap

The Values grid prioritizes exact numbers, empty cells, and evidence. Heatmap prioritizes patterns through colour. Use the grid when precise values or missingness matter. Use the heatmap to find clusters worth investigating. Colour strength is not a substitute for checking scale, direction, and sample size.

Large split comparisons default to a grid-based view because crowded grouped bars would hide values and empty intersections.

Break-even map

The specialized Break-even map plots each group's R-based average win/loss ratio against decisive win rate:

decisive win rate = W / (W + L)
break-even win rate = 1 / (payoff ratio + 1)

Break-even trades are excluded from the plotted rate even if your normal win rate setting counts them as a win or loss. This keeps every point mathematically comparable with the two-outcome boundary. Point size represents group trade count, and groups without an available payoff ratio cannot be plotted.

Result size and specificity

A comparison can contain up to 2,000 primary and child nodes. If a request is too large, narrow the filters or use broader dimensions.

Even below that limit, more detail is not always more insight. Highly specific intersections often produce tiny samples. Start with one dimension, identify a pattern, add one split to test it, and then inspect the underlying trades.

A practical comparison workflow

1. Define a question and apply the relevant trade filters.
2. Choose one primary dimension.
3. Keep Trades visible and add a small set of complementary metrics.
4. Compare rows with the All selected trades baseline.
5. Check each important cell's evidence and effective span.
6. Add a second dimension only when it tests a specific explanation.
7. Choose the visualization that matches the question.
8. Open or filter the contributing trades before acting on the pattern.
9. Repeat the analysis after more data to see whether the relationship persists.

Comparison reveals associations in the journal. It does not establish causation, remove regime effects, or guarantee that the strongest historical group will remain strongest.