Skip to content
Guides
FROpen Studio

Test and understand

Understand metrics

Connect each number to a concrete decision for your next experiment.

4 min read · 7 sections
Contents of this guide

Choose the right study view

Study views answer complementary questions. Begin with the baseline report and use the view suited to your next question. A comment or score does not replace operations and assumptions.

ViewMain questionEvidence to seek
BacktestWhat operations do these rules produce on this history?Test report and journal
AnalyticsHow is the result distributed?Available periods, groups and contributions
AI LabWhich variant deserves comparison?Experiment, difference and trade-off
AuditWhich reservations need checking?Review points and missing evidence
Stress testWhat changes under less favourable assumptions?Difference from baseline and sensitivity

Net result and return

Net result is a gain or loss in the indicated currency after the costs included in the report. Return relates an outcome to a capital reference. Check that reference before comparing percentages.

Text
Teaching example: initial capital 10,000 and net result +500.
Relative to initial capital: 500 / 10,000 × 100 = +5%.
Moving from +5% to +7% is an increase of 2 percentage points.
It is not merely a 2% increase in the previous return.

Currency outcomes from very different account sizes are not directly comparable. Percentages also need context when exposures or risks differ substantially.

Win rate, average wins and average losses

Win rate describes the proportion of winning trades according to the report’s convention. Read it alongside average gain and average loss. Frequent wins can still be outweighed by occasional large losses.

Expectancy per trade summarises the average result in a sample. In a simplified example with 40% wins averaging 120 and 60% losses averaging 60, the sample average is 0.40 × 120 − 0.60 × 60 = 12, before any costs not already included.

That average does not describe the order of outcomes or remove uncertainty about the next trade. Inspect sample size and extreme values too.

Profit factor and unavailable values

The usual profit factor definition divides the sum of positive outcomes by the absolute sum of negative outcomes. Total gains of 1,200 against losses of 800 give 1.5. Check the displayed metric’s scope to understand which outcomes and costs it includes.

Drawdown and the capital path

Drawdown describes a decline from a previous peak of the relevant curve. Depth and duration convey different information: a sharp brief loss and a long stagnation can be difficult for different reasons.

Text
Example: peak at 12,000 followed by a low at 10,800.
Decline: 1,200; relative to the peak: 1,200 / 12,000 = 10%.
A 10% decline then needs approximately 11.1% to regain the peak.

Check whether the measure uses closed balance or includes open positions. Worst trade and worst drawdown differ: several losses in sequence can create a larger decline than any one trade.

Distribution and concentration

Inspect available breakdowns by period, direction or group. Compare amounts and observation counts. A group with three trades supports less confidence than a much larger sample, even when its apparent return is higher.

Ask whether one week or a few trades explains most of the result. Segmentation should generate a question; removing every losing segment after seeing the data can overfit the strategy to that history.

Turn an observation into an experiment

“Losses are concentrated in this window” can be a measured observation. “This window is always unfavourable” goes beyond the sample.

Example prompt
Using the available metrics, separate:
1. numerical observations and their sample sizes;
2. possible explanations, without presenting them as established facts;
3. one experiment to compare with the baseline on another period.
Do not fill missing metrics with assumed values.

Search the guides