THE ST EDGE FRAMEWORK

How the numbers get judged

The ST EDGE Assessment starts from the closed days, computes 15 metrics and compares them with thresholds written in advance and fitted to the declared way of trading. Out come a 0 to 100 score, how precise it is and a position on the scale for the operator type. The method is the same for a year, a quarter, a month or any two dates.

The Metrics tab: the score with its interval strip, the four groups, warning signs and strengths.

Next to the score there's an interval showing where it would land over as many days traded of the same kind. ST EDGE computes it by resampling the days hundreds of times and redoing the calculation each time. Its width matters less than whether it stays within one rating band: if it spans three or more, the verdict itself isn't settled yet. With 300 resamples the interval is 90%, from 500 up it is 95%. Below 20 days the interval isn't computed, and the screen says so.

The percentile isn't a count of how many people were beaten: it shows how high the result sits on a scale calibrated for the operator type and style. It isn't a grade, and on the same period it can say something different from the score, because it answers another question. Alongside it come sensitivity to order (if the days are reshuffled, does the result hold?), diagnostics that read the metrics in pairs, unexpressed potential and a short verdict. Past the last mark the scale stops and says so, instead of inventing precision it doesn't have.

Over several years two different questions can be asked. The single block shows what the track record is worth as a whole. The year-by-year reading assesses each year on its own capital and then builds a verdict from them, so the gap between the worst and best year becomes a figure in its own right. Two records with the same total can have years that all look alike, or one year carrying the rest, and the cumulative view can't show that difference. At least three years are needed.

It puts everything shown on screen into one document: score with band and interval, metrics with their bands, positioning, diagnostics, verdict and an appendix explaining each calculation. Fixed rules produce it, with no AI involved, so the same data and the same version give an identical document, as the reproducibility seal attests. Before generating it, the sections, the operator sheet rows and amount redaction are chosen; optionally, the data fingerprint can be added under the seal.

ST EDGE Assessment: score, interval and four groups

Analytical Agent: composite percentile, positioning and radar chart

The year-by-year reading: each year on its own capital, then a single verdict

Before the Assessment PDF: sections, operator card, amounts, fingerprint

A click on the screen opens it full size.

HOW THE VERDICT IS BUILT

From the days to a verdict, in five steps

01DaysEach row in the archive is one closed day. The days are what gets judged, so more detail on single trades doesn't change the grade.

0215 metricsComputed on the chosen period, on an annual basis (√252) or on trading days only (√N).

03ThresholdsEach metric is compared with written thresholds, some of them fitted to the declared way of trading.

04ScoreThe metrics vote in four groups with different weights. Out comes a number from 0 to 100 with its rating band.

05Precision and positionResampling the days measures how stable the score is. The Analytical Agent places it on a percentile scale.

THE 15 METRICS

What the framework measures, group by group

Each group answers one question and weighs differently on the score: Risk and Return / risk count more, and the exact weights aren't published. No metric is enough on its own: a high return with a huge drawdown, or a very high win rate with rare but heavy losses, are fragile profiles that a single row doesn't show. That's why the score weighs them together.

lower is bettercloser to 0 is betteradaptive thresholdfrom 30 daysfrom 60 daysat least a year

Performance

«How much did the capital return?»

Annual net return %
What the capital returned over a year, with every cost already taken out. It is the only metric in the group.

Risk

«How much risk did the result cost?»

Max Drawdownlower is betteradaptive threshold
The largest loss from the highest point reached, measured close to close. Declared leverage tightens the threshold.
Ulcer Indexlower is betteradaptive thresholdfrom 60 days
How far the equity fell and how long it stayed down. Max Drawdown is a single episode, this looks at the whole path.
Daily volatilitylower is betteradaptive threshold
How much results swing from one day to the next. It doesn't say whether money is being made, it says how bumpy the path is.
VaR 95%closer to 0 is betteradaptive threshold
Where the bad days begin: one day in twenty went worse than this.
Skewnessfrom 60 days
Which way surprises lean. Positive means the big moves are gains; negative means many small wins and a few very bad days.
Kurtosis (excess)lower is betterfrom 60 days
How often extreme days happen, either way.

Return / risk

«Was the gain worth the risk?»

Sharpe Ratiofrom 30 days
Return per unit of swing, up and down alike. The higher it is, the smoother the path to the gain.
Sortino Ratiofrom 30 days
Like Sharpe, but it only counts downward swings: very good days aren't a risk.
Calmar Ratioat least a year
Annual growth measured against Max Drawdown. It is judged only on a period of at least one year that has already closed.
Daily profit factor
How much is earned for every unit of currency lost, counted by day. 2 means twice as much.

Consistency

«Did it happen regularly?»

Daily win rateadaptive threshold
The share of days closed in profit. Days, not trades: a winning day can contain many losing trades.
% positive weeksadaptive threshold
The same question over weeks: trading weeks, or Monday to Sunday for crypto.
% positive monthsadaptive threshold
The same question over months. It tells steady earners from occasional ones.
Max consecutive wins
The longest run of winning days in a row.

THE RATING BANDS

Criticalbelow 40

Weak40–55

Sufficient55–65

Good65–75

Very good75–85

Excellent85+

What the score isn't. It isn't a forecast of future returns, it doesn't certify the data and it isn't a ranking among traders: it's a summary of the 15 metrics over the observed period, to be read together with the rows behind it and its precision interval.

Thresholds are not the same for everyone. The market, style and leverage that are set move the risk and regularity thresholds: a drawdown that's normal on futures isn't normal on bonds, and leverage tightens the limits on Max Drawdown and Ulcer Index. Ratios, return and costs use the same thresholds for everyone. Each row shows the value, the metric's 0 to 100 score, the band reached and the direction («higher is better», «lower is better»). The numerical band limits are part of the methodology and aren't published.

Costs don't vote. The net return has already taken them out, and under Performance a line of facts shows gross, cost weight and net. Sharpe, Sortino and volatility follow the calculation basis chosen: annual (√252, or 365 days for crypto) or recorded days only (√N). On short periods the return is scaled to a year in proportion to the days, only so the same thresholds apply; the report always prints the period's real return.

WHEN «N/A» APPEARS

A metric that abstains

When a metric can't be defined, ST EDGE doesn't print a meaningless number. It shows n/a, the reason and the current value. The metric drops out of the score and its weight moves to the others, so the grade isn't pulled down where there isn't enough data.

Sharpe RatioIf every day returned the same amount, there’s no fluctuation and so nothing to measure.

Sortino RatioIf no day ever closed at a loss. Writing «zero» would be wrong, because it would give a bad grade to someone who never lost.

Calmar RatioOn a period shorter than a year, on a year still running, if the period closes at a loss or if there was never a loss.

With only a few days, the number is shown but doesn't count yet. Sharpe and Sortino count towards the score from 30 recorded days; Ulcer Index, skewness and kurtosis from 60. Until then the row shows the current value and when it will start to count, since on so few days it would have no statistical meaning.

GOOD TO KNOW

The full Assessment is available from the Premium plan; on Base the Dashboard shows a summary of it. During the beta it's open to everyone.