Everything worth knowing before getting started

A tool that measures trading performance. It takes the record of what has been done and returns a score, a position on a reference scale, a diagnosis, and a set of reports that can be handed to a partner, an allocator, a bank or an auditor.

No, to all three. ST EDGE doesn't connect to any broker, doesn't receive quotes, doesn't generate trading signals and doesn't make predictions. It analyzes only what the user records or imports. It isn't a trading system; it's used after trading, to see how things are actually going.

It does record the data, but that's not really what it's for. A journal is for logging; ST EDGE is for judging. It uses stated thresholds, a comparison scale calibrated to the trader's style, and a diagnosis that looks at how the metrics relate to each other rather than listing them one by one.

Excel will always produce a number. If the sample is too short, if a ratio's denominator is close to zero, if the period isn't long enough to judge a metric, a number still comes out, and whoever reads it can't tell a solid result from one the spreadsheet produced because it had to produce something.

ST EDGE works the other way round: when a metric can't be calculated in a defensible way, it leaves it out and explains why. The comparison scale also changes based on the declared trading style, every report shows the software version that generated it, and each threshold comes with its source.

The analysis engine doesn't. Score, percentile, diagnosis, Monte Carlo and accounting reports all run on fixed rules and thresholds, so the same data always gives the same result.

There's a separate, optional feature, AI Analyst, that uses a language model to comment on the numbers in plain words. It's the only one that does, it needs the user's own key with any preferred provider, and its report states on the cover that it isn't reproducible. The two are always kept apart.

Anyone who wants to know how they're actually trading, rather than how they think they're trading. That means private traders, prop traders who need to keep risk and drawdowns in check, strategy developers comparing backtests, simulations and live trading with the same metrics, and anyone who has to present results to an investor, a partner, a prop firm or a bank.

No. Archives stay on the user's computer. The only things that go out are:

• the account credentials (the email; the password is never sent in the clear), to verify the license;
• an aggregated usage count, meaning which features were used and how many times a day, when the app is online.
• only if the AI Analyst is used with an external provider, the summary of statistics described in the question on AI.

The numbers and the rows in the archives never leave.

With a model that runs locally on the computer (Ollama), nothing leaves it and no key is needed. With other providers, the chosen one receives a bundle of pre-computed statistics (return, drawdown, Sharpe, win rate and the like) together with the configured trading profile. The raw rows stay on the machine: no days, no trades, no orders. The model never sees the archive, only a numerical summary of it.

The API key is the user's own. ST EDGE doesn't come with one and doesn't sit between the user and the provider.

Mostly, yes. Assessment, Monte Carlo, reports and import send nothing to anyone and work without a network.

The app does check the license at startup and then once an hour. If the connection drops, work can continue for up to 24 hours: the light at the bottom left turns amber and shows how many hours are left, and a warning appears an hour before time runs out. After 24 hours the app stops until the connection is back. Archives stay where they are and nothing is deleted.

This grace period is there so a brief outage doesn't interrupt work, not for using the app offline all the time. It's a real limit, so it's better to know about it upfront.

Yes. A local lock can be set with a dedicated password, separate from the account password. The app asks for it at startup, and it works offline too. ST EDGE only stores a cryptographic fingerprint of it, never the password itself, and never sends it to any server.

There are two things to know before turning it on:

• It can't be recovered. There's no "forgot password": the lock can only be removed from inside the app, which means after the password has been entered.
• It's not a vault. It stops people from using the app, but it doesn't encrypt the database file, so anyone who opens the file with another program can still read the data. That's also the way out if the password is forgotten: the data isn't lost, but getting back in means working on the file from outside the app.

No, it applies to the active profile, because it's stored inside that archive. With more than one profile, it has to be set on each of them. Protecting one doesn't protect the others.

The database belongs to the trader, not to ST EDGE. The .db file isn't encrypted, and that's a deliberate choice: it's an open format readable with any SQL tool, with no keys or access controls held by ST EDGE.

That way the history stays with its owner, who can still open and read it on the day they stop using ST EDGE. It isn't tied to the software, and it won't become unusable if the service shuts down.

Looking after the file is up to whoever owns it. ST EDGE guarantees that the data doesn't leave the machine, but it can't guarantee the machine itself is secure. Keeping the computer, operating system, backups and access to the file secure is the responsibility of the device's owner.

Just two things: the capital and daily data.

Capital is set year by year, and it's the only figure ST EDGE can't work out from the numbers: trades show how much was made, not how much capital it was made on. Without it, percentages can't be calculated and show as "—".

For the data, a summary of each day is enough (fills, gross, commissions, net). Everything else is optional.

The app opens on the demonstration profile, the same one shown in the manual's screenshots, with test data, read-only: it doesn't count toward the plan's profile limit and can be deleted when no longer needed. For one's own data, the default profile is selected from the Active database menu.

ST EDGE doesn't depend on any trading platform: it analyzes data exported from any broker or platform, as long as the required fields are there.

In practice it works with any broker that exports a CSV or text file, MetaTrader included. MetaTrader's closed-trade history goes in through the Trades import, with swap, taxes and fees if the file has them. There's no special format to follow: each column in the file is matched to a field, and the mapping is remembered, so the next export from the same broker is already set up.

It detects the separator and the date format (European or American) on its own, and if the values are ambiguous they can still be fixed by hand.

Three, and they can live side by side. The choice depends on what the broker provides:

Level	A row is…	Comes from
Daily	a day of trading	filled in by hand, or imported
Operations	a closed trade (entry, exit, direction, size)	the broker's trade statement
Transactions	a single filled order	the broker's raw order tape

Transactions never go straight into the daily record. ST EDGE first matches them in chronological order to rebuild the trades, then builds the days from those trades. It happens in two steps.

No. The engine always analyzes the daily record, whatever level of detail the data came in at, so importing trades or orders won't change the score.

There are two concrete gains, though. One is verifiability: a day built from orders can be reconciled with the broker statement, and the report shows where it came from. The other is Per-Trade Analysis, a view that measures trades one by one: win rate, profit factor, duration, which instruments and which hours do best. Daily data simply can't provide those figures.

The engine was validated by hand against real broker statements, reconciling every trade by ID and size.

It also works when the history comes in several separate imports rather than one file. A position opened today and closed two weeks later, imported day by day, is still recognized as a single trade. This was checked by comparing incremental imports with a rebuild of the whole archive in one go, and the result matched trade for trade.

It can be fixed. Each level has its own view in Records & Positions: days are edited with the pencil icon and deleted with the trash icon, and trades and transactions work the same way. "Delete period" follows whatever filters are set.

During an import, nothing is written to disk until the import button is pressed. The data can be reviewed, the columns mapped and the preview checked, starting over as many times as needed.

Yes to both. Each profile is a separate archive (a real account and a demo one, different accounts, a personal account and someone else's) with its own currency (twenty-two are supported), its own costs and its own report headers. Only one is active at a time, and the bar at the top of the menu always shows which one is in use. How many profiles can be opened depends on the plan: 1 with Free, 3 with Base, 9 with Premium, unlimited with Pro.

Several profiles can also be combined into a consolidated one, a portfolio with its own statistics, Assessment and reports that shows each system's weight and contribution. The ST EDGE Composer tries out combinations of the profiles, suggests which ones to group together and tests them on the following years. Both are in the Pro and EDU plans.

The archives live on this computer, in the program's «data» folder. To use a profile on another computer, it is saved with Export DB and put back into use there with Restore backup…, in the Database screen.

Four kinds of document, each with its own job:

• The Assessment judges, and it's the only one that gives a rating.
• Accounting reports disclose how a period closed (annual, quarterly or cumulative), with no score and no benchmark.
• Monte Carlo explores how much the result depended on the order in which things happened.
• AI Analyst comments on the numbers. It's optional, and it says openly that it isn't deterministic.

For anyone with several accounts there's also the Composer, which combines: it suggests which systems to put together in a portfolio and tests them on the years after the choice was made.

It comes from 15 metrics split into four groups (Performance, Risk, Return / risk, Consistency). Each one is compared against thresholds, weighted and combined with the others. The groups don't carry the same weight, and the panel under the score shows each group's score from 0 to 100, so it's clear where points were lost. Risk and Return / risk weigh the most; the exact weights are part of the methodology and aren't published.

Next to the number there's a confidence interval (90% with 300 resamples, 95% from 500 up), so the score comes with its uncertainty attached. The more days of data there are, the narrower it gets.

Because ST EDGE has chosen not to calculate it. A ratio with no denominator (no losing days, no drawdown, too short a sample) isn't zero; it's simply undefined.

In those cases ST EDGE writes the reason next to it and gives that metric zero weight, rather than making up a number that would pull the score down exactly where the data is strongest. The whole app works this way, and everything else is built on that principle.

Because the comparison scale is calibrated on the combination of the trader type that is set (retail, prop trader, institutional, family office, market maker) and the style (scalping, day trading, swing, trend following, market making, order flow, mean reversion).

A 25% drawdown is a serious problem for a scalper and perfectly normal for a trend follower, so measuring very different kinds of trading with the same yardstick would give a meaningless verdict. Likewise, a Sharpe of 1.8 is excellent for a prop-firm scalper and modest for a fund.

In practice, changing these settings changes the yardstick, not the numbers. If a ranking looks odd, those settings are the first thing to check.

Yes. The Assessment is fully rule-based and deterministic: the same data and the same program version always produce the same document. The PDF says so with a seal on the cover. If the data fingerprint is turned on, it also lists the days it was built from and the version that produced it, so whoever receives it can check it came from the same data.

The only exception is AI Analyst, which states clearly that it isn't reproducible.

No. Reshuffling the days doesn't reveal anything new about the future; it shows the variability already in the history.

That's why the labels say "share of paths" and never "probability of profit". It's a count of what has been done, not an estimate of what will happen. For the same reason Monte Carlo doesn't feed into the score and is read separately.

Note: with only a few days, or if they all come from one market regime, the range shown is narrow because the sample is narrow, not because the risk is small.

A year, a quarter, a month, a cumulative range of years, or any two dates. Assessment, Monte Carlo and AI Analyst all use the same periods, since it's the same engine looking at a different window.

On a single month, many metrics will be marked as not assessable. That's not a software limitation, just the same abstention rule described above.

Because an average can hide dangerous trade-offs, like strong returns covering up risk that's out of control. The score comes from combinatorial scoring instead, where metrics are weighted and combined rather than added up. That way imbalances between the four groups are penalized, and a high score only comes when the system holds up on every front at once.

Yes, that's what the accounting reports are for. They present the data without judging it, show where it comes from, and include a technical appendix with the formula behind every number instead of a vague reference to the literature.

There are three of them, one for each kind of period (Annual, Quarterly and Cumulative), and each section can be switched on or off. Consolidated profiles also get a Portfolio systems section, with each account's weight and contribution.

Yes, at three levels:

• field by field in the trader profile: the eye icon removes the row from the report, while the toggle keeps it but shows "redacted" instead of the value;
• all amounts at once, with a single switch;
• the Public excerpt: ST EDGE builds the full document and shows only the first and last pages. The page numbers stay the original ones and the gaps are visible ("page 4 of 19", then "page 17 of 19"), so it's clear the document is reduced and nobody mistakes it for the complete one.

No. ST EDGE doesn't certify anything; it reports what the user declares.

It asks for an explicit statement of what kind of data it is (real trading, a test environment, or not specified). If it's a test environment, it prints a red band on every page, because a PDF can get split up and even a single sheet should make clear what it contains.

If nothing is stated, the report stays silent on the point rather than claiming something it was never told.

No. When generation finishes, a preview of every page appears. It can be paged through and checked, and the file is only written when "Save PDF" is pressed. If the preview is closed, nothing is left on disk.

Accounting summary, multi-year history, quarterly and monthly trends (tables, bars, return and drawdown heatmaps), full metrics table, equity & drawdown, length of underwater periods and time to recover, return distribution, autocorrelation, rolling analysis, the Monte Carlo test, a per-trade analysis, a data integrity check, and a technical appendix that typesets every formula used.

Each section has its own checkbox, and the ones not needed stay out of the document.

The first is produced by rules and is reproducible: same data, same document. The second is written by a language model and isn't reproducible. When it's regenerated the text may change, because no provider guarantees an identical answer.

The two never get mixed, and each says which one it is on the cover. For a repeatable verdict, the first is the one to use.

Yes. Reports come out in the language selected in the app, Italian or English. That covers the Assessment, accounting reports, Monte Carlo and AI Analyst, including the technical appendix and disclaimers.

No, and the difference matters. ST EDGE rebuilds the equity curve from daily results, so the drawdown it shows is measured from one close to the next. The loss actually taken during the day is bound to be deeper.

Prop firms measure the limit intraday, so the two numbers can't be compared. ST EDGE says this upfront rather than leaving it to be found out the hard way.

The same goes for Per-Trade Analysis: the drawdown is measured between closed trades, not inside each one. A trade opened at 100 that dips to 95 and closes at 101 shows as +1, and how far it dipped while open isn't recorded in any archive.

No. ST EDGE doesn't know market prices and doesn't pretend to. For open positions it shows their carrying price and how many days they've been open, but never a current value or an unrealized gain.

It's in beta, and the software says so. The version appears when hovering over the license light; with the data fingerprint turned on, the document shows it too. Some areas are finished but haven't yet been tested through extensive live use.

Some have a verifiable primary source; others are industry conventions, and are presented as such. ST EDGE publishes what its scale is calibrated on and doesn't claim a source for a number when there isn't one.

For the same reason the comparison is called the "ST EDGE evaluation scale" and not "the world's top 1%": nobody has ever observed that population, so anyone quoting it is using a number they don't have.

Taxes and financing, no: ST EDGE measures performance based on the data it receives. Commissions are either read from the imported file or calculated from the per-fill rate that has been set (by period too, if the rate has changed over time), and when it calculates them it always says so.

Swap, fees and taxes can be recorded, though: they come from the imported file or can be entered manually on the day, and net becomes gross + swap − commissions − fees − taxes. Anything not recorded simply isn't there, because ST EDGE never adds it behind the scenes.

Yes, it's required. It takes an email and password, a short optional profile (country, trader type, years of experience, style and markets) and two separate consents, one to the Terms of Service and one to the Privacy Policy. Each can only be ticked after the document has been opened. The email is verified with a six-digit code before the first login.

During the beta everything is free: until 31 March 2027 every feature is available and every account has the Pro plan. Paid plans will start after the beta, and for now they can't be purchased. From 1 April 2027, beta accounts move to the Free plan unless a paid plan is chosen: nothing is charged automatically and the data stays where it is. The academic plan doesn't move to Free (details below).

There will be four plans, each including everything in the one below it: Free (free, 1 profile), Base (€9.90/month or €99.00/year, up to 3 profiles), Premium (€19.90/month or €199.90/year, up to 9 profiles) and Pro (€39.90/month or €399.00/year, unlimited profiles, consolidated profiles and Composer). The full comparison is in the Plans and pricing section. Moving to a lower plan doesn't mean losing anything, and the data stays where it is.

Academia: professors, researchers and students can use the Pro plan for free by signing up with their institutional email address. If the domain is one of the recognized ones, the plan activates automatically; if not, it can be reported through the Contact page and will be added before activation. The plan expires every year on 31 December and doesn't renew on its own: when it expires the app shows «EDU licence expired»; pressing «Reactivate EDU» sends a 6-digit code by email, and confirming it there renews the plan for another year. If the domain is no longer academic, the plan can't be reactivated. For any problem, the Contact page is the place to write.

No. The account is separate from the database profiles, so switching profiles or moving the installation to another computer doesn't mean registering again. Login stays the same; only the data being analyzed changes.

Italian and English, with the same features in both. Numbers and dates follow the conventions of the selected language.

Yes, in two forms with the same content. Inside the app there's the guide, with a walkthrough on first launch, and contextual help: the "?" button at the top opens the card for the current screen. The PDF user manual is the printable version of those same texts. It's generated from the help cards, so it says what appears in the app and gets updated along with them.

User manual download (PDF)

The public beta is available for Windows 64-bit and for Macs with Apple silicon, M1 or later. It's a desktop application that works locally and needs no browser.

That year's capital is missing. It's set in Active DB Settings › Starting capital per year. It's the only figure ST EDGE can't work out from the trades, and rather than calculate on a made-up value it flags it as missing.

Market and style haven't been set in Active DB Settings › Strategy. Nothing breaks: ST EDGE uses a generic scale and flags it, but the comparison is less reliable.

Almost always it's because ST EDGE doesn't know which value in the column means "buy" and which means "sell". It has to be told, by entering the two strings to look for (B/S, BUY/SELL or whatever the broker uses).

Discarded rows don't just disappear: they're listed in a panel with the reason, and the field that doesn't match is highlighted.

It's still there, in the other profile. Switching profiles changes more than the days: capital per year, currency, costs, report headers, the strategy that was set and the detail archives are all per profile. That's intentional, because two accounts shouldn't share anything.

They're separate archives, and the engine reads the days. Once the detail is fixed, the day has to be rebuilt from it, or the analysis will keep using the old value.