ORBIT
Inside the platform

Eighteen charts, a copilot with hands, and an interface made of glass.

This is the actual application — every image on this page is a screenshot, not a mockup. Here is what is in it, why each piece exists, and how the whole thing is put together.

01 · The cockpit

Up to eighteen live charts. Each one its own trading desk.

Most platforms give you one account and many windows. ORBIT inverts it: every chart carries its own account, its own risk rails and its own bot. Run the same strategy on three timeframes side by side, or three different opening-range windows against the same instrument, and watch which configuration is actually working — on the same screen, at the same time, on the same tape.

The ORBIT cockpit: three live NQ charts side by side, each with its own account panel, order ticket and volume profile, with a combined account rail on the right. CLICK TO ENLARGE
The cockpit. Three independent charts, each with its own paper account, order ticket, risk rails and bot — and an account rail on the right tracking every book at once.
18

Charts on one screen

Arranged in any layout you like, saved between sessions. Link charts with a click of an LED and they share a wallet, a symbol and settings; unlink and they go their own way.

1 : 1

Chart to account

Paper or live, chosen per chart. A chart armed on paper cannot reach your broker at all — not a dry run, not a demo endpoint. The block lives in the engine, not in a badge.

6

Opening-range windows

Sydney, Tokyo, Frankfurt, London, New York, Globex — each drawn on the chart as it forms, each with its own trade budget, each able to run a different configuration of the same bot.

ORBIT's top status bar showing feed health, session countdowns, ATR, volume and the ORBIT score. CLICK TO ENLARGE
The status bar never stops talking. Feed health and messages per second, which session is open and how long until the next one, today's ATR, volume, range and average volume — and how many signals each window has fired today.
Two ORBIT charts side by side on the same instrument at different timeframes, showing opening range levels, value area and volume profile. CLICK TO ENLARGE
Same instrument, two lenses. The 5-minute on the left for structure, the 1-minute on the right for the entry — the opening range, value area and volume profile drawn identically on both, so your eye never has to re-orient.
02 · The copilot

Just say… “Hey ORBIT”

ORBIT is a voice-native assistant that can see your whole screen — not just the chart in front of you, but every window you have open, and the relationship between them. The levels, the profile, your position, your working orders, your risk budget. Ask her out loud and she answers out loud, so your eyes never leave the tape. She will place an order when you tell her to, call your fills, and tell you plainly when she does not know.

She also knows the books. Every account on the desk, what each one is carrying, and the P&L of the book behind whichever chart you are asking about — so “how is that one doing?” is a question with an actual answer rather than an invitation to go and look it up yourself.

TAP HER — SHE’LL SPEAK

She reads the chart, not a script

“The range formed at 20:05 and broke above 29529.25 at 22:49. You're now in the waiting phase — price is below the break point, and the retest window is open for another 55 minutes. That is the setup sequence working as designed.”

“Price is at 29522.75, inside the range. The visible profile shows the POC at 29523.00 — right where you are now. The strongest support line has six touches and runs down from 21:34 to 22:53.”

“You're flat and have 600 dollars of daily loss room left. The rules are waiting: if price retests the range low and holds, the retest confirms and I fire. If it breaks below the low before the retest window closes at 23:49, the setup dies for today.”

Three answers, one question each, and every number in them came off the chart in front of her — the range, the break price, the profile's point of control, the support line's touch count, the money left in today's budget, and the minute the setup expires. Then she stops talking.
Voice in, voice out
Press the button or say her name. Pick the voice you want to hear — several are available, and she keeps talking while you work.
Spoken orders
“Buy two at market.” “Place a sell limit at 29630.” “Flatten.” She parses price from quantity, reads the order back, and places it through the same guarded path your own clicks use.
Silence by default
She answers what you asked and stops. No unrequested chart commentary, no filler while you are in a position — a deliberate design rule, not an accident of the model.
No hands of her own
She cannot claim to have placed an order she did not place. Order claims are checked against what actually reached the book before she is allowed to say them.
Not a signal service
A coach, not an oracle. She explains; you decide. Every answer is logged.
03 · Vision & hands

Her hands work.

This is no longer a thing she is going to be able to do. She reaches the chart. Ask her about a level and she draws on it — a glowing arrow whose point sits on the exact price she is naming, a dashed halo around the zone she is explaining, labelled in her own words.

And her reach is not one pane wide. With the vision system integrated she can point across windows — at the chart, at the panel beside it, at the heatmap on the other monitor — and walk through how the things on your screen relate to each other, marking each one as she goes.

It is the difference between an assistant that describes your screen and one that is in it. Say “clear your marks” and they all go.

30160.25 · 15m ORB HIGH “Where’s the retest level?” ORBIT THE RETEST “Right here — the top of the opening range. I’ve marked it, and circled where it has to hold.”
She points
A glowing arrow whose apex sits on the exact price she is talking about — reaching down from above or up from below, with her label riding the tail. Not "around 30160". There.
She circles
A dashed halo around the price zone she is explaining, sized with the bars so it hugs the action at any zoom. It stays put while you keep talking.
She labels
Every mark carries a word: THE RETEST, ABSORPTION, YOUR STOP. You can read the chart afterwards without replaying the conversation.
You clear it
“ORBIT, clear your marks.” All of it goes, in one sentence. Her drawings are hers — yours are never touched.
Coming next
Highlighting any control in the app while she explains it, and clicking through menus herself while you watch. In testing now.

Pointing and drawing work today. Highlighting controls and driving menus are in active testing and are not finished — they are on the roadmap, not in your hands yet.

04 · Execution

The whole order ticket, docked to the chart it belongs to.

Market, bid, ask, limit and stop — buy and sell, one click each, always in the same place. Underneath: a full bracket builder with take profit, stop loss, breakeven and trailing, in ticks or in dollars, with multi-level scale-outs. Set it once and every entry inherits it.

ORBIT's order ticket docked beside a live chart, showing both sides, four order types, balance, open quantity and P/L, the OCO bracket builder, and the ASK ORBIT panel below it. ENLARGE
One ticket, every order type.
The order ticket with a live position, showing open quantity, open P/L, day P/L and the resting stop and target orders. ENLARGE
With a position on. Open quantity, open P/L, day P/L, and every resting order listed by name.
Price pills on the ORBIT chart showing the entry, take profit and stop loss levels with dollar values and points to target. ENLARGE
Your bracket, on the price axis. Entry, target and stop as live pills — with points and dollars to each.
OCO brackets
Stop and target attached at entry, cancelled together, in ticks or dollars. Up to four scale-out levels.
Breakeven & trailing
Move the stop to entry after a set gain; trail by a fixed distance with an offset. Both configured in money if that is how you think.
Cancel & flatten
One button that pulls every working order and closes the position. It is the largest control on the panel on purpose.
Risk-based sizing
Tell the engine what a trade may cost you, and it sizes from the distance to the stop — not from a fixed contract count. Three separate rails cap the result.
Fire drills
A test-fire button that pushes a synthetic signal through the entire pipeline, so you know the plumbing works before anything real is armed.
Real orders take a deliberate act. Live trading requires two separate flags set by hand in a configuration file on your own machine — there is no switch in the interface that can turn it on, by design and permanently. Until you set them, every chart in the building is paper.
05 · Your screen

Every panel collapses to exactly as much as you need.

Screen space is the scarcest thing a trader has. So nothing in ORBIT is one fixed size: each panel folds down to a headline, opens to the numbers that matter, or opens all the way — and remembers which you chose, per chart.

MINIMISED
The account panel collapsed to balance, net change and a health bar.
Balance, net, health. When you are watching price and nothing else.
OPEN
The account panel showing daily and open P/L, position, and the daily loss, drawdown and target rails with remaining room.
The three rails, with how much room is left in each — daily loss, drawdown, target. Read in one glance.
TICKET · COMPACT
The order ticket collapsed to buy and sell market, bid, ask, limit and stop buttons plus cancel and flatten.
Eight buttons and the flatten. Everything you need mid-trade, nothing you don't.
TICKET · FULL
The fully expanded order ticket with OCO brackets, breakeven, trailing and Ask ORBIT docked at the bottom. CLICK TO ENLARGE
And the whole bracket builder when you are setting up, with ORBIT waiting at the bottom.

Macro Market Watch — and yes, there is a light theme.

A strip across the top of the cockpit showing the names that move index futures — NVDA, AAPL, MSFT, GOOGL, AMZN, AVGO, META, MU — each with last, high, low, VWAP distance, previous close and volume. When five or more sit on one side of VWAP, it says so, because that is when the break behind you has something behind it.

The Macro Market Watch strip in dark theme, showing eight large-cap tickers with sparklines, VWAP and volume.
THE WHOLE APP SWITCHES, NOT JUST THIS STRIP
06 · The Brain Engine

An AI inside the bot — and you are the one training it.

This is the part of ORBIT that gets better while you use it. Not a model somebody else fitted to somebody else's trades: a brain that learns from your tape, one labelled setup at a time, and is built to keep learning across thousands of them.

Every time you mark a setup you believe in and commit it to memory, the corpus grows by one and the engine has one more real example of how you read a chart. It never stops being your data, and it never stops being the thing that makes the model yours rather than generic.

YOUR CHART BUY 🧠 SEND TO MEMORY 🧠 ORBIT’S MEMORY 197 198 LESSONS · ALL YOURS

Drop a label. It gets a brain tick the moment it is saved. The corpus goes up by one — and the next time she trains, she trains on that too.

A live ORBIT chart with BUY labels placed on it and a SEND TO MEMORY pill showing five unsent labels. CLICK TO ENLARGE

And here it is on the real chart

The SEND TO MEMORY pill with a pink brain icon and a badge reading five.

The pill only appears when there is something to give her, and the number is how many labels are waiting. Press it and they go — each one carrying the chart around it, not just a price and a direction.

The pink brain is deliberate. It is the same mark that appears beside every label already in the book, so at a glance you can see which of your calls she has learned from and which are still yours alone.

The Brain Engine panel open over a live ORBIT chart showing the training corpus, hardware picker and a glowing brain graphic with six skill percentages. CLICK TO ENLARGE

It is measurably getting smarter. Here is the same panel, days apart.

Twelve more labelled setups went into the corpus between these two screenshots. Nothing else changed — same model, same machine, same six skills. Watch what moved.

BEFORE · 198 LESSONS · 67%
The Brain Engine panel showing 198 lessons and 67 percent learned, with range position at 6 percent and value-area map at 12 percent. CLICK TO ENLARGE
AFTER · 210 LESSONS · 74%
The Brain Engine panel showing 210 lessons and 74 percent learned, with range position risen to 60 percent and value-area map to 22 percent. CLICK TO ENLARGE
Range position
6% → 60%. The big one. Twelve labels taught it more about where in the recent range you like to act than the previous hundred and ninety-eight had.
Value-area map
12% → 22%. Slower, and honest about it.
Momentum reading
78% → 78%. Already learned. It did not pretend to improve to make the picture nicer.
Opening range & directional bias
0% → 0%. Still nothing. On this corpus, so far, those two have taught it nothing — and it says so rather than rounding up.
Overall
67% → 74% learned, and held-out accuracy from 85% to 88% — now measured by 5-fold cross-validation across all 210 lessons instead of a single held-back fifth.
This is what "trained on thousands of data points over time" actually looks like from the inside: not a number that only goes up, but a number that moves when you teach it something and sits still when you do not.
Close-up of the Brain Engine panel: 198 lessons, 85% on held-back labels, a GPU picker and six skill percentages around a glowing brain. CLICK TO ENLARGE

Six skills, scored against held-back lessons

Momentum reading. Range position. Volume pressure. Value-area map. Opening range. Directional bias. Each one is scored by hiding a fifth of your labels, training without them, and checking — so a skill sitting at 0% is telling you the truth: on your corpus, so far, it has learned nothing there. It only moves when you teach it something it did not already know.

The panel shows the corpus size, when it last trained, how it scored, and every GPU in your machine so you can choose which one does the work.

And it does not trade. The Brain Engine studies your labels and reports what it found. It does not gate the engine, place orders, or touch a live decision. Promoting it to a voice in live trading is a step taken deliberately, later, with your eyes open — not a thing that happens to you one morning because a model got confident.

Only your labels ever enter the book. The platform's own calls are kept in a separate store by construction and can never be merged in — because a training corpus contaminated with the machine's own guesses teaches the machine to agree with itself.

Where this goes next: a brain that learns from a fleet

Right now your Brain Engine learns from exactly one trader — you. That is the correct place to start, and for a lot of traders it is the whole point. But there is an obvious next step, and it is worth saying out loud rather than shipping quietly.

Think of how a fleet of cars makes the next release of the driving software better than the last: thousands of drivers encountering situations no single driver would meet in a lifetime, and the lessons coming back to a shared model. A shared Brain Engine is the same idea applied to setups — a model that has seen how thousands of traders labelled the same opening range, the same failed retest, the same chop, and is therefore better at recognising all three than a model taught by one person in one market on one screen.

Opt in, or nothing happens
This is off unless you switch it on, and switching it on is a decision you make deliberately, not a checkbox buried in an installer or a term you agreed to by using the product. Never retroactive: turning it on does not reach back for labels you drew before you did.
Labels, never your market data
What a contributing trader would share is their own judgment — the labels they drew and the shape of the setup under them. Not the feed. Your market data is licensed to you by your provider, it is not ours to move, and no server of ours takes it in today or in this plan. Not your credentials, not your positions, not your account numbers.
Your own brain keeps working regardless
A trader who never opts in loses nothing they have today. The local model still trains on their labels, on their machine, and still improves with every one they add. Contributing is how you help the shared model — it is not the price of having your own.
Where the fleet analogy stops
The usual version of this collects by default and asks later. That is the part we are not copying. A trading label is a record of how someone thinks about risk, and that belongs to them until they say otherwise.
This is direction, not a feature you are buying today. It is described here so nobody is surprised by it later, and so the boundary is on the record before there is any code that could cross it.
07 · The bot library

Twenty strategies, ready to arm. Every one born disarmed.

The opening range is the house strategy, but it is not the only one in the building. Pick a classic from the library — crossovers, mean reversion, volatility breakouts, VWAP, channels, momentum — and each arrives with a plain-English explanation, a suggested timeframe, an ATR-based stop and target, and a cooldown.

ORBIT's NEW BOT panel listing twenty strategies including EMA crossover, golden cross, MACD, RSI reversal, Bollinger breakout, VWAP bounce, Donchian and Turtle breakouts. CLICK TO ENLARGE

Illustrated, explained, and honest about itself

Each entry draws its own idea — you can see what a squeeze breakout or an engulfing reversal actually looks like before you pick it. The description says what the strategy believes, in one sentence, without jargon.

Then, at the bottom of the panel, the line that matters:

“Every new bot is born DISARMED and on PAPER — tune it in the BOT MENU, then arm it. A strategy being popular is not a promise it works.”

That sentence is printed inside the product, where a customer sees it every time. It is the same standard the rest of the platform is held to.

08 · Indicators & order flow

The standard library, plus what the tape is actually doing.

Thirty-two classical studies, each placeable on the price panel or in its own pane with a single toggle. And underneath them, a second deck that most retail platforms do not have at all: reaction-based order flow, read from the published prints.

ORBIT's standard indicator library with TOP and BOT placement toggles for each of 32 studies. CLICK TO ENLARGE
Thirty-two classics. TOP puts it on price, BOT gives it its own pane. No dialogs.
ORBIT's advanced order flow indicators: big-trade bubbles, absorption marks, stacked imbalance, iceberg watch, VWAP bands, cumulative big-print delta and buy/sell pressure. CLICK TO ENLARGE
Advanced order flow. “Print-based ones read the published big prints — reaction, never prediction.”
Big-trade bubbles
Every print above your minimum, drawn as a circle at its traded price — size and glow carry the contract count.
Absorption marks
Big prints hammering a bar's extreme that closes back away: rings where aggression was absorbed.
Stacked imbalance
Three or more same-side big prints laddering with price — stacked arrows of initiative conviction.
Iceberg watch
The same-size print refilling at one price again and again — diamonds on suspected hidden liquidity.
Cumulative delta
Running sum of signed big-print contracts, from the feed floor up, plus per-bar delta and a 0–100 buy/sell pressure line.
Volume profile
POC, value-area high and low, built live from real order flow and drawn on the price axis of every chart.
A break with aggressive buyers stacked behind it prints differently in the flow than one running on air. The difference between a break and a fake is written in the tape — for anyone equipped to read it.
09 · The trade journal

Five pages that grade the trader, not just the account.

Every fill lands in a journal that separates your accounts, computes the statistics that actually predict survival, and then does the part nobody else does: tells you what it found, in sentences.

The ORBIT trade journal overview page showing the ORBIT SCORE dial, statistics tiles, account comparison, session breakdown and four charts. CLICK TO ENLARGE
Overview. Paper accounts across 114 trades. Note that this book is down — the journal is built to show you that clearly, not to flatter you.

The ORBIT SCORE, and the distance between two of them

One number, out of 100, built from six weighted components — each shown with its weight so you can see exactly why the number is what it is. The ranks climb from LIFTOFF to ESCAPE VELOCITY. Below are two real score cards from the same trader's own books, days apart. This is the point of the whole journal.

BEFORE · 40 / 100 · LIFTOFF
The ORBIT SCORE dial reading 40 out of 100, labelled LIFTOFF, with six weighted component bars mostly empty. CLICK TO ENLARGE
Profit factor 0.91 — the book is paying out more than it takes in. Recovery 0.00. Consistency 0/100. Drawdown −$2,379 against a win rate of 48%. The win rate was never the problem.
AFTER · 87 / 100 · ESCAPE VELOCITY
The ORBIT SCORE dial reading 87 out of 100, labelled ESCAPE VELOCITY, with five of six component bars full. CLICK TO ENLARGE
Profit factor 2.75. Average win/loss 3.01×. Drawdown cut to −$365. Recovery 5.84. Win rate: 48% — exactly the same. Consistency still red at 29/100, and the card says so.
Both books won 48% of their trades. One scored 40 and one scored 87. Everything that moved was the size of the losses and the speed of the recovery — which is precisely what a trader controls, and precisely what a win rate hides.

How the score is designed to be improved

The weights are published on the card because a score you cannot reverse-engineer is a score you cannot act on. Read left to right, they tell you what to fix first — and the cheapest fixes are almost never “win more often.”

Profit factor · 25%
Gross wins over gross losses. The heaviest single weight, and the one that moves when you stop widening a stop. Above 1.0 the book pays; the card shows the number and the raw ratio side by side so you cannot fool yourself about which side of 1.0 you are on.
Average win / loss · 20%
How big a winner is against a loser. In the pair above this went from 1.35× to 3.01× and did more work than anything else. You raise it by cutting losers at the level you decided on before you entered — not by holding winners longer and hoping.
Max drawdown · 20%
The deepest hole from an equity peak. This is the component that punishes the one oversized loss, and it is why the card's own caption warns that “one oversized loss dents nearly all six at once.” Position sizing and the daily-loss rail are the levers.
Win rate · 15%
Deliberately the fourth-heaviest, not the first. Chasing it is the classic way to wreck the other five — you widen stops, you cut winners early, the hit rate climbs and the account bleeds. The card is built so that strategy scores worse.
Recovery factor · 10%
Net profit against the worst drawdown: how well the book digs itself out. It answers the question a curve cannot — not “did it recover” but “how much did it cost to.”
Consistency · 10%
Whether the results are spread across many days or carried by one. In both cards above this is the red bar — and in the 87 book it is red because a single day carried more than 100% of the profit. The score keeps saying so at 87 exactly as loudly as it did at 40.

And the discipline page carries the other half of the answer: rule adherence is tracked separately from money, because a green day on a broken process is not progress. The app's own guidance is to aim for 90% adherence across 30 to 50 days before chasing the score — get the process clean, and the score follows it. Not the other way around.

The journal analysis page listing thirteen written observations about the trader's sessions, timing, losing streaks and expectancy. CLICK TO ENLARGE
Thirteen observations, written out.

It says the uncomfortable part

“Right after two straight losses, the next trade has averaged −$43.41 over 46 tries. A short pause after the second loss is the cheapest protection in trading.”

“Days with more than 8 trades have earned −$14.76 per trade, against +$44.48 on lighter days. More shots has not meant more game — selectivity is the strategy's own first rule.”

“The 10:00 ET hour has cost −$779.50 across 13 trades so far. Nothing says it always will — but knowing your reddest hour is how a personal rule gets made.”

Each observation names its own sample size, because an observation drawn from four trades is a rumour.

ORBIT's trading calendar for August 2026, with each day showing net P/L, trade count, win rate and percentage of daily goal, and weekly totals down the right. CLICK TO ENLARGE
Calendar & goals. Daily, weekly and monthly targets you set, with weekly totals and event days flagged.
The journal discipline page showing a daily rule checklist and a rule adherence score of 50 percent. CLICK TO ENLARGE
Discipline. Tick your own rules, finish the day, and the adherence score builds from clean days — separately from whether the day was green.
“A profitable revenge trade is a process FAILURE. A controlled stop-out on your setup is a process SUCCESS. Mark the rules, not the money.”
— printed under the Progress Tracker, in the app

Every figure on this page comes from paper and broker test accounts — they show what the journal measures, not what anyone earned.

10 · The architecture

Three systems. One engine. Nothing to keep in sync.

ORBIT is now three windows that each do one job properly — the trading platform, the BOT LAB, and the HEATMAP — and the thing that makes them worth having is that they are not three applications. They are three views onto one engine, one market-data connection, one set of books and one brain.

That sounds like an implementation detail. It is the whole product. Anyone who has run a chart package beside a strategy tester beside a depth tool knows the tax: three logins, three data subscriptions, three ideas of what your position is, and a reconciliation problem every evening. Here there is nothing to reconcile, because nothing was ever separate.

ORBIT's main trading chart on MNQ with opening-range levels, volume profile, a docked order desk, risk rails panel and a glowing SELL signal marked at a specific price. CLICK TO ENLARGE
ORBIT. Where you trade — charts, order desk, risk rails, and a copilot that can see all of it.
ORBIT's account panel listing nineteen separate paper accounts, each with its own day P/L and equity, above combined performance statistics and risk rails. CLICK TO ENLARGE
One set of books, visible from every window. No window has a private idea of what you are carrying.
ORBIT — the platform
Where you trade. Up to eighteen live charts, each its own desk with its own order ticket, its own risk rails and its own account. The opening range drawn, the profile built from real prints, and a voice-native copilot that can see your whole screen and mark it up while she explains it.
BOT LAB — the proving ground
Where strategies argue it out. Twenty-five variations running at once on the same live feed, each on its own paper book, plus ORBIT LEARN keeping shadow copies of every model in the library and scoring them on their own closed trades.
HEATMAP — the microscope
Where you go to see what a move had to eat through. The recorded resting book as heat, the prints that went off as bubbles, on its own monitor, with no order path anywhere in it.
One connection, one brain, one journal
All three read the same feed and write to the same books, so the journal totals the lot and ORBIT answers from any of them with the same numbers. Switch windows and nothing reloads, nothing re-syncs, nothing disagrees. Jump straight between them from the header of whichever one you are in.

And the part that gets better while you use it

Underneath all three sits the loop that makes ORBIT different from a platform that ships the same on day 400 as it did on day one. It learns how you trade, from you, on your machine. Here is exactly how, because a claim like that is worthless without the mechanism.

1 · You teach, by marking
Four tools in the chart toolbar: BUY, SELL, the BAD TRADE indicator, and a CHOP ZONE you drag as a region across the bars it covers. You drop them where your judgment says they belong — including on the trades you would not have taken, which is the half most systems never capture.
2 · Each mark carries its setup
A label alone is nearly useless: the same setup means different things on different settings. So every mark is stamped with the exact configuration underneath it — the entry model, the ranges armed, the bot family and edition, the strategies and modifier, the instrument, the timeframe, and that copy's actual tuned numbers, read fresh from the engine at the moment you send it.
3 · Only your labels enter the book
Your marks and the platform's own calls live in two separate stores, and the guard that keeps them apart runs on both sides of the wire. A corpus contaminated with the machine's own guesses teaches the machine to agree with itself — the most expensive mistake in this entire field, and the one it is structurally impossible to make here.
4 · It trains, on your hardware
Press TRAIN ON MY LABELS NOW and the Brain Engine studies the corpus, on your GPU if you have one. It reads six skills off the market around each label: momentum, where price sat in its range, whether volume was running heavy, where price stood against the volume profile, where it stood against the opening range, and the lean of the tape.
5 · It grades itself, honestly
Scored by cross-validation on labels it was not shown, against the score for simply guessing the commoner side. That gap is the only number that means anything, and it is the one the panel puts in front of you — per skill, so you can see which parts of your read it has actually picked up and which it has not.
6 · It does not trade. Yet.
The model measures, reports, and stops. It does not gate a signal, size a position or touch the engine. Promoting it to a voice in live decisions is a deliberate step, taken later, on purpose — not something that happens to you one morning because a percentage got high enough.
ORBIT's Brain Engine panel showing corpus size, training run details, a brain diagram surrounded by six named skills with percentage scores, and a list of available training hardware. CLICK TO ENLARGE

Six skills, scored separately

A single accuracy number hides everything interesting. Splitting it six ways tells you something you can act on: that it has learned to read momentum the way you do but has barely started on where you stand against the opening range, say, and that the next forty labels are best spent there.

Each skill is a plain sentence about the market, not a black box — how price was pushing into your label over the last 1, 5, 15 and 30 minutes; whether trading was heavier or lighter than the last half hour. You can read the list and know exactly what it is looking at.

And every one moves as you teach. The corpus and the score in this shot are from a development machine on 24 August 2026; the number that matters is not its value, it is that it changes when you feed it.

Nothing here promises a result. What it promises is a mechanism you can inspect — your labels, your machine, your hardware, a score measured on data the model never saw, and a model that is not allowed near your orders until you decide otherwise.
11 · The Bot Lab

Twenty-five strategies, running at once, each on its own book.

Every trader has the same argument with themselves: is this setup actually working, or do I just remember the times it did? The BOT LAB settles it by running the variations side by side, in the live market, on separate paper accounts — so at the end of a session there is a table instead of a feeling.

The BOT LAB wall: rows of MNQ and NQ charts on 5m, 15m and 30m, each carrying an opening-range bot preset with its own open and day P/L and its own paper account. CLICK TO ENLARGE
One row, one experiment, one book. Same market, same moment, different settings — which is the only way a comparison between them means anything.
The ORBIT LEARN scoreboard listing models with fires, wins and losses, win percentage, average R, profit factor and conviction, beside an ORBIT SCORE dial reading 85. CLICK TO ENLARGE

Measured, never projected

ORBIT LEARN keeps a shadow copy of every model in the library. Those copies watch the same live feed you do and place no orders at all — they are scored on their own closed paper trades, so a strategy earns its row by trading it, not by being backfitted to a chart that already happened.

Each row carries fires, wins and losses, win rate, average R, profit factor and how much conviction the model had. The status column is blunt about which of them have simply not traded enough yet to be judged.

“Self-training scoreboard — shadow tests measure on the live feed and place no orders. Measured, never projected.”

That line is printed at the top of the panel inside the product, where it is read every session.

Its own account, per row
Every cell is bound to a separate paper book with its own balance, its own risk rails and its own trade log. Two cells cannot quietly share a book and double-count a result — the totals walk unique account ids, not rows.
Paper only, by construction
The lab is where ideas prove themselves before they ever touch a live route. A signal from a paper window does not reach the order client at all — not a dry run, not a demo endpoint. It is a property of the engine, not a badge on a screen.
FIX & FILL
One press rebuilds the wall: it repoints duplicated cells onto missing ones, fills the gaps, and rebinds each chart to the book it now answers to. A wall that has drifted after a day of tinkering repairs itself.
ARM ALL · DISARM ALL
Arms every cell that actually has a strategy on it, and says how many it left alone. The button face follows the wall, so it always states what it will do next rather than leaving you to guess.
One total line, deduped
Realised, unrealised, total and today, for the whole wall, above the wall. It counts each book once. A total that lies is worse than no total at all.
Ask her about it
“How’s the bot lab doing?” from any window, and ORBIT reads the same numbers off the same books — so she and the wall can never disagree.

And then it grades the trader, not just the bots

LAB RESULTS opens the journal across every book on the wall at once. Not a P&L readout — a report card, built from six weighted components, where win rate is deliberately fourth, because chasing it is the classic way to wreck the other five.

The Trade Journal overview across six paper accounts, showing an ORBIT SCORE of 85 with its component bars, a grid of measured statistics, and charts for cumulative net P/L, trade distribution, daily P/L and drawdown from peak. CLICK TO ENLARGE
Paper books. Every figure here is from simulated accounts — what the journal measures, not what anyone earned.
ORBIT's analysis panel listing thirteen observations about the trading record, covering sessions, timing, behaviour after losses, trade count, streaks, expectancy, fees, drawdown, instrument and consistency. CLICK TO ENLARGE
Thirteen observations, in sentences. Each one names the number it came from, so it can be checked rather than believed.

The figures in these shots come from paper accounts during development. They show what the journal measures on simulated books over a short window — they are not a track record, they are not live results, and they do not promise either. ORBIT is still being built; this is where it had got to on 24 August 2026.

12 · The Heatmap

Where the resting orders were, and what actually traded through them.

A candle tells you where price ended up. It does not tell you what was sitting in its way. The HEATMAP window records the resting book as heat and paints the prints that went off as bubbles, so the liquidity a move had to eat through stays on the screen after the move is over.

ORBIT's heatmap window showing horizontal coloured liquidity bands across a time axis, with circular bubbles marking executed trades and a live price ladder pinned to the right edge. CLICK TO ENLARGE
Its own window. Put it on the second monitor and leave it there — it does not borrow a pane from the chart you are trading.

Tuned by hand, because every instrument reads differently

Six controls, all live, none of them buried in a dialog. HEAT and DIM shape how the liquidity reads. SENSITIVITY, SIZE, GROUPING and OPACITY shape the bubbles — which prints are worth drawing at all, how big they are, how tightly they cluster in time, and how much of the heat behind them shows through.

THERMAL swaps the palette. PAUSE freezes the right edge so a burst can be studied while the market keeps going without you.

Wheel over an axis zooms that axis. Drag pans. Double-click goes back to following the live edge. The axes are locked where they belong — price on the right, time along the bottom — so the picture never wanders off while you are reading it.

A close view of the heatmap showing coloured horizontal liquidity bands with overlapping translucent bubbles marking individual executed trades. CLICK TO ENLARGE
Eyes only, by construction
There is no order path from this window at all. Not a hidden button, not a keyboard shortcut — the code to place an order does not exist here. A window built for staring at liquidity is a window you should never be able to fat-finger a position from.
It records, it does not replay someone else's
The heat is built from the book as your own feed reported it, on your own machine, while you were watching. Nothing is fetched from a server of ours, because there isn't one in the path.
Big trades, marked on the chart too
The same prints that raise bubbles here surface on the trading chart as the big-trades indicator — numbered circles on the bar that carried them, so a print worth noticing is noticed without leaving the chart.
Its own flag
The window boots under the HEATMAP artwork rather than flashing the ORBIT splash first, and carries its own light and dark marks. On a desk of eight windows, knowing which one you just alt-tabbed into matters.

What the heatmap can show depends on the market-data entitlements you hold with your own provider. The window in these shots is drawing the book as this feed reported it during development, without a deeper depth entitlement enabled — a fuller book yields a fuller picture, and that is a subscription you arrange in your own name, not something ORBIT supplies.

13 · The material

Liquid glass: an interface you can see through.

A trading screen has a problem no other software has. Everything on it is urgent, and all of it is competing for the same square inch — the tape underneath, the levels drawn on it, your position, your orders, the panel you just opened. Stack it in opaque boxes and you spend the session closing windows to see the chart you opened them to read.

So ORBIT's panels are made of glass. They float above the chart rather than replacing it. The market keeps moving underneath them, blurred and saturated so it reads as depth instead of noise. Nothing you need is ever fully hidden by something you opened.

OPEN POSITION · NQ
+$672.50
Long 3 from 30246.75. Stop is under the retest low; target sits at the value-area high.
STOP30246.75
TARGET30292.50
DAY ROOM$600
TURN THEM OFF — WATCH THE GLASS BECOME A BOX

Blur & saturation

What is behind the panel is smeared and its colour pushed, the way light pools inside real glass. It reads as distance, which is why your eye stops trying to focus on it and the panel becomes legible.

Refraction

A gradient tint running bright at the top-left corner and dark at the bottom-right — the direction a real lens would gather light. It gives a flat rectangle a shape.

The specular edge

A one-pixel bright line inside the top edge and a dark one inside the bottom. Almost the entire illusion lives in those two hairlines: they are what makes the surface look thick.

Depth

A long, soft shadow underneath. Not decoration — it is the cue that tells you the panel is above the chart and the chart is still there, which is the whole argument for the material.

Light that follows you

Surfaces catch a highlight where your cursor is, so the glass responds to the hand. Move over the screenshots on this page — the frames are doing it right now.

Colour with a job

Green is yours and working. Purple is the machine's. Red is a rail you are near. Amber is the clock. No colour is used decoratively anywhere in the app, so a colour appearing always means something.

Every panel moves
Drag it by its header, resize from the corner, and it remembers where you left it — the Brain Engine, the journal, the bot menu, the engine's day, the order ticket. Your screen is yours.
Live, not static
The brain graphic pulses along its own pathways. The score dial has a rocket that climbs it. The starfield behind the loading screen drifts. None of it blocks the tape.
Motion with manners
Animation throttles itself when the app is ingesting hundreds of thousands of bars, and honours your system's reduced-motion setting. A pretty interface that drops frames during the open is a broken one.
Easter eggs
There are a few. We are not going to tell you where. One of them dances.

ORBIT is in private development.

Early access is limited to the first 500 applicants, with founder pricing locked for as long as the subscription runs. One email — first access, launch news, and a say in what gets built next.

Join the launch list