One trade, from the exchange
to the record it leaves behind.
Seven stages, in this order, every time. Scroll and the diagram follows — each stage is drawn as it is described, and the page changes colour as the trade moves through it.
01
Exchange
02
Analysis
03
AI
04
Risk
05
Execution
06
Journal
07
Analytics
It starts with a connection you control
Nexus connects to Binance USDⓈ-M for live market data today, with more venues on the roadmap. Candles arrive over Binance's public websocket, with REST history to warm indicators up, and only closed candles are used. Duplicate, out-of-order and missing candles are detected, a stale feed pauses new entries rather than trading on a gap, and after a restart missed candles are replayed in order before trading resumes.
- Live venue
- Binance · more on the roadmap
- Key scope
- Trade only, never withdraw
- Feed
- Closed candles · gap-checked
Prices become a description of the market
Every close, the strategy reads what the chart is doing rather than what it costs: trend on each timeframe it declares, swing structure, support and resistance, and the market regime. This is the layer that lets a strategy say “only in a trend” and have that mean something enforceable.
- Extracted
- Trend · structure · S/R
- Timeframes
- As declared per strategy
- Regime
- Trend · range · volatility
The Decision Brain scores it
The strategy decides where a trade could be; the Decision Brain decides whether it is worth taking. It scores the setup out of 100 from eight weighted factors — higher-timeframe alignment, regime fit, momentum, reward to risk, stop safety, volatility, structure and volume — keeps the rules it passed and failed, and blocks some setups outright, such as reward to risk below 1.
- Factors
- 8 · weighted to 100
- Minimum
- 60 of 100, configurable
- Output
- Score + passed and failed rules
Then it has to get past risk
Every order intent passes the risk checks: open-position limit, no pyramiding, correlation, event blackouts, trading day and session, daily and weekly loss limits, cooldown after a loss, trades per day, stop validity, sizing, exposure and the venue's lot rules, then the global risk manager. Any one failing rejects the order with a written reason, and if a check cannot run, nothing trades.
- Checks
- 20 · all mandatory
- On failure
- Fails closed
- Result
- Accepted · or a reason
The order fills on paper, at the live price
Approved orders fill on the paper broker against the live Binance price. A limit entry fills only when price trades through it, and expires unfilled after a set number of candles rather than being chased; market and stop orders pay modelled spread and slippage, and every fill pays a fee. Stops and targets are managed by the engine rather than held at an exchange, and live order routing is locked.
- Broker
- Paper
- Costs
- Spread · slippage · fees
- Live routing
- Locked
Every outcome is written down and kept
The trade closes and is kept: the decision that opened it, the reasoning at the time, the outcome and a review of what went right or wrong. Rejections are recorded too — the record of what the system nearly did is often more instructive than the record of what it did.
lesson: retest held — size normally
mistake: third retest in compression
lesson: retest held — size normally
recalled at the next similar setup
- Stored
- Context · reasoning · outcome
- Rejections
- Recorded as decisions
- Where
- Journal · dashboard
And the record tells you what is actually working
The dashboard's analytics break results down by strategy, symbol and session, with fees already inside every paper fill. An equity curve tells you whether something is working; the breakdown tells you what — and often that one symbol in one session is carrying the rest.
a breakdown, not a single curve
- Breakdown
- Strategy · symbol · session
- Costs
- Inside every fill
- Source
- Your paper trades
What feeds back, and what does not
What feeds back on its own is size, never rules. After every closed trade an instance re-reads its own record: a losing pattern that repeats — on a symbol, in a regime, in one direction — trades smaller, never below half size; two losses in a row halve risk and four quarter it; and trading below its recent equity average halves it again. A pattern that has proven itself may add at most 25%, and only when nothing else is cutting size. Lessons lapse after 14 days unless newer trades confirm them. A strategy's rules, its risk settings and where it trades change only when a person changes them.
- Exchange→
- Analysis→
- AI→
- Risk→
- Execution→
- Journal→
- Analytics
- ↺back to 03