How This Actually Works
Not every workflow works the same way. Here's what's actually happening behind each one.
TradeAnalysis has two kinds of workflows, and they're built differently on purpose. The 4 trading bots apply fixed smart money rules to live market data, with no AI in the buy/sell decision. The 10 research workflows, most of them AI-powered, work through the inputs you give them against institutional-style criteria. This page explains both, plainly.
The Trading Bots: Our SMC Trading Strategy
Scalper, Basic Day, Advance Day, and Swing all run the same smart money concepts (SMC) rules, written as fixed code and backtested on 62 markets from 2020 to 2026. No AI model and no news feed decides the direction. The bots differ only in the chart they read, the trend they follow, and the liquidity levels they watch:
| Bot | Order block chart | Trend | Liquidity it watches | Best sessions |
|---|---|---|---|---|
| Scalper | 15-minute | 4-hour | Previous day, Asian session | London, New York |
| Basic Day | 1-hour | 4-hour | Previous day, Asian session | London, New York |
| Advance Day | 4-hour | Daily | Previous day, previous week | Any |
| Swing | Daily | Weekly | Previous week, previous month | Any |
Order blocks and market structure: how a reaction area forms
On every request the bot pulls the last 500 closed candles for your symbol, fresh, plus the higher-timeframe trend (is the last closed trend candle above or below its 50-period EMA).
- An order block. The last opposite-coloured candle before a strong move: price leaves it by at least 1.5 ATR within three candles. Price often comes back to that candle, so the bot marks it as a reaction area.
- The top tier: sweep, shift, order block. When the block formed right after a liquidity sweep (a wick through one of the levels in the table that closes back inside) and a break of market structure, it is the classic SMC setup and gets the extra star.
- Still valid. A zone ends when a candle closes through its far side. A zone that has already been touched five times is dropped as worn out.
What every area gives you
- Entry zone: the order block. Enter anywhere inside it, with a limit order or after a lower-timeframe break of structure.
- Trap area: the space just beyond the zone where price often pokes, taking stops, before it turns.
- The line where the idea is wrong: your stop. For top-tier blocks it sits beyond the sweep wick; otherwise 0.5 ATR beyond the zone.
- TP1 and TP2: 1× and 2× the risk, measured from the zone edge.
- Rating, 1 to 4 stars: one each for trading with the higher-timeframe trend, a first or second touch, and a liquidity sweep or liquidity level in the trap area.
The bot shows the three best areas near price, plus the liquidity levels and today's range. When price comes back into a top-tier block for the first time, with the trend (and, for Scalper and Basic Day, during London or New York), the card headline says BUY or SELL; otherwise it says HOLD and shows the areas to watch. If the price feed is out of date, the bot says so instead of guessing. The same candles always produce the same answer.
How results are counted
Every area we post on Telegram and X is logged and tracked on live prices. TP1 hit is a +1R win, TP2 hit is a +2R win, and a stop hit is a 1R loss. The evening recap and the weekly scorecard list every area, losses included.
The Analysis Suite
The 10 institutional-style research workflows — stock screening, DCF valuation, risk assessment, pattern detection, and the rest — work differently. You give the workflow your inputs: risk tolerance, portfolio holdings, a ticker, a sector, a time horizon, whatever that specific workflow asks for. The AI reasons through those inputs against a defined set of institutional-style criteria for that workflow, and returns a structured result — the same fields every time, not free-form text.
Each workflow's exact criteria are described on its own page — see the "How it works" section on any agent page for the specifics of that workflow.
Why It's Built This Way
An AI model that's free to call a trade or invent a stop-loss is only as reliable as its last guess, and it can give a different answer to the same chart twice. Fixed rules can be checked: every bot area traces back to a specific order block, sweep and structure break on real candles, and the same rules can be tested on years of history. AI is used where judgment over your inputs is the point, in the analysis suite.
Testing rules on history is not a promise about the future. We don't publish win rates or profit figures for the bots, and no rule set wins every market or every month.
One More Thing
None of this is financial advice, and none of it guarantees an outcome — see the disclaimer in the footer of every page for the full detail. What this page is for is answering the question we think matters most before you trust any signal: what is this number actually based on?