The problem with most setups
A typical algorithmic trader today runs at least three separate tools: a charting platform to analyze the market and visualize ideas, a broker account to execute trades, and an automation layer to connect the two. Each tool has its own subscription, its own interface, and its own way of breaking.
More importantly, none of these tools talk to each other natively. A signal generated in TradingView doesn't automatically become a live order. A backtest result doesn't automatically tell you how to configure your risk. You're the connector — and every manual step between idea and execution is a point of failure.
Step 1 — Building the strategy
In Charton, strategy building happens through a conversation. You describe what you want and the system translates it into structured trading logic. There are three ways to start:
Goal-based
You describe a desired outcome: "I need a strategy on gold with a minimum 70% win rate and a maximum drawdown under 8%." The system scans millions of combinations across over 150 technical tools and returns the closest viable result based on historical data.
Knowledge-based
You describe a specific setup: "Build a strategy based on order blocks inside the London session, 1:1 risk/reward, stop below the order block." If the logic is supported in the system, it structures the rules, tests them, and returns results.
Image-based
You upload a screenshot of a chart or a hand-drawn strategy sketch. The system reads it, converts it into structured logic, and builds a strategy from it automatically.
At the end of every build, the AI reviews the strategy and surfaces up to five specific improvement suggestions — focused on win rate, drawdown reduction, or profit factor. You decide which ones to apply.
Step 2 — Backtesting
Once the strategy is built, you get an immediate summary in the chat: win rate, total trades, net P&L. This is enough to know whether it's worth exploring further.
For the full picture, the strategy moves to the Workspace. Here you see the strategy drawn directly on the chart — every entry, exit, and trigger point — alongside the complete backtesting report. This is where you verify that the logic behaved the way you intended, not just that the numbers look good.
The backtesting engine runs at tick level — processing every price movement, not just candle opens and closes. You can add commissions and slippage before running to get results that reflect real execution conditions. Any change you make through the chat updates the backtest in real time.
Step 3 — Setting risk parameters
Before deploying anything, you define the risk rules. These operate at two levels:
At the strategy level, you set stop loss, take profit, risk per trade (as a percentage or fixed amount), maximum simultaneous positions, and session filters if you only want the strategy active during specific market hours.
At the agent level — which applies across all strategies on the same account — you set a maximum daily loss, a maximum total drawdown, and a cap on total open positions. When any of these limits are hit, the agent stops automatically. This is especially important for prop firm accounts where breaching a risk rule means losing the funded account entirely.
Step 4 — Launching an agent
An agent is the entity that connects your strategy to a real (or demo) broker account and executes trades. You create an agent, connect it to a broker account via API key or OAuth 2.0 — a one-time setup — and assign the strategies you want it to trade.
Before going live, you have two options to validate in real market conditions without capital at risk: Charton's built-in demo portfolios (available on all plans) or your broker's own demo account.
When you're ready, you choose the execution mode: fully automatic (agent trades on its own), semi-automatic (agent waits for your approval per trade), or alerts only (no orders sent, just notifications). This can be changed at any time.
Step 5 — Monitoring
Once the agent is running, you can monitor it from the agent dashboard. You see overall performance, a breakdown per strategy, and live open positions. Every trade the agent executes is drawn on the chart in the Workspace — so you can verify at any point that it's entering exactly when the strategy logic says it should.
Alerts are sent by email and through the web app when something happens — a trade opened, a position closed, a limit reached, or an agent stopped. If you're in semi-automatic mode, the alert takes you directly to the approval page.
The complete picture
The entire flow — from describing an idea to a running agent — happens inside one system. The chart you analyze is the same chart where you verify execution. The strategy you built in the chat is the same one the agent trades. There's no manual export, no webhook configuration, no third-party connector to maintain.
That's the point of having everything in one place: not just convenience, but a shorter distance between what you intended and what actually runs.