How your agent thinks

Most automated trading sits at one of two extremes. On one end: rigid bots that execute static rules with no judgment. Any market state that doesn't match a pre-set threshold is invisible to them. On the other end: opaque "AI" black boxes you have to trust because they won't show their work.

Every agent you run on Engine is neither. It reads the market the way a thoughtful trader would, follows its strategy, and trades inside its limits. Every move it makes is traced, in plain English, in your decision log. This is the trading agent we wanted to use ourselves and couldn't find anywhere else.

It reads the market holistically

The agent doesn't act on one indicator at a time. It weighs price, momentum, liquidity, volume, market regime, and the signals relevant to its strategy together for every market in its universe.

A static "if RSI < 30 then buy" bot sees one number. The agent sees a market and interprets it.

It trades on the data it can actually verify

Engine includes a broad catalog of market, technical, positioning, liquidity, macro, news, and token data. A strategy can also declare private data feeds and on-demand tools when its edge depends on something Engine does not provide.

Description alone does not make a missing data point real. Before an agent can publish, Engine checks that the important indicators and timeframes in its strategy are available. It reports a clear error instead of silently swapping in a different signal. During live trading, temporarily unavailable evidence stays marked unavailable, so the agent waits rather than pretending it saw a confirmation.

That keeps the system flexible without making it vague: if you can describe an idea and connect the evidence it requires, the agent can apply it.

It adapts to changing regimes

When volatility expands, liquidity moves, or participation changes, the agent's reading of its rules updates with the conditions. Nobody has to manually retune. What the strategy says keeps meaning what it meant.

This is the difference between an agent that works for two weeks and one that works for a year.

It explains every decision

Trades and decisions are recorded in plain English: what the agent saw, what rule matched, and why it acted or stayed out. If something confuses you, ask the agent in chat.

Other "AI traders" don't show their work because there isn't any work to show. The agent's transparency isn't a feature; it's the feature.

One market view, applied to your account

Everyone running the same agent shares its market view and trade ideas. Engine then checks each account separately using its current balance, positions, limits, and approval setting. That is why the same valid idea can open for one user, wait for another user's approval, and be skipped for someone whose account is already at the strategy's limit.

Once a position is open, it stays linked to the exact strategy version and trade idea that opened it. Position reviews use that original plan and current evidence, not an unrelated newer idea for the same market.

It gets sharper, the strategy doesn't change

We keep shipping updates to make the agent's judgment better. You don't have to do anything to pick them up. And crucially: an agent's strategy stays exactly as written. The what is its to define. The how is ours to keep improving.