> For the complete documentation index, see [llms.txt](https://docs.tallyprotocol.xyz/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.tallyprotocol.xyz/architecture/trade-plan.md).

# The trade plan

`engine.trade.prepare({ ticker, issuer, usd, tolerancePct, user })` turns "buy $X of this token" into something a wallet can sign. It re-quotes on every call and answers one of three states.

| Status           | Meaning                                                    | What the app does                                                 |
| ---------------- | ---------------------------------------------------------- | ----------------------------------------------------------------- |
| `needs_funds`    | Not enough USDT or BNB, with the shortfall                 | Shows the shortfall and the top-up options                        |
| `needs_approval` | An exact-amount `approve(guard, amountIn)` is needed first | Asks the wallet to sign it, waits for it to mine, then asks again |
| `ready`          | Guard calldata, gas limit and simulation result            | Shows the review and asks the wallet to sign                      |

## Why two phases

The gas estimate and the simulation both need the allowance to exist. Without it, the guarded call reverts on `transferFrom` and the error says nothing useful. So the plan is asked for again after the approval mines, and again when its 15-second freshness window passes.

## What "ready" contains

* **The calldata** for `swapForShares` (or `swapForSharesWithFeed` for Ondo), with the router data the aggregator built for ShareGuard as the user.
* **The floor:** `minShares`, the quoted shares minus the user's tolerance.
* **The gas limit:** `eth_estimateGas` from the user × 1.25, simulated with `eth_call` at that exact limit. Binance's own simulate endpoint is also called; if it fails, that is a warning, not a gate.
* **Warnings:** for example "Market-maker quotes expire in a few seconds. Confirm promptly."

## The Ondo feed signer

Ondo has no onchain multiplier, so the contract keeps a stored value that the owner seeded and a bounded, signed feed keeps current. There is no separate sign endpoint: `prepare` signs a bounded `FeedUpdate` itself, and only when the stored value is stale or differs from the engine's accepted multiplier by more than 1 ppm. `FEED_SIGNER_PK` is read only on the server. See [Multiplier sources](/smart-contracts/multiplier-sources.md).

## Route choice

The plan chooses a route with the shared route chooser (pool route within 0.5% of the best market-maker route, otherwise the market-maker route flagged). The sell plan uses the same chooser. See [Share-true quotes](/concepts/share-true-quotes.md).

## Errors become plain words

Errors are typed (`price_moved`, `token_paused`, `feed_stale`, `rfq_required`, `not_buyable`, `below_minimum`, and so on) and mapped to short sentences by one function, so the web app, the bot and the agent tools explain failures the same way. A revert reason from the chain is decoded rather than shown raw.

## After the swap

`engine.trade.receipt(hash, ticker)` decodes the `Guarded` event: the buyer, the stock, tokens received, shares credited, the multiplier used and the router. The receipt page shows shares received against the floor, read from the chain. See [Receipts](/product-modules/receipts.md).


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.tallyprotocol.xyz/architecture/trade-plan.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.