AI Assistant Tools: SweepTick

Ask a connected assistant "what if I nudged this setting?" and it can build a SweepTick for you: one core position plus a copy for every value of a setting across a range, run as a single Backtick. This article explains the four SweepTick tools and how the assistant uses them with the Backtick tools you already know.

The plan and the connection you need are in Connect tradeSteward to Claude, ChatGPT or Codex. Each tool below names the permission it needs; Choose What a Connection May Do explains them. To use the same tools from your own code, see User API: SweepTick Endpoints.

What a SweepTick is

A SweepTick starts from one core position, the position as you would normally test it. You choose a setting (the short strike delta, the profit target, the stop level, the entry time) and a range of values, and the SweepTick generates one position per value, keeping everything else the same as the core. All of them run together in one Backtick, so the results page lines the variants up side by side and shows whether your chosen value sits on a stable plateau or on a knife edge. On the website this is SweepTick under the Backtick menu; the assistant builds the same thing.

• A one-setting sweep steps one setting: for example the short put delta from 10 to 30.
• A two-setting sweep (a grid) steps two settings and generates one position for every pair: for example five deltas by six stop levels, 30 positions. The results page draws these as heat maps.

A range is given either as a count (so many values from the first to the last, evenly spaced) or as an interval (a fixed step from the first to the last). For a position with several legs, a setting can be swept on one leg, or on several legs moved together.

The core can come from three places: a position of one of your Backticks, one of your bots (the bot is only read; this needs the View bot settings permission), or a set of position settings the assistant writes out itself, in the same form get_backtest_position returns. With settings as the source, no Backtick has to be made or changed first. A Multi-Entry bot or position is swept through one of its entries.

A SweepTick is a Backtick, so everything in Building and Running Backticks applies: it is a draft until it is run, it runs with run_backtest, the assistant waits for it with wait_for_backtest, it appears in the API/MCP tab of your Library and moves to the Trash 48 hours after its last change unless it is saved, and a completed one is cloned, not edited. Its positions are generated from its definition, so update_backtest cannot change them one by one: a SweepTick is changed with update_sweep.

The tools, in the order the assistant uses them

SweepTick settings get_sweep_settings

The settings that can be swept, so the assistant knows what it may step and how to name it. Ask: "What can we sweep on this condor?" On its own, it lists every sweepable setting with its label, type, unit and whether it is swept per leg, how a leg is named in a sweep, and your plan's limit (257: up to 256 combinations plus the core position). When the assistant names a source (a Backtick position or a bot), it also returns that source's current value for each setting, the options it offers, and whether sweeping a setting would switch a feature on that the core has off (for example sweeping a profit target on a position that has none). It needs View Backticks and results.

Preview SweepTick preview_sweep

Shows what a sweep would generate before anything is saved: how many positions, their names and their values, the step between values, your iteration allowance and, for a grid, the values on each axis. Ask: "Preview a sweep of the short put delta from 10 to 30 in steps of 2." It saves nothing and runs nothing.

The assistant previews first, every time. A create that is refused because of its source has already spent some of your change allowance, while a preview spends none of it. For a large sweep the assistant asks for a summary: the count, the step, the allowance, the first and last few positions and the axis values, with a note of how many positions were left out of the list. Previews are limited, and each also counts toward the validation limit. It needs View Backticks and results.

Create SweepTick create_sweep

Creates the SweepTick as a draft: the core position and every generated position, with the name, the source and the one or two swept ranges from the preview. Ask: "Create it." The assistant can also set the SweepTick's own fields (test period, starting balance, commission profile, skipped days, price rounding, quantity-of-one mode, the 49-contract SPX cap and the slippage and delay overrides), put your tags on it, give the generated positions a common starting name, and attach a signals file (position open and close signals, included with Backtick Professional). Notes, the position size minimums and strike conflict checking are not fields of a SweepTick.

It does not run anything. The answer gives the new SweepTick, its edit version, how many positions it holds (core included) and which one is the core. A new SweepTick always starts unsaved, on the API/MCP tab of your Library; the assistant saves it afterwards with set_backtest_library_status if you ask. A repeat of the same request within 10 minutes returns the first SweepTick instead of making another. Creates and edits are limited. It needs Create Backticks.

Update SweepTick update_sweep

Changes a draft or failed SweepTick: a new core, new ranges, a new name or starting name, its own fields, its tags, its signals file or its library section. Ask: "Widen that to 5 through 35" or "Sweep the stop instead." Changing the core or the ranges generates the positions again from scratch. The assistant sends the edit version from get_backtest and gets the next one back. Moving the SweepTick to another library section needs Change library status; everything else needs Edit Backticks and positions. A completed SweepTick is not changed: the assistant clones it, and the copy is still a SweepTick.

Running it and reading the results

After the SweepTick is created, the assistant runs it with run_backtest and waits with wait_for_backtest, exactly as for any Backtick. Running a SweepTick needs Backtick Professional and the Run Backticks permission. When it completes, the positions view of get_backtest_results is the natural read: one row per variant with its metrics, so the assistant can tell you where the best values cluster and whether the neighbours agree. For the series or breakdown views of a SweepTick the assistant names the positions it wants, because its positions are alternatives to compare rather than parts of one portfolio. Chart links and portfolio links work the same way; see Backtick Results, Charts and Portfolio Links.

Limits at a glance

Worked example: short put delta from 10 to 30 in steps of 2

You say: "Take the put spread position in my SPX Backtick and sweep the short put delta from 10 to 30 in steps of 2."

1. The assistant opens the Backtick with get_backtest to find the position, then calls get_sweep_settings with that position as the source. It sees the short delta is sweepable, that the position's current value is 15, and how to name the short leg.
2. It calls preview_sweep with that source and one range: short delta, first value 10, last value 30, interval 2. The preview answers 11 positions, named with the delta they test (10, 12, 14 and so on up to 30), a step of 2 and a limit of 257 (256 combinations plus the core).
3. It shows you the list and you say "Create and run it." create_sweep makes the SweepTick as a draft, "SPX Put Spread Delta Sweep", with 12 positions: the core at 15 delta and the 11 variants.
4. run_backtest queues it; wait_for_backtest holds until it completes.
5. The assistant reads the positions view and tells you, for instance, that total profit rises from 10 to 18 delta and falls after 22, with max drawdown growing steadily, so 16 to 20 is the stable range. It links the results page so you can see the variants side by side.

If you then say "Now do that against three stop levels too", the assistant previews a grid with a second range on the stop setting (11 deltas by 3 stops, 33 positions), and since the first SweepTick has completed, it creates a new one rather than changing it.