Strategy reference

The Strategy tab (owner-only) is where you build, tune, test, and publish the algorithm your bot runs. This section is a reference for each surface in that workspace.

Your strategies and their settings are yours. Parachute runs your bot on your own Kalshi account using credentials you provide; it never takes custody of your funds or trades on its own behalf. See Money safety.

The Strategy tab

The tab has two modes, switched with the toggle at the top right:

  • Tune — build and edit the strategy (this page and its sub-pages).
  • Backtest — run the simulation engine over historical data. See Backtests.

Tune opens on the strategy roster: a card for each strategy your account can run. The strategy currently deployed to your bot is marked Active. Selecting a card shows what that strategy does and its settings.

Tune: canvas and table

The Tune surface has two views, switched with a Canvas ⇄ Table control. Both views edit one shared draft — switching between them never loses your edits.

  • Canvas — the strategy drawn as a visual pipeline. A component palette lets you pick the pricing model and add or remove admission gates; clicking a stage opens its parameters.
  • Table — every strategy parameter in a single list. This is the same draft as the canvas.

Schedule (the trading calendar) and money-safety limits like the daily loss limit are set in Bot Control. The entry window is edited on the strategy's composition surface (it is an Entry-stage filter), and the ticker blacklist has its own instant editor. Deploying from Tune publishes the whole configuration.

Save vs Deploy

Editing follows a lock → edit → save → deploy lifecycle:

  • Unlock and Edit to open the draft for changes.
  • Save stores your changes as a draft. A draft does not change what the live bot is doing.
  • Deploy publishes the draft as a new version. The bot picks up a newly deployed version on its next loop.

The evidence gate

Some changes are structural — swapping the pricing model, adding or removing admission gates, or deploying a brand-new strategy. Deploying a structural change asks for evidence first:

  • Backtested — a completed backtest of this exact composition is linked. The Deploy button shows a "Backtested ✓" note, with an honest caveat if the run relied mostly on approximation.
  • Deploy untested (audited) — you can override and deploy without a backtest by giving an explicit reason. The override is recorded.
  • Blocked — until you do one of the above, Deploy stays blocked for a structural change.

Simple dial tweaks that do not change the strategy's structure keep the normal Deploy behavior and do not require a backtest.

Named strategies and versions

Each strategy is a product card with a name, description, and the markets it trades. Every time you publish, the strategy records an immutable version in its history, newest first. Published versions cannot be edited after the fact — publishing again creates a new version rather than overwriting an old one.

In this section

| Page | What it covers | |---|---| | Strategy components | The pipeline, the model calculator, and admission gates | | Trading calendar | When and how the bot trades each hour (slots and modes) | | Deep-ITM tuning | The Deep-ITM ruleset and its stricter thresholds | | Maximum edge ceiling | The upper limit on modeled edge (a winner's-curse guard) | | ETH-specific overrides | Optional ETH-only settings | | Take-profit & stop-loss (Exit Bracket) | Per-position exit rules | | Backtests | Simulating a strategy over historical data | | What-if replay | A fast, free preview of a threshold change |