Hyperliquid Grid Bot

This bot builds a two-sided limit-order grid around a center price. It places buy orders below the center, sell orders above it, and then replaces filled orders with the opposite side one grid step away.

Because this runs on perpetuals, the sell side is not just passive cash liquidation. Filled sell orders can create short exposure. That means the bot is carrying directional inventory in both breakout directions depending on what gets filled first.

So this is a perpetual grid bot, not a spot inventory balancer. The distinction matters for risk and for how you think about margin usage.

The bot uses `--center`, `--levels`, and `--spacing` to build the grid. Before placing orders, it validates that the rounded levels stay distinct, stay on the correct side of the center, and do not collapse into non-positive prices.

After launch, it monitors fills on a loop. When a buy fills it places a replacement sell above; when a sell fills it places a replacement buy below. That keeps the grid cycling while price oscillates.

Once the grid is live, the center stays fixed. The implementation does not try to cancel and rebuild the grid automatically, which avoids quietly leaving stale orders behind during a reshape.

`--center` is the grid midpoint. If omitted, the bot uses the current market price on startup. `--spacing` is a percentage of the center price, and `--levels` is the number of buy and sell steps per side.

`--size` is the size of each individual order in asset units, not USD notional. If you set `--size 0.01`, every rung uses 0.01 units of the symbol.

`--interval` controls how often the bot checks for fills. If you want to reposition the grid later, you do that manually by stopping the bot and restarting it with a new center.

The bot does best when price chops around a range and repeatedly fills both sides. It does badly when the market trends hard enough that inventory accumulates on one side and the replacement logic keeps leaning into a losing direction.

That breakout risk is real in both directions. In a strong rally, sell orders can leave you net short. In a strong selloff, buy orders can leave you with a heavy long inventory at poor average prices.

So again, the value here is a clean open-source baseline, not a promise that the grid is self-protecting in all market regimes.