> For the complete documentation index, see [llms.txt](https://arbital-1.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://arbital-1.gitbook.io/docs/automated-strategies/inventory-management.md).

# Inventory Management

## What is Inventory?

Inventory = how much of the asset the bot is currently holding (long or short).

**Example:**

* If the bot buys more than it sells → it becomes long
* If it sells more than it buys → it becomes short

## How Arbital Bots Manage Inventory Automatically

The bot actively prevents runaway exposure by:

* Reducing buys when inventory is too long
* Reducing sells when inventory is too short
* Adjusting spreads dynamically
* Stopping one-sided trading if limits are hit

You don't need to manually rebalance - the bot does this continuously.

## Directional Bias

Directional bias allows you to tilt execution toward buying or selling earlier in time, while still running a two-sided market making strategy.

* The bot will always place both buy and sell orders.
* Bias only affects **when** buys and sells happen - not **whether** they happen.

## Neutral

* Buy and sell orders are distributed evenly over time
* Inventory stays close to zero
* No directional exposure

This is pure market making.

## Long Bias

* Buy orders are executed earlier than sell orders
* Sell orders are executed later, not removed
* Inventory becomes temporarily long

Long bias causes the bot to build a long position early, then even out over time through continued market making.

## Short Bias

* Sell orders are executed earlier than buy orders
* Buy orders are executed later, not removed
* Inventory becomes temporarily short

Short bias causes the bot to build a short position early, then even out over time through continued market making.

## Important Notes

* The strategy always market makes on both sides
* Total buy and sell volume still converges over the full run
* Bias increases temporary inventory and margin usage
* Margin requirements increase by up to 20% at higher bias levels