bptf-autopricer

An open-source solution that generates item prices for selected items by utilising listing data from backpack.tf. Each price created is a result of the meticulous evaluation of both the underlying data and the actual prices, incorporating multiple checks and balances to counteract potential price manipulation.

Features

• Automated Pricing: Automatically generates item prices using both real-time and snapshot backpack.tf listing data, ensuring a profit margin and performing various sanity checks.

• Baseline Comparison: Compares generated prices against those from Prices.tf - disregarding prices that go over percentage thresholds configured.

• Trusted/Blacklisted Steam IDs: Prioritises listings from trusted bots, and filters out untrusted bots when calculating prices. Fully configurable.

• Excluded Listing Descriptions: Filters out listings with descriptions containing configured keywords. Useful for removing listings from calculations that include special attributes, such as spells.

• Filters Outliers: With a sufficient number of listings, filters out outliers from the mean. Removes listings with prices that deviate too much from the average.

• API Functionality:

  • Add and Delete Items: The API can be used to add or remove items for the auto pricer to track.
  • Retrieve Prices: Prices can be requested and retrieved through the API.

• Socket.IO Server:

  • Emits item prices to any listeners using a Socket.IO server.
  • Prices are stored and emitted in a format fully supported by the TF2 Auto Bot custom-pricer interface.

Requirements

• Install dependencies by running npm install in the project directory with package.json.
• A PostgreSQL database is required. You have two options in setting this up.
  1. Keep the default options for the database in config.json and follow along with the tutorial provided in INITIALIZE-DB.md to satisfy all the database requirements.
  2. Create a database and schema according to your preferences and specify them in config.json. A table named listings must be created using the following SQL statement:

CREATE TABLE your_schema_name.listings (
    name character varying NOT NULL,
    sku character varying NOT NULL,
    currencies json NOT NULL,
    intent character varying NOT NULL,
    updated bigint NOT NULL,
    steamid character varying NOT NULL,
    CONSTRAINT listings_pkey PRIMARY KEY (name, sku, intent, steamid)
);

Regardless of the option you choose. Make sure you have properly specified the database name, schema, user, and other relevant details that you are using in the config.json file.

Configuration

To configure the application you need to specify the values for all the fields in config.json.

{
    "bptfAPIKey": "your bptf api key",
    "bptfToken": "your bptf token",
    "steamAPIKey": "your steam api key",
    "database": {
        "schema": "tf2",
        "host": "localhost",
        "port": 5432,
        "name": "bptf-autopricer",
        "user": "postgres",
        "password": "database password"
    },
    "pricerPort": 3456,
    "maxPercentageDifferences": {
        "buy": 5,
        "sell": -8
    },
    "alwaysQuerySnapshotAPI": true,
    "excludedSteamIDs": [
        "76561199384015307"
    ],
    "trustedSteamIDs": [
        "76561199110778355"
    ],
    "excludedListingDescriptions": [
        "exorcism",
    ]
}

The majority of these fields are self-explanatory. I will explain the ones that may not be.

maxPercentageDifferences

Contains two fields, buy and sell. These values represent the maximum difference from the baseline (prices.tf) you will accept before a price is rejected. Adjust the maximum percentage differences for buy and sell according to your preferences.

• A higher buy percentage means that the auto pricer is willing to buy items for a higher price than prices.tf.
• A lower sell percentage means that the auto pricer is willing to sell items for a lower price than prices.tf.

The default in the config is the settings I used while using this auto pricer with a trading bot.

"maxPercentageDifferences": {
  "buy": 5,
  "sell": -8
}

excludedSteamIDs

A list of Steam ID 64s that you don't want to use the listings of in pricing calculations. This is useful for stopping bad actors that you know of from attempting to manipulate the prices created. It's important to note that this is not a fool-proof solution and because of this other methods are also used to reduce the risks of price manipulation.

"excludedSteamIDs": [
    "76561199384015307"
]

trustedSteamIDs

A list of Steam ID 64s used to prioritise listings owned by these IDs over others during pricing calculations. This feature is beneficial when you want to give preference to bots or users that consistently provide accurate pricing data.

"trustedSteamIDs": [
    "76561199110778355"
]

excludedListingDescriptions

A list of descriptions that, when detected within a listing's details, causes the listing to be disregarded and not used during pricing calculations. This is useful for excluding listings involving items with 'special attributes,' such as spells, when calculating prices. Leaving such listings in the calculations can risk affecting the average price in unexpected ways.

"excludedListingDescriptions": [
    "exorcism",
]

alwaysQuerySnapshotAPI

This setting determines whether the pricer should consistently call the snapshot API for each item during the pricing process, regardless of the number of listings available in the database. By default, this setting is set to true.

Behavior:

• true: The pricer always calls the snapshot API for every item, ensuring the most up-to-date information. However, this may result in slower pricing processes due to the API rate limits.

• false: The pricer avoids unnecessary API calls if there are already a suitable number of listings in the database to generate a price. This speeds up the pricing process significantly as it bypasses the need to call the snapshot API. However, there are trade-offs to consider:

Trade-offs:

1. Accuracy vs. Speed: By setting alwaysQuerySnapshotAPI to false, the pricing process becomes faster but may use a smaller pool of listings to calculate prices. The logic ensures that if there are at least 10 buy listings and 1 sell listing, no further data retrieval is attempted.

2. Timeliness of Data: When the API is not called every time, there's a risk of using listings with prices that are up to approximately 35 minutes old, in the worst-case scenario.

Benefits:

• Improved Performance: Setting alwaysQuerySnapshotAPI to false makes the pricer more efficient, especially when dealing with large item lists, as it avoids the minimum 2-second wait time per item.

Consider your requirements for pricing accuracy, speed, and data freshness when configuring this setting.

"alwaysQuerySnapshotAPI": true

API Routes & Socket IO

The socket io server will emit events called 'price' with an item object as the value. The item objects are structured like the following:

{
  "name": "Strange Australium Minigun",
  "sku": "202;11;australium",
  "source": "bptf",
  "time": 1700403492,
  "buy": {
    "keys": 25,
    "metal": 21.33
  },
  "sell": {
    "keys": 26,
    "metal": 61.77
  }
}

• Note that the same JSON structure is used when an item or an array of items is returned by the API.
• This JSON format is fully compatible with what TF2 Auto Bot requires for a custom pricer implementation.

Now I'll highlight the different API routes you can make queries to, and what responses you can expect to receive.
Please note that both the Socket IO server and API run locally (localhost) on the port defined in config.json.

GET /items/:sku

Retrieves a particular item object from the pricelist using the Stock Keeping Unit (SKU) provided. Item object returned contains the prices for the item.

Request:

• Parameters:
  • sku (String): The Stock Keeping Unit of the item. E.g., Mann Co. Supply Crate Key has an SKU of 5021;6.

Response:

• Success (200):
  • JSON item object containing information about the item, including the prices.
• Failure (404):
  • If the requested item is not found in the pricelist.
• Failure (400):
  • Where 5021;6 is the SKU provided and there is an issue with fetching the key price from Prices.tf.

GET /items/

Retrieves the entire pricelist.

Response:

• Success (200):
  • JSON object containing the entire pricelist. An array of JSON item objects.
• Failure (400):
  • If there is an issue loading the pricelist.

POST /items/:sku

An endpoint that returns a status code of 200 for each request. Exists so there's no issue in integrating with TF2 Auto Bot.

Request:

• Parameters:
  • sku (String): The Stock Keeping Unit of the item.

Response:

• Success (200):
  • JSON object indicating the SKU.

POST /items/add/:name

Adds the item to the list of items to auto price.

Request:

• Parameters:
  • name (String): The name of the item to add.

Response:

• Success (200):
  • Item successfully added. Will now be automatically priced.
• Failure (400):
  • If the item already exists in the item list.

POST /items/delete/:name

Deletes an item from the list of items to automatically price.

Request:

• Parameters:
  • name (String): The name of the item to delete.

Response:

• Success (200):
  • Item successfully deleted.
• Failure (400):
  • If the item does not exist in the item list.

Running the Auto Pricer

Once all the requirements have been met, and you have provided the values required in config.json, simply run:

node bptf-autopricer.js

If you want to use PM2 to run the application even after closing the terminal run:

pm2 start bptf-autopricer.js

Adding Items to Price.

• Using the API Routes: Send requests to the relevant route specified in the API section.
• Manual Additions: Add items to price manually to item_list.json. Make sure to use the valid format shown.

Each item name given should be the same as the one used for its listings on backpack.tf. For example, Non-Craftable Tour of Duty Ticket NOT Uncraftable Tour of Duty Ticket.