pdr-backend

Run bots (agents)

• Run predictoor bot - make predictions, make $
• Run trader bot - consume predictions, trade, make $

(If you're a predictoor or trader, you can safely ignore the rest of this README.)

Settings: PPSS

A "ppss" yaml file, like ppss.yaml, holds parameters for all bots and simulation flows.

• We follow the idiom "pp" = problem setup (what to solve), "ss" = solution strategy (how to solve).
• PRIVATE_KEY is an exception; it's set as an envvar.

When you run a bot from the CLI, you specify your PPSS YAML file.

pdr has basic stdout logging, but supports customisations. To customise logging, copy and edit the existing logging.yaml:

cp logging.yaml my_logging.yaml

CLI

(First, install pdr-backend first.)

To see CLI options, in console:

pdr

This will output something like:

Usage: pdr sim|predictoor|trader|..

Main tools:
  pdr sim YAML_FILE
  pdr predictoor YAML_FILE NETWORK
  pdr trader APPROACH YAML_FILE NETWORK
...

Atomic READMEs

• Get tokens: testnet faucet, mainnet ROSE & OCEAN
• Claim payout for predictoor bot
• Predictoor subgraph. Subgraph filters
• Run barge locally

Flows for core team

• Backend-dev - for pdr-backend itself
  • Local dev flow
  • VPS dev flow
  • Release process
  • Clean code guidelines
• Run dfbuyer bot - runs Predictoor DF rewards
• Run publisher - publish new feeds
• Run trueval - run trueval bot

Repo structure

This repo implements all bots in Predictoor ecosystem. Here are each of the sub-directories in the repo.

Main bots & user tools:

• predictoor - submit individual predictions
• trader - buy aggregated predictions, then trade
• sim - experiments / simulation flow
• payout - OCEAN & ROSE payout

OPF-run bots & higher-level tools:

• trueval - report true values to contract
• dfbuyer - buy feeds on behalf of Predictoor DF
• publisher - publish pdr data feeds
• analytics - analytics tools
• deployer - deployer tool
• accuracy - calculates % correct, for display in predictoor.ai webapp

Mid-level building blocks:

• cli - implementation of CLI
• ppss - implements settings
• lake - data lake and data pipeline
• subgraph - blockchain queries, complements lake
• aimodel - AI/ML modeling engine
• accuracy - to report % correct in webapp
• pred_submitter - for predictoor bots to submit >>1 predictions in 1 tx

Lower-level utilities:

• contract - classes to wrap blockchain contracts; some simple data structures
• util - function-based tools