| title | 2020-05-22 | |
|---|---|---|
| description | This overview highlights Kelp GUI v1—how it simplifies launching and managing market-making bots on the SDEX, from bot creation and price feeds to advanced safeguards and clean shutdown. | |
| authors | john | |
| tags |
|
import YouTube from "@site/src/components/YouTube";
Nikhil Saraf introduces Kelp GUI v1, the desktop interface for Kelp—an open-source trading bot originally built for the Stellar decentralized exchange (and later extended to many centralized exchanges via the CLI). The focus of the talk is how the GUI lowers the barrier for non-technical liquidity providers by turning configuration-heavy market-making into a guided workflow: generate a starter bot, or create a custom one by selecting assets, setting spreads and levels, and wiring up price feeds.
He explains how pricing in the buy/sell strategy works by composing a numerator and denominator feed (for example, valuing XLM via an exchange feed while treating a “coupon” asset as a fixed $1). The GUI displays the derived mid price and lets users define multiple “levels” around that mid, which translate into ladders of bids and asks. A live demo shows creating two bots on testnet, connecting a fiat price feed (Currencylayer), and using “advanced” price adjustments (offsets) to intentionally misprice one bot so it trades through the other—illustrating how bots place, cross, and refresh offers over time.
Nicholas also emphasizes operational hygiene: if you quit without stopping a bot, its offers can remain on the orderbook, potentially filling unexpectedly. Stopping a bot should both halt the process and cancel outstanding offers so the market doesn’t retain stray bids/asks after you’re done.
- What Kelp is (CLI + strategies in Go) and why the GUI matters for first-time market makers on the SDEX.
- Bot setup workflow: secret key handling, asset/issuer selection, and the “starter bot” vs custom configuration paths.
- Price feeds and derived pricing: composing numerator/denominator sources, sanity-checking quotes, and understanding how the bot refreshes prices on an interval.
- Market-making controls: spreads, multi-level ladders, order sizing, and how thin/no markets affect displayed spread.
- Advanced safeguards and tuning: correction knobs (offset % / offset), multipliers, and preventing bad quoting when price data drifts.
- Safe shutdown: stopping bots to cancel offers so you don’t leave live orders behind.