Prompting Crypto Market Data Agents for Scans and Analysis
Summary
The document presents prompt patterns for using a crypto market data assistant. It covers broad ticker scans, candle and indicator analysis, comparisons across timeframes or venues, order book depth checks, news context, and recurring reports. Its central method is to state the asset, exchange, timeframe, lookback, comparison, and desired output clearly while letting the assistant select the data tools. It also recommends escalating from inexpensive ticker scans to candle or order book queries only when a scan identifies something worth examining.
Examples illustrate how to request relative strength, abnormal volume, slippage, and cross-source price checks. The article offers practical workflow advice, such as limiting lookback and calls when quota matters, and keeping a versioned, cost-tagged prompt library. It provides examples rather than measured evidence that the patterns improve results. Outcomes depend on data quality, tool availability, and the assistant's interpretation; indicator readings and apparent news reactions are not independently validated trading signals.
Key ideas
- State the asset, venue, timeframe, lookback, and desired output to reduce ambiguity.
- Let the agent select the appropriate market data tools unless correcting a prior tool choice.
- Use broad ticker scans first, then request candles or order book data for selected candidates.
- Compare multiple timeframes or venues when assessing trend agreement, relative strength, or liquidity.
- Constrain data requests and call counts when managing quota, and cross-check important prices.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.