Fetch live OHLCV data from a crypto exchange and run a PyneCore indicator on it in real time. No API keys needed — uses public market data.

uv run run.py

This will:

1. Fetch the last 100 closed BTC/USDT 1-minute candles from Binance
2. Run RSI on the historical data (warmup)
3. Poll every 10 seconds and process each newly closed candle, until 3 have arrived (about 3 minutes)

Only closed candles are fed to the script: the exchange also returns the candle that is still forming, whose values change until it closes.

The runner works in live mode (ScriptRunner(..., live=True)). The generator yields the history, then LIVE_TRANSITION, then the new candles, and each new candle's RSI is printed right after the candle arrives. Without live mode the runner treats the feed as history: it reads one bar ahead to know which bar is the last, so every result would come one candle late.

Requires PyneCore 6.10.7 or newer (live=True).

Edit the constants at the top of run.py:

EXCHANGE = "binance"          # any CCXT-supported exchange
SYMBOL = "BTC/USDT"           # trading pair
TIMEFRAME = "1m"              # candle timeframe (CCXT notation)
PERIOD = "1"                  # the same timeframe as a Pine Script period ("60" = 1h, "D" = 1 day)
HISTORY_BARS = 100            # closed historical bars for indicator warmup
LIVE_BARS = 3                 # number of new closed bars to wait for
POLL_INTERVAL_SEC = 10        # seconds between polls

CCXT supports 100+ exchanges. Just change the EXCHANGE variable:

EXCHANGE = "coinbase"         # Coinbase
EXCHANGE = "kraken"           # Kraken
EXCHANGE = "bybit"            # Bybit
EXCHANGE = "okx"              # OKX

• Replace polling with websocket streaming for real-time updates: yield OHLCV(..., is_closed=False) for each price update of the forming candle and is_closed=True when it closes
• Add multiple indicators by creating more scripts
• Connect to a trading bot (see 05-freqtrade-indicators/ for a complete example)