Quant Agent: RSI and EMA technical analysis signals

The Quant Agent downloads OHLC candlestick data from Binance and applies two indicators — Wilder RSI and an exponential moving average — to produce a directional verdict for any trading pair. Stage 2/6 in the full Elyra pipeline runs it against BTCUSDT, but you can point it at any symbol Binance supports.

CLI usage

Pass the trading pair as a positional argument:

python3 agents/quant_agent.py BTCUSDT
python3 agents/quant_agent.py ETHUSDT
python3 agents/quant_agent.py SOLUSDT

If you omit the argument, the agent defaults to BTCUSDT.

Python usage

from agents.quant_agent import analyze_market

result = analyze_market(symbol="BTCUSDT", interval="1d", limit=100)
# Returns: {"trend": "bullish", "rsi": 58, "signal": "possible breakout"}

`analyze_market()`

Downloads up to limit candlesticks for symbol at the given interval, computes RSI and EMA, and returns a structured signal dictionary. Signature

def analyze_market(
    symbol: str = "BTCUSDT",
    interval: str = "1d",
    limit: int = 100,
    rsi_period: int = 14,
    ema_period: int = 20,
) -> dict[str, str | int]

Parameters

string

default:"BTCUSDT"

Binance trading pair in uppercase (e.g. "BTCUSDT", "ETHUSDT", "SOLUSDT"). Any valid Binance spot or futures symbol is accepted.

string

default:"1d"

Candlestick interval. Accepts any Binance kline interval string — common values are "1m", "5m", "1h", "4h", "1d", "1w".

number

default:"100"

Number of candlesticks to fetch from Binance (maximum 1000). Must be greater than max(rsi_period, ema_period) + 1; otherwise the function returns "insufficient data".

number

default:"14"

Lookback window for the Wilder RSI calculation.

number

default:"20"

Lookback window for the exponential moving average. Trend is determined by comparing the last close price against the last EMA value.

Return value

Returns dict[str, str | int].

string

required

Directional bias derived by comparing the most recent close against the EMA. One of "bullish" (close ≥ EMA) or "bearish" (close < EMA). Returns "neutral" when there is insufficient data.

number

required

Most recent RSI value rounded to the nearest integer, in the range 0–100. Returns 50 when there is insufficient data.

string

required

Human-readable trading signal derived from RSI and trend. Possible values:

Value	Condition
`"overbought"`	RSI ≥ 70
`"oversold"`	RSI ≤ 30
`"possible breakout"`	Trend is bullish and 50 < RSI < 70
`"possible breakdown"`	Trend is bearish and 30 < RSI < 50
`"uptrend"`	Trend is bullish and RSI ≤ 50
`"downtrend"`	Trend is bearish and RSI ≥ 50
`"insufficient data"`	Not enough candles to compute indicators

Example response

{
  "trend": "bullish",
  "rsi": 58,
  "signal": "possible breakout"
}

Insufficient data response

If limit is smaller than max(rsi_period, ema_period) + 1, the function returns safe defaults instead of raising an exception:

{
  "trend": "neutral",
  "rsi": 50,
  "signal": "insufficient data"
}

Signal logic reference

The signal is computed in two steps: first determine trend from price vs. EMA, then apply RSI thresholds.

# Trend: last close vs. last EMA value
trend = "bullish" if last_close >= last_ema else "bearish"

# Signal: RSI thresholds take priority
if rsi >= 70:   signal = "overbought"
elif rsi <= 30: signal = "oversold"
elif trend == "bullish" and 50 < rsi < 70: signal = "possible breakout"
elif trend == "bearish" and 30 < rsi < 50: signal = "possible breakdown"
elif trend == "bullish": signal = "uptrend"
else:           signal = "downtrend"

No API key is required. Binance kline data is publicly accessible.

Research Pipeline

Prediction Markets

Quant Agent: RSI and EMA technical analysis signals

CLI usage

Python usage

`analyze_market()`

Parameters

Return value

Example response

Insufficient data response

Signal logic reference

​CLI usage

​Python usage

​analyze_market()

​Parameters

​Return value

​Example response

​Insufficient data response

​Signal logic reference

CLI usage

Python usage

`analyze_market()`

Parameters

Return value

Example response

Insufficient data response

Signal logic reference