Trading Signals

Overview

The Trading Signals module generates eight distinct algorithmic signals derived from real-time and historical NEM data. These signals are informational — they highlight potential market opportunities based on quantitative rules — and are designed to support, not replace, trader judgement.

Signal Types

1. Momentum Signal

The momentum signal identifies regions where prices have sustained upward or downward movement:

Momentum_Score = (Price_now - Price_n_periods_ago) / Price_volatility_n_periods

Parameters:

Short window: 6 periods (30 minutes)
Long window: 48 periods (4 hours)
Signal trigger: |Momentum_Score| > 2 standard deviations

Interpretation: A positive momentum score suggests continued price strength. A negative score suggests mean reversion may be due.

2. Mean Reversion Signal

Based on the historically strong mean-reverting behaviour of NEM spot prices:

Z_score = (Price_now - Rolling_mean_24h) / Rolling_std_24h
Signal = "BUY" if Z_score < -2 else "SELL" if Z_score > 3 else "NEUTRAL"

The asymmetric thresholds (−2 vs +3) reflect the right-skewed distribution of NEM prices — spikes are more extreme than troughs.

3. Weather-Adjusted Demand Signal

Compares actual demand to weather-expected demand:

Expected_Demand = f(Temperature, DoW, PublicHoliday, Solar_Output)
Demand_Anomaly = Actual_Demand - Expected_Demand
Signal_Threshold = 1.5 × Residual_StdDev

When actual demand exceeds weather-adjusted expectations, it suggests an unmodelled demand driver (major event, forecast error) that could support elevated prices.

4. Renewable Ramp Signal

Identifies rapid changes in renewable output that historically precede price movements:

Wind_Ramp = Wind_Output_t - Wind_Output_(t-30min)   [MW]
Solar_Ramp = Solar_Output_t - Solar_Output_(t-30min) [MW]

Signal = "PRICE_RISE_RISK" if (Wind_Ramp < -300) or (Solar_Ramp < -150)
Signal = "PRICE_FALL_RISK" if (Wind_Ramp > 500) or (Solar_Ramp > 300)

Rapid negative renewable ramps (wind dropping) in SA1 and VIC1 have historically preceded significant price spikes.

5. FCAS Arbitrage Signal

Identifies opportunities where FCAS prices are elevated relative to the cost of providing FCAS:

FCAS_Opp_Score = max(0, FCAS_Price_service - Opportunity_Cost_MWh)

For a battery providing RAISE6SEC:

FCAS revenue: enabled_mw × price ($/MW/h)
Opportunity cost: energy not dispatched × spot_price
Net: if FCAS_revenue > energy_opportunity_cost, FCAS preferred

6. Interconnector Basis Signal

Identifies price divergence between regions that share an interconnector, subject to congestion:

Basis = Price_Region_A - Price_Region_B
Congestion_Flag = (Interconnector_Flow / Interconnector_Limit > 0.95)

Signal = "ARBITRAGE_POSSIBLE" if (|Basis| > 50) and NOT Congestion_Flag
Signal = "CONGESTION_PREMIUM" if (|Basis| > 100) and Congestion_Flag

7. Demand Shock Signal

Detects sudden large demand changes not explained by weather or time-of-day patterns:

Demand_Change_30min = Demand_t - Demand_(t-6_intervals)  [MW]
Normalised_Change = Demand_Change_30min / Region_Peak_Demand

Signal = "DEMAND_SHOCK" if |Normalised_Change| > 0.05  (5% of peak demand in 30 min)

8. Time-of-Day Pattern Signal

Exploits statistically significant intraday price patterns:

Morning ramp: 07:00–08:30 AEST — demand rise, solar not yet offsetting
Solar trough: 12:00–14:00 AEST in summer — rooftop solar suppresses prices
Evening peak: 17:30–19:00 AEST — solar drops, demand peaks
Overnight trough: 01:00–05:00 AEST — lowest demand, coal baseload excess

Signal Dashboard

Signals Feed (`/middle-office/trading-signals`)

Live feed of all active signals across all 8 types
Region filter to focus on specific market
Signal strength indicator (1–5 stars)
Historical hit rate: percentage of times this signal type was followed by the predicted price move

Screenshot: Trading signals feed showing 3 active signals with strength, region, signal type, and historical hit rate.

Signal Backtesting (`/middle-office/signals-backtest`)

Test any signal’s historical performance over configurable lookback period
P&L simulation: if you had traded every signal of a given type, what would the outcome be?
Risk-adjusted return: Sharpe ratio, maximum drawdown, win rate

API Endpoints

# All current signals
GET /api/trading-signals/latest

# Specific signal type
GET /api/trading-signals?type=mean_reversion&region=SA1

# Signal history
GET /api/trading-signals/history?type=momentum&region=VIC1&start=2025-03-01

# Backtest signal
POST /api/trading-signals/backtest
{
  "signal_type": "renewable_ramp",
  "region": "SA1",
  "lookback_days": 90,
  "trade_horizon_minutes": 30
}