Copper Momentum Strategy

LMEmini Copper (the 5-tonne contract) is better for most retail traders. The standard LME Copper lot is 25 tonnes with roughly US$25,000+ overnight margin and a US$6.25 tick value (US$25 per US$1/tonne move). LMEmini is 5 tonnes with roughly US$5,000+ margin and a US$1.25 tick value (US$5 per US$1/tonne move). Start with LMEmini for better position-sizing flexibility. Many UK retail traders instead use copper spread bets (priced in GBP per point) or CFDs for easier access and, for spread bets, tax-free profits - though leverage risk is high.

Copper trends are smoother and more sustained than gold because copper is driven by industrial demand, not speculation. Gold has safe-haven spikes and is more sensitive to USD/rates. Copper momentum tends to persist longer.

For momentum trading, use RSI 50 as your key level. RSI above 50 = positive momentum (stay bullish). RSI below 50 = negative momentum (stay bearish). Don't fade copper just because RSI is above 70 - in trends, it can stay there.

Approximately 11:40-17:00 London time is optimal. This window covers the Ring (where the LME Official Price - the global benchmark - is set in the second morning session) and the afternoon overlap with COMEX in New York, providing the highest volume and most significant moves. Avoid the thin overnight electronic session (01:00-08:00) and the post-Ring taper (17:00-19:00).

ADX tells you IF a trend exists, not the direction. ADX > 25 means a tradeable trend exists. ADX < 20 means no trend (ranging). Use ADX to filter - only take momentum trades when ADX confirms a trend.

Create a scoring system: RSI position/direction (0-3 points), MACD position/histogram (0-3 points), ADX strength (0-3 points), Trend/EMA (0-3 points), Volume (0-2 points). Trade when total score >= 10/15 for high confidence.

Divergence is when price and momentum indicator disagree (e.g., price makes higher high, RSI makes lower high). Use it for exits and stop tightening, not immediate reversals. Wait for price confirmation before trading the reversal.

Pullbacks are generally preferred - better entry price, confirmed trend, better R:R. But breakouts work for strong signals with volume surge. Use breakouts for initial entry in new trends, pullbacks for adding to established trends.

The LME Official Price is the global benchmark, but COMEX (New York) and SHFE (Shanghai) provide valuable cross-confirmation. COMEX tracks the LME almost exactly via arbitrage, so if COMEX is rallying your LME long has higher conviction. SHFE reflects Chinese demand and often leads sentiment in the European morning. Use them for confirmation, not contradiction.

Use ATR-based sizing: Risk Amount / (1.5 x ATR x point value) = Lots, where the point value is US$25/tonne for LME Copper and US$5/tonne for LMEmini. Then adjust for signal quality (score/15) and regime (weak trend = 75%). This normalizes risk and allocates more to high-conviction setups.

Score each component: RSI (0-3 based on level, direction, rate of change), MACD (0-3 based on position, histogram, expansion), ADX (0-3 based on level, direction), Trend (0-3 based on EMA position, slope, breakout), Volume (0-2). Total 15 points, trade >= 10.

Higher TFs (weekly/daily) determine direction - only trade that way. Middle TFs (4-hour) show structure. Lower TFs (1-hour/15-min) provide entry timing. Score each TF's momentum, trade when 5+ of 7 points align across timeframes.

Factor models decompose returns into components: Price momentum (40% weight), Volume momentum (20%), Cross-asset momentum (20%), Fundamental momentum (20%). Calculate normalized scores for each, combine with weights. Score > 0.6 = strong signal.

Classify using ADX: Strong trend (>35), Weak trend (25-35), Mean reverting (<20), Transitional (rising from <20). Adjust parameters per regime - strong trend uses wide stops/full position, weak trend uses moderate stops/75% position.

Log every trade with: regime, score, entry/exit, result. Analyze performance by regime type, signal score, timeframe, and session. Look for pattern - which regimes/scores perform best? Optimize weights and thresholds based on data, not intuition.