Momentum Scanner

Trend refers to the overall direction of price movement (up, down, or sideways), while momentum measures the speed and strength of that movement. A stock can be in an uptrend but have weak momentum (slow grinding up) or strong momentum (fast sharp moves up). Momentum tells you how powerful the trend is, not just its direction. Think of trend as the direction of travel and momentum as the speed.

Start with a manageable universe of 50-100 stocks, such as FTSE 100. This gives you enough variety to find opportunities while not being overwhelming. As you gain experience interpreting scanner results, you can expand to larger universes like the FTSE 350 (~350) or FTSE All-Share. Scanning too many stocks initially can lead to analysis paralysis.

A momentum score of zero indicates neutral momentum - the stock is neither showing strong upward nor downward momentum. This could mean the stock is consolidating in a range, has mixed signals across timeframes, or is simply moving in line with the market. Momentum traders typically avoid zero-score stocks and focus on strong positive (>+50) or negative (<-50) scores.

Volume represents the conviction behind price moves. High volume during a momentum move means many participants are driving the price, making the move more likely to sustain. Low volume momentum is often 'fake' - prices may have moved but without broad participation, they can reverse quickly. Always look for volume ratio above 1.5x average to confirm momentum validity.

Momentum scanning works for multiple timeframes. For positional trading, use longer period settings (20-day and 50-day ROC emphasis) and scan less frequently (once daily). For day trading, use shorter periods (5-day emphasis) with real-time or 1-minute scanning. The scanner is flexible - adjust timeframe weights based on your trading horizon.

False breakouts are inevitable, but you can reduce their impact by: (1) Requiring volume confirmation >1.5x average, (2) Waiting for price to close above breakout level rather than just intraday spike, (3) Checking that acceleration is positive not negative, (4) Using stop losses consistently. Accept that 30-40% of momentum signals may fail, but the winners should more than compensate with proper risk management.

The default weights work well for most traders, but you can customize based on your style. UK markets have some unique characteristics - institutional flows and disclosed short interest influence price momentum, so you might create a custom factor for short-squeeze potential. Also, FCA short-position data (specific to the UK and EU) can be incorporated. Start with defaults, then experiment with changes only after gaining experience.

Use sector momentum in two ways: (1) Top-down: First identify strongest momentum sectors, then find strongest stocks within those sectors. Trading strong stocks in strong sectors gives you 'wind at your back.' (2) Relative: Compare stock's momentum to its sector average - stocks significantly outperforming their sector often continue to lead. Avoid buying stocks in weak sectors even if individual momentum looks good.

For swing trading (holding 2-10 days), scanning every 15-30 minutes during market hours is optimal. This catches momentum shifts in time to act without the noise of tick-by-tick updates. Run a comprehensive scan at market open (8:15 AM) and before close (4:00 PM), with intermittent checks during the day. Daily scanning is insufficient as you'll miss intraday momentum developments.

Divergences are warning signals, not immediate action signals. When scanner flags a bearish divergence: (1) Do not initiate new longs in that stock, (2) Tighten stops on existing positions, (3) Wait for price confirmation (breaking support) before shorting. Divergences can persist for days or weeks before price reacts, so patience is essential. Use divergence as a filter to avoid low-probability momentum trades.

Monitor Information Coefficient (IC) of your momentum factors monthly. A sustained decline in IC over 3-6 months indicates decay. Causes include: other traders discovering the same factor, market regime changes, or data source changes. Response strategies: (1) Reduce weight of decaying factor, (2) Combine with orthogonal factors to maintain edge, (3) Develop new proprietary factors to replace decayed ones. Factor decay is inevitable - continuous research and development is required for sustained edge.

Gradient boosting methods (XGBoost, LightGBM) typically outperform deep learning for momentum prediction due to tabular data structure and interpretability needs. Key success factors: (1) Feature engineering matters more than model complexity - include UK-specific features like disclosed short interest and institutional ownership change, (2) Use time-series cross-validation strictly - random CV causes severe overfitting, (3) Regularize heavily and use early stopping, (4) Monitor IC decay post-deployment and retrain monthly. Start simple with logistic regression baseline before adding complexity.

Build a regime detection layer using: (1) Momentum breadth indicator (% stocks with positive momentum), (2) VFTSE/volatility level, (3) Momentum autocorrelation over 20 days. Classify into 4 regimes: Strong Trend (breadth >70%, autocorr >0.3), Choppy (breadth 40-60%, autocorr <0.1), Crisis (VFTSE >25, breadth <30%), Recovery (breadth rising from lows). Map each regime to parameter presets - threshold levels, holding periods, position sizes. Backtest each regime's optimal parameters separately.

Key considerations: (1) Regulatory: If providing personal recommendations on investments, FCA authorisation under FSMA is required; information-only tools that provide scores without personal recommendations may fall outside the perimeter - take legal advice and comply with financial promotion rules. (2) Track record: Build 12+ months verified track record before launch. (3) Infrastructure: Cloud-based servers with redundancy, 99.9% uptime target, disaster recovery. (4) Customer acquisition: Content marketing (YouTube, blogs) + broker partnerships work well in the UK. (5) Pricing: £20-150/month for retail, enterprise pricing for institutions. (6) Support: English-language support across UK time zones.

Order flow analysis enhances momentum scanning by revealing institutional activity: (1) Trade imbalance: Track aggressive buying (trades at ask) vs selling (trades at bid) - sustained imbalance supports momentum direction. (2) Large trade detection: Flag trades >£100k as potential institutional orders - their direction often predicts near-term momentum. (3) Order book depth analysis: Growing bid depth vs ask depth indicates accumulation. For UK markets, LSE and vendor feeds provide tick data with sub-second to ~500ms delay for retail, usable for 1-5 minute aggregated order flow signals. Pure HFT speeds are inaccessible to retail, but aggregated order flow signals remain valuable for entry timing.