Unusual Volume Scanner

Daily volume is available from TMX Money and most financial websites/trading platforms; because trading is fragmented across marketplaces, prefer consolidated volume. For short-sale data, CIRO publishes the Short Sale Trading Statistics Summary twice monthly (showing each security's short-sale volume as a percentage of total trading) and the Consolidated Short Position Report (CSPR) semi-monthly. For real-time volume during market hours, use your broker's platform (Questrade, Interactive Brokers, TD Direct, etc.) or sites like TMX Money, TradingView, or Stockwatch. Note that the short-sale summary is published after the period, not intraday.

A good starting threshold is 2x the 20-day average volume. This means if a stock typically trades 500,000 shares daily and today it traded 1,000,000+ shares, it qualifies as unusual. For more significant signals, look for 3x average or higher. Additionally, check the short-sale percentage - if it is well below the stock's average (a long-demand tilt), this adds conviction to a bullish read. Start with these thresholds and adjust based on your experience with what produces actionable signals for your trading style.

No, unusual volume alone is not a buy signal. Volume tells you something is happening, but not what or whether it's positive. Always investigate why volume is unusual: Is there news? Is it a technical breakout? Is it sector-wide or stock-specific? Also check the price direction and short-sale percentage. High volume with rising price and a low short-sale percentage is more bullish than high volume with falling price and a high short-sale percentage. Use unusual volume as a screening tool to identify stocks worth investigating, then apply fundamental and technical analysis before deciding to trade.

A high short-sale percentage with high volume means a large share of the selling is short-marked. This can reflect bearish positioning, but very often it is benign - market makers, ETF arbitrageurs and inter-listed (TSX/US) arbitrageurs short constantly as part of neutral strategies, and CIRO even uses a 'short-marking exempt' designation to filter this non-directional flow. So it is not automatically bearish. Investigate: is the name inter-listed or ETF-heavy (likely arbitrage), or is it a less-liquid stock with directional short interest? Compare to the security's own average before drawing conclusions.

On the Montreal Exchange, monthly equity and index options expire on the third Friday and index futures (SXF) expire quarterly (Mar/Jun/Sep/Dec); some products have weekly options. Volume typically rises as traders roll over or close positions, and this expiry-related volume is mechanical rather than informative about future direction. For inter-listed names, US expiries (also the third Friday) and quarterly 'quad-witching' add to the effect. When analyzing volume on or near expiry days, be cautious about interpreting it as a directional signal; compare to previous expiry days for context.

RVOL calculation requires building an intraday volume profile. Steps: (1) Collect 10-20 days of intraday volume data in fixed intervals (e.g., 15-minute buckets), (2) For each bucket (9:30-9:45, 9:45-10:00, etc.), calculate the average volume, (3) Calculate expected cumulative volume at any time by summing averages from open to that time, (4) RVOL = Actual cumulative volume / Expected cumulative volume. Example: If by 11 AM ET you expect 200,000 shares (based on historical averages) and actual is 400,000, RVOL = 2.0. An RVOL of 1.5+ indicates elevated activity given the time of day.

Accumulation and manipulation can look similar on the surface (rising volume and price) but have distinguishing characteristics. Accumulation signs: Gradual volume increase over weeks, a falling short-sale percentage (genuine long buying), institutional names in block/cross trades, volume higher on up days than down days, price base forming. Manipulation signs: Sudden volume explosion in an illiquid stock, volume with no clear arbitrage rationale, unknown counterparties, promotional activity on social media or paid stock promotions, volume concentrated in few large trades, patterns too regular. If in doubt, avoid the stock - the risk of being caught in manipulation outweighs potential gains.

When a Single-Stock Circuit Breaker (SSCB) triggers - a price move of at least 10% and 20 trading increments in five minutes (20% in the 9:30-9:50 AM post-open) - trading is halted for five minutes (extendable by another five). Key considerations: (1) Volume is interrupted during the halt, so reported volume understates true interest, (2) Order-book imbalance during the halt represents pending demand or supply, (3) Trades more than 5% beyond the trigger level are cancelled, which can affect prints, (4) Market-wide breakers (harmonized with the U.S.; the S&P/TSX Composite is used when U.S. markets are closed) can halt the whole market. For halted stocks, focus on the order-book imbalance and the move's catalyst rather than the interrupted traded volume.

Volume analysis enhances options trading in several ways: (1) Entry timing: Unusual equity volume often precedes options premium moves - enter options positions when equity volume confirms your thesis, (2) Strike selection: Options volume concentration at specific strikes may indicate institutional price targets, (3) Direction confirmation: Call volume surge with equity volume confirms bullish bias; put volume surge confirms bearish, (4) Event positioning: Pre-event equity volume intensity helps gauge expected move size, informing straddle/strangle pricing assessment, (5) Exit signals: If equity volume shows distribution patterns, consider closing options positions even if not yet at target. Always check options OI alongside equity volume; for inter-listed names, monitor the more-liquid US options too.

Key volume warning signs for trend reversal: (1) Volume divergence: Price makes new high but volume is lower than previous high - buying power waning, (2) Climax volume: Extreme volume (5x+) at price extreme often marks exhaustion - blow-off top or selling climax, (3) Distribution pattern: High volume on down days during uptrend - institutions selling into strength, (4) Declining volume trend: Each rally has less volume than the previous - participation fading, (5) Failed breakout on low volume: Price breaks key level but volume doesn't confirm - likely to fail. These are warning signs requiring heightened attention, not immediate reversal signals. Wait for price confirmation before acting against the trend.

Volume factor construction: (1) Define metric: e.g., 20-day volume z-score x ((1 - short%) / (1 - avg short%)) - captures both unusual volume and a long-side tilt; an inter-listed Canadian volume share can be an additional feature, (2) Universe selection: Apply to stocks meeting liquidity criteria (minimum volume, market cap), (3) Monthly ranking: Rank stocks by factor, form quintile portfolios, (4) Factor return calculation: Long top quintile, short bottom quintile, calculate return, (5) Evaluation: Measure mean return, Sharpe ratio, correlation with other factors (market, value, momentum), (6) Combination: If factor shows positive alpha and low factor correlation, include in multi-factor model with appropriate weight. Backtest with transaction costs, validate out-of-sample, and monitor for factor decay.

Effective ML approaches for volume analysis: (1) Isolation Forest: Good for unsupervised anomaly detection - identifies unusual volume patterns without labeled data, (2) Random Forest/XGBoost: For classification (unusual vs normal) or regression (predicting returns from volume features), handles non-linear relationships well, (3) Autoencoders: Learn 'normal' volume patterns and flag deviations as anomalies, (4) LSTM: For capturing sequential patterns in volume time series. Feature engineering is crucial: include volume ratio, z-score, short-sale percentage, inter-listed volume share, time-of-day factors, sector relative measures. Use time-series cross-validation to avoid lookahead bias. Start with simpler models (tree-based) before attempting deep learning.

Cross-market volume integration framework: (1) Identify relationships: For each stock/sector, map related markets (MX options/futures, the US-listed twin for inter-listed names, commodities, USDCAD), (2) Normalize across markets: Different markets have different volume scales - use z-scores or percentile ranks, (3) Calculate correlation baseline: Understand normal relationships between cash and derivative volume and the TSX/US volume share, (4) Detect anomalies: Flag when cross-market volume relationships deviate significantly from baseline, (5) Interpret divergences: Options volume leading equity may indicate informed positioning; futures without cash may be hedging; the US line leading the TSX often presages the open, (6) Build composite signals: Weight signals from each market by their historical predictive power. Implementation is helped by Canada and the U.S. sharing the Eastern time zone, but requires currency handling and careful timestamp alignment.

Key technical challenges: (1) Data latency and consolidation: Marketplace feeds have varying delays and trading is fragmented across venues, so consolidating volume in real time is non-trivial; broker APIs may throttle, (2) Scalability: Scanning hundreds of stocks with minute-level updates requires efficient algorithms and proper data structures, (3) RVOL complexity: Building accurate intraday profiles requires historical data and careful handling of session variations (open auction, MOC), (4) False positive management: Real-time has more noise; balance speed vs accuracy to avoid alert fatigue, (5) System reliability: Network failures, data gaps, and edge cases must be handled gracefully, (6) Integration: Connecting to multiple data sources, broker APIs, and alerting systems introduces complexity. Start with a focused watchlist (20-50 stocks) rather than full market, and expand as the system matures.

Validation methodology: (1) Information Coefficient (IC): Calculate correlation between volume signal and forward returns - should be consistently positive across time periods, (2) Quintile analysis: Form portfolios based on volume signal, calculate returns by quintile - top quintile should outperform bottom, (3) Event study: Measure average return following unusual volume events with statistical significance tests, (4) Out-of-sample testing: Reserve 20-30% of data for true out-of-sample validation, (5) Walk-forward analysis: Simulate real-time by training on past data, testing on next period, rolling forward, (6) Regime analysis: Test if signals work in different market conditions (bull, bear, high/low volatility), (7) Decay analysis: Check if predictive power is declining over time (signal arbitraged away). Signals should pass all tests; failure in any suggests overfitting or lack of real edge.