CGI Inc. RSI Strategy

No. RSI < 30 indicates the stock is oversold but does not guarantee a bounce. Wait for RSI to cross back above 30 (exit oversold) with a bullish candle confirmation, and apply filters: trend context, volume, sector. This is doubly important for CGI, which spent long stretches of 2024-2025 in a downtrend where oversold readings simply persisted. Blindly buying at RSI < 30 leads to catching falling knives.

RSI 14 (14-period) is the standard - smoother, fewer signals, more reliable. RSI 9 is faster, with more signals but also more noise and false signals. For swing trading CGI, RSI 14 is recommended; given CGI's already-low signal frequency, the smoother 14 keeps quality high. RSI 9 might suit intraday but demands quicker reaction.

Yes, especially in strong downtrends - CGI's 2024-2025 de-rating is a textbook case, with RSI staying low for weeks. This is exactly why waiting for RSI to EXIT oversold (cross back above 30) matters: it confirms that selling pressure is actually abating rather than just pausing.

CGI is an IT services and consulting firm with low beta (~0.6) - it behaves like a steady compounder. Shopify, Constellation Software, OpenText and Celestica are platform, software or hardware businesses with higher volatility. For RSI trading, CGI produces FEWER extremes (fewer signals) but they tend to be cleaner with less violent gapping. This reverses the logic of the Indian version, where the MORE volatile IT name (Tech Mahindra) was preferred - here CGI is deliberately the steadier choice. Match it to a patient, lower-frequency style.

The daily timeframe is recommended for beginners - less noisy, requiring only one check per day at the 4:00 PM ET close. The hourly can refine entry timing once you have a daily signal. Avoid lower timeframes (15-min, 5-min) until experienced; they generate too many false RSI signals, and CGI's modest intraday range makes those signals especially marginal.

Divergence shows potential, not an immediate signal. Wait for price confirmation: for bullish divergence (lower price low, higher RSI low) wait for price to break above a recent swing high. Combine with support/resistance - divergence at support is far more actionable than divergence in 'air'. On CGI in a downtrend, insist on that confirmation, since momentum can keep deteriorating.

Use mean reversion (extremes) when ADX < 20-22 (ranging). Use momentum (RSI > 55 or < 45 with the trend) when ADX > 25 (trending). Check trend context too - mean reversion works best when the RSI extreme runs counter to a primary trend that is otherwise intact (e.g. an oversold pullback inside an uptrend). When CGI itself is trending hard, lean momentum, not mean reversion.

Weekly RSI sets the primary bias (above/below 50), daily RSI provides signals, hourly RSI times entries. Best trades: weekly bullish plus a daily oversold bounce = high-probability long. Avoid buying a daily oversold while the weekly is bearish - lower probability because the primary trend is against you, which described CGI for much of its recent downtrend.

For an oversold bounce, buy ATM calls or a bull call spread (lower cost). For an overbought reversal, buy ATM puts or a bear put spread. Use the nearest monthly expiry for quick mean reversion (3-5 days) - CGI has no weeklies. Always check the bid-ask first: MX single-stock options are thinner than US, so use limit orders, and defined-leg spreads are often easier to fill and manage than naked premium near expiry.

Very important - volume confirms RSI signals. An oversold bounce with a volume spike means buyers are arriving (good); with no volume it is a weak signal. High-volume new lows despite RSI < 30 usually means more downside. Always check the volume pattern before acting, and remember CGI's average volume is lower than a mega-cap, so a genuine spike stands out clearly.

Calculate the ATR percentile over 100 days. High ATR (top 20%): use 25/75. Low ATR (bottom 20%): use 35/65. Because CGI is low-beta and often sits at the low-volatility end, the classic 30 may rarely print - so a percentile approach (oversold when RSI touches its own lower band or its bottom decile of recent readings) adapts automatically and is especially valuable for this name.

A failure swing is an internal RSI pattern. Bullish: RSI < 30, bounces to 40-50, retests but stays above 30, then breaks above the prior bounce high - a momentum shift before price confirms. Trade when RSI breaks the prior bounce high, with a stop below the retest low. It is higher-probability than a simple oversold reading, and for a lower-signal name like CGI it is worth waiting for this higher-quality pattern.

Train a classification model on RSI signal features (RSI value, rate of change, ADX, volume ratio, sector RSI, days to earnings) with the target being signal success/failure. Tree-based models learn non-linear patterns like 'RSI works when ADX < 22 but fails when ADX > 30,' and the probability output can guide position sizing. The honest caveat for CGI: its low signal count makes overfitting easy, so pool a long history, keep features parsimonious, and treat ML as a refinement on top of sound discretionary filters rather than a standalone system.

Use walk-forward analysis: train on 3 years, test on 1, roll forward and repeat. Test multiple parameters but avoid over-optimising. Measure Sharpe ratio and max drawdown, not just win rate, and require 100+ trades for significance - which for low-signal CGI may mean many years of data. Segment by regime (exclude oversold signals in strong downtrends), compare to buy-and-hold, and recalibrate every 1-2 years.

Calculate the strategy's historical volatility and target an overall volatility contribution. If the strategy runs at 16% volatility and you target 12%, scale the allocation by 12/16 = 0.75. Use fractional Kelly for sizing within the allocation. Crucially, limit correlation with other Canadian tech strategies - they move together, so treat them as a single strategy for diversification purposes, and pair them with uncorrelated sleeves such as banks or energy.