CFG · Citizens Financial Group Inc.
Stock Universe · global stock marketQuantitative Strategy Analysis: CFG · CFG Algorithmic Trading Model Evaluation
This analytical study evaluates the historical backtest performance of the primary algorithmic trading strategy for CFG on the 15-minute timeframe. Spanning over seven years of simulated market data from September 2019 through September 2026, the strategy generated 158 completed trades with a high win rate of 75.95 percent and an exceptional profit factor of 3.75. A defining feature of this model is its highly balanced return distribution: average trade returns (+3.05 percent) align almost perfectly with median returns (+3.03 percent), and the top three winning trades account for a minimal 8.17 percent of gross profit. This confirms that historical equity growth was driven by consistent, broad-based trade performance rather than isolated statistical outliers. Operating as a multi-day swing trading system with an average position duration of 186.54 hours, the strategy demonstrates strong temporal stability across varying market regimes, maintaining positive annual baseline gains even during its weakest period in 2023. With recent execution data since June 2024 producing 54 closed trades and +142.85 percent in cumulative trade returns, this model ranks first for the asset on DevioLab with an overall analytical score of 77.61, serving as a robust quantitative foundation for automated trading bots and algorithmic execution agents.
Analysis as of 01.10.2026. Strategy data is updated separately.Read full analysis
Strategy profile
The quantitative model evaluated in this report represents the top-ranked algorithmic trading strategy for CFG on DevioLab, achieving an overall score of 77.61 and securing the primary rank for the symbol. Operating on a 15-minute price bar interval, the strategy spans a comprehensive dataset of 7.03 years, beginning on September 16, 2019, and running through September 25, 2026. Across this extended historical horizon, the decision logic generated a total of 158 closed trades, yielding an overall cumulative performance metric of +6932.01 percent in closed trade percentage additions, which translates to a projected annualized trade-sum metric of 259.76 percent. Designed as a systematic trading framework, this decision architecture provides the underlying logic required for automated trading bots, trading robots, and quantitative agents seeking objective, rules-based execution in stock market assets. Rather than relying on hyper-frequent market noise, the algorithm filters 15-minute price action to isolate structured medium-term opportunities, striking a calculated balance between selectivity and statistical significance.
Trading rhythm and position duration
Despite utilizing a 15-minute execution chart, the strategy behaves as a classic multi-day swing trading framework rather than a rapid intraday system. Over the 7.03-year backtest window, the algorithm completed 158 trades, averaging 22.48 trades per year and approximately 1.93 trades per active month. Position holding times reflect a deliberate, patient capture of trend movements. The average holding duration stands at 186.54 hours (roughly 7.77 days), while the median holding duration is 148.38 hours (approximately 6.18 days). The slight positive skew between average and median holding times indicates that while most trades settle within six to seven days, a subset of positions remains open longer to fully exploit sustained price trends. This spatial spacing is further reflected in exit frequency metrics: the average time between trade exits is 16.28 days, with a median of 13.78 days. For quantitative traders implementing automated trading robots, this low-to-moderate transaction frequency provides a serene operational footprint, minimizing exposure to rapid execution churn while maintaining clear spacing between exposure cycles.
Quality of historical results
The statistical distribution of trade gains reveals an exceptionally high-quality return profile characterized by consistency and minimal outlier dependency. Out of 158 total trades, 120 resulted in profits while 38 ended in losses, establishing a baseline win rate of 75.95 percent. The profit factor of 3.75 demonstrates that total closed gross gains exceeded gross losses by nearly fourfold. What sets this strategy apart is the striking symmetry between its mean and median expectations: the average trade return is +3.05 percent, while the median trade return stands virtually identical at +3.03 percent. In quantitative equity research, close alignment between average and median figures indicates a symmetrical Gaussian-like distribution free from heavily distorted tails. This structural health is further confirmed by profit concentration metrics: the top three winning trades (+19.15 percent best trade) contributed just 8.17 percent of gross profit. Consequently, the strategy's equity curve relies on repeated, medium-sized winning trades rather than single extraordinary market events.
Risk, drawdown and losing behavior
A thorough evaluation of risk parameters reveals a disciplined loss-containment structure alongside manageable drawdown boundaries. The maximum historical drawdown recorded over the 7.03-year period was 26.81 percent. When contextualized against the overall total return profile and win rate, this drawdown figure represents a tolerable equity retracement. The strategy's worst single trade loss was -18.96 percent, which mirrors the magnitude of its best single winning trade (+19.15 percent), demonstrating symmetrical structural boundaries at the extreme tails. Losing sequence dynamics reflect strong resilience: the longest losing streak observed across the entire backtest was restricted to just 3 consecutive trades. In contrast, the strategy achieved a maximum winning streak of 18 consecutive trades. The short duration of losing streaks explains why drawdowns remain constrained; equity dips stem primarily from individual adverse moves or brief, localized clusters of losses rather than prolonged structural breakdowns. For automated trading agents, brief losing sequences significantly reduce the risk of panic interventions or operational compounding during negative performance cycles.
Behavior through time and yearly stability
Examining the strategy's annual breakdown demonstrates consistent positive drift across diverse market environments between 2019 and 2026. Every single calendar year generated a positive net cumulative trade sum. Performance opened in late 2019 with 5 trades, an 80.00 percent win rate, and +10.23 percent sum. The system expanded activity in 2020, logging 22 trades with a 77.27 percent win rate and a peak annual sum of +140.18 percent. Stability continued in 2021 (24 trades, 83.33 percent win rate, +84.72 percent) and 2022 (19 trades, 68.42 percent win rate, +53.00 percent). In 2023, market conditions presented greater resistance; trade frequency remained normal at 25 trades, but the win rate contracted to a history-low of 52.00 percent, producing a modest annual sum of +14.58 percent. Crucially, even during its most challenging regime, the algorithm avoided net capital erosion. Performance rebounded strongly in 2024 (20 trades, 80.00 percent win rate, +59.57 percent), peaked in win rate quality in 2025 (24 trades, 91.67 percent win rate, +88.14 percent), and maintained momentum into 2026 (19 trades, 78.95 percent win rate, +31.90 percent).
Recent period since 2024-06-01 versus full history
A dedicated analysis of closed trades completed on or after June 1, 2024, provides vital insight into modern strategy health and sample recency. Out of the 158 total historical trades, 54 positions closed within this recent window, representing over one-third of the complete lifetime dataset. These 54 recent trades accumulated a combined positive trade return of +142.85 percent. Comparing this recent window to the full 7.03-year history confirms that the strategy's edge has not deteriorated over time. The trade execution rate accelerated slightly during this period, generating approximately 2.3 trades per month compared to the long-term historical baseline of 1.93 trades per month. This active post-June 2024 performance confirms that the algorithm's multi-day swing logic continues to successfully capture trend mechanics under current stock market dynamics, making it highly relevant for live monitoring via automated trading bots.
Strengths and limitations
The primary analytical strength of this strategy lies in its outstanding win rate (75.95 percent), high profit factor (3.75), and balanced trade return profile. The structural lack of profit concentration—where the top three trades deliver less than 9 percent of gross profit—proves that historical success is deeply systemic rather than accidental. Furthermore, a maximum losing streak of only 3 trades provides excellent psychological and mathematical cushion during drawdowns. Conversely, notable limitations must be recognized. First, a maximum drawdown of 26.81 percent requires sufficient risk tolerance and equity capitalization. Second, with an average trade frequency of under two trades per month, the system demands strict discipline from operators, as extended flat periods between trades are common. Finally, as evidenced by the 2023 results where win rate fell to 52.00 percent, prolonged sideways or choppy price action can compress profit margins significantly.
DevioLab analytical conclusion
With a DevioLab score of 77.61 and a top rank of #1 for CFG, this algorithmic strategy stands out as a high-performing quantitative model. Its combination of multi-day position holding times, robust 75.95 percent win rate, and exceptionally uniform profit distribution offers a compelling historical profile. The model's historical ability to limit losing streaks to 3 trades while generating sustained positive annual performance from 2019 through 2026 underlines its structural robustness. Supported by strong recent validation since June 2024, the strategy represents an appealing logic set for deployment within automated trading agents and quantitative trading robots designed for medium-term stock swings.
Data scope and methodology
The metrics presented in this analysis are derived from backtested, historical simulated trade records for CFG across a 15-minute chart resolution from September 16, 2019, through September 25, 2026. All statistics reflect closed trade evaluations calculated in UTC standard time. Recent execution figures post-dating June 1, 2024, are compiled strictly from trades whose exit timestamps occurred within that timeframe. These figures represent simulated backtest outputs and do not account for live order execution slippage, brokerage commissions, overnight financing fees, or unexpected liquidity constraints. Past performance results in quantitative backtesting serve purely as analytical indicators and do not guarantee future live performance.