BG · Bunge Global SA
Stock Universe · 全球股票市场Quantitative Analysis: BG · BG 15-Minute Positional Strategy Achieves 84.85% Win Rate and 6.25 Profit Factor Across Multi-Year Backtest
This analytical study evaluates the top-ranked algorithmic trading strategy for BG stock on the 15-minute timeframe. Over a 6.98-year testing window from September 2019 to September 2026, the strategy generated a cumulative closed-trade gain of +2569.09% (135.35% annualized) across 33 completed trades. Defined by extreme signal selectivity, the strategy averages only 4.73 trades per year while holding positions for an average of 1170.89 hours (approximately 48.8 days). With an 84.85% historical win rate (28 wins, 5 losses), a profit factor of 6.25, and a maximum drawdown capped at 15.52%, the quantitative profile reflects exceptional payoff asymmetry and structural stability across multiple market regimes.
分析日期： 23.09.2026. 策略数据单独更新。阅读完整分析
Strategy profile
The strategy under examination represents the primary core strategy for BG within the DevioLab quantitative database, earning a DevioLab score of 84.92 and ranking first among evaluated models for this asset. Operating on a 15-minute execution interval, the system utilizes high-frequency price data to pinpoint entry and exit conditions, yet its actual trade execution frequency aligns with macro positional swing trading rather than high-frequency intraday trading. Across a 6.98-year evaluation window spanning from September 16, 2019, to September 8, 2026, the algorithm completed 33 closed trades. Out of these, 28 trades resulted in positive gains while 5 trades realized losses, establishing an overall win rate of 84.85%. Designed for automated implementation through trading bots, trading robots, or quantitative execution agents, the framework prioritizes trade selectivity over transaction volume, yielding a cumulative backtested return of +2569.09% and an annualized equivalent of 135.35%.
Trading rhythm and position duration
A defining characteristic of this algorithm is the stark contrast between its underlying 15-minute chart resolution and its extended multi-week holding times. Despite scanning intraday 15-minute bars, the average holding duration per position reaches 1170.89 hours (approximately 48.79 days), with a median holding period of 980.00 hours (roughly 40.83 days). The strategy executes an average of 4.73 trades per year, or approximately 1.03 trades per active month. Spacing between position exits is similarly spaced, averaging 73.22 days between trade completions with a median interval of 62.98 days. This low trade frequency indicates that the algorithm filters out short-term market noise, waiting for high-conviction macro setups before deploying capital. For automated trading agents, this operational pace minimizes exposure to intraday turbulence while concentrating capital in sustained multi-week trends.
Quality of historical results
The strategy demonstrates high payoff quality, highlighted by a profit factor of 6.25 and an average trade return of +11.52% compared to a median trade return of +7.84%. The peak historical performance was marked by a best trade return of +53.32%, while the worst trade yielded -15.52%. A critical indicator of systemic health is the dispersion of returns across trades: the top three winning trades collectively account for 29.48% of total gross profit. This figure confirms that performance is not dependent on a single speculative outlier; rather, profitability is broadly distributed across the 28 winning closed positions. The positive gap between the average trade return (+11.52%) and median trade return (+7.84%) highlights a right-skewed payoff distribution where large upside trends periodic expand the baseline profitability without sacrificing consistency.
Risk, drawdown and losing behavior
Risk containment within the model is evidenced by a maximum historical peak-to-trough drawdown of 15.52%. Notably, this drawdown figure matches the strategy's worst single trade loss of -15.52%, indicating that drawdown events have historically been bound by individual stop-outs or trade exits rather than compound multi-trade equity collapses. The algorithm recorded a longest losing streak of exactly 1 trade, meaning that throughout the entire 6.98-year history, the system never experienced consecutive losing trades. Conversely, the longest winning streak reached 12 consecutive profitable trades. Combined with an 84.85% win rate and a 6.25 profit factor, the loss distribution demonstrates that risk exposure is strictly controlled, allowing automated trading bots operating the logic to recover equity efficiently following negative trade events.
Behavior through time and yearly stability
An analysis of yearly performance figures across the 2020–2026 period reveals broad annual stability. The algorithm achieved positive cumulative net gains in every single calendar year evaluated. In 2020, 3 trades yielded a +59.38% return with a 66.67% win rate. In 2021, 4 trades produced +57.59% (75.00% win rate). Performance accelerated in 2022 with 9 trades generating +106.84% (88.89% win rate). In 2023, 5 out of 5 trades were profitable (100.00% win rate) summing to +46.60%. The year 2024 logged 2 wins out of 2 trades (+11.94%), while 2025 delivered 5 trades (4 wins, 1 loss) for +28.64% (80.00% win rate). In the partial year of 2026 through September, 5 trades produced +69.34% (80.00% win rate). Trade activity varied between 2 and 9 trades per year, reflecting dynamic adaptiveness where signal frequency expands during favorable trend regimes and contracts during choppy markets.
Recent period since 2024-06-01 versus full history
The recent sample window from June 1, 2024, to September 8, 2026, provides valuable insight into the strategy's ongoing efficacy. During this recent window, the strategy completed 11 trades, generating a cumulative return of +102.08%. This accounts for one-third of the total 33 historical trades over the full 6.98-year period, representing a moderate increase in trade density relative to earlier years. The strong return of +102.08% across 11 recent trades confirms that the model's underlying edge has remained active and effective in recent market conditions, showing no sign of structural decay or strategy degradation.
Strengths and limitations
The strategy's principal strengths include an exceptional 84.85% win rate, a strong profit factor of 6.25, a modest 15.52% maximum drawdown, and positive return consistency across every evaluated year. Additionally, gross profits are well distributed, with the top three trades contributing under 30% of total gains. The main limitation lies in the small absolute trade volume of 33 trades over nearly seven years. With an average holding period of ~48.8 days and over two months between trade exits on average, capital remains locked in individual positions for long durations, requiring patience from operators. Furthermore, low sample sizes per year necessitate ongoing statistical validation as additional live trades complete over time.
DevioLab analytical conclusion
With a DevioLab score of 84.92 and a rank of #1 for BG, this 15-minute positional model stands out as a robust quantitative strategy. By marrying granular intraday bar monitoring with multi-week position management, the system filters out short-term market noise while capturing substantial medium-term trends. Its combination of high win rate (84.85%), low drawdown (15.52%), excellent payoff asymmetry, and positive annual returns across seven distinct years provides a compelling foundation for deployment within automated trading environments. Quantitative traders seeking low-frequency, high-conviction systematic models will find this strategy's historical metrics highly representative of effective trend capture.
Data scope and methodology
This analysis is based on historical backtested strategy metrics for BG over the period from September 16, 2019, to September 8, 2026 (6.98 total years). All reported figures, including win rates, trade counts, holding durations, drawdowns, and return percentages, are derived from completed, closed positions executed on a 15-minute bar framework. Performance data reflects simulated backtested results and does not account for live slippage, market impact, or exchange commission schedules. Historical simulated outcomes serve strictly for quantitative evaluation and do not guarantee or predict future trading performance.