Simple beats complex. We have engineered dividend capture down to a single control: the probability threshold. Based on real 2025 data across 85,878 dividend candidates, pick your risk tolerance and everything else optimizes itself. Higher probability means safer trades but fewer opportunities; lower probability means more trades but occasional losses.
Read the numbers below as a global reference ceiling, not a personal target. Every figure in this article comes from picking the single highest-yield qualifying dividend across all world markets each day, tens of thousands of candidates. No broker gives you access to all of them, so you cannot personally capture these returns, and they are shown gross, before tax. A realistic, tradable benchmark is your own accessible market: US, our most conservative, backtested at about 220% gross a year, roughly 128 to 167% after tax. Use this article to understand how the threshold behaves, not to set your return expectations.
How Perfect Hunting Works
Every trading day during 2025 (295 trading days total), the algorithm identifies dividend capture opportunities using a straightforward two-step process:
- Step 1: Probability Filter: All global equity instruments with ex-dividend dates that day are evaluated. We keep only those where our AI model predicts same-day price recovery with your chosen probability threshold (0.5 = 50%+, 0.6 = 60%+, etc.).
- Step 2: Pick the Best: From the filtered candidates, we automatically select the one with the highest dividend yield.
That's it. No second lever. No yield adjustments. Set your probability threshold, and the system finds the highest-yield opportunity that meets it.
The ceiling: If we had perfect hindsight (knowing exactly which trades would succeed) we could achieve 909.9% annual return in 2025. This is our reference point. Everything below is the realistic "prediction path" where we're making actual bets based on model confidence.
The Results: By Probability Threshold
Adjust probability threshold to balance accuracy against opportunity frequency. Higher thresholds filter more aggressively (fewer trades, higher win rate). Lower thresholds trade more often (more opportunities, but more losses).
Global reference, gross of tax. The rows below assume you can trade every qualifying dividend on earth and always take the single best one, which no broker allows. Treat them as a ceiling, not a target. Your real, tradable return depends on the markets your broker covers; on US, our most conservative market, the backtest is about 220% gross, roughly 128 to 167% after tax.
| Probability | Annual Return | Win Rate | Trades/Year | Failures | Avg Dividend |
|---|---|---|---|---|---|
| 0.50 | 394.02% | 88.66% | 238 | 27 | 1.66% |
| 0.55 | 364.85% | 91.95% | 236 | 19 | 1.55% |
| 0.60 | 320.68% | 92.41% | 237 | 18 | 1.35% |
| 0.65 | 254.04% | 92.95% | 227 | 16 | 1.12% |
| 0.70 | 147.12% | 95.14% | 185 | 9 | 0.80% |
| 0.75 | 28.08% | 95.56% | 45 | 2 | 0.62% |
Key insight: Win rate improves steadily (88.66% → 95.56%), but returns collapse at high thresholds. Moving from 0.5 to 0.75 increases accuracy by only 6.9% while sacrificing 365% in annual returns. The low probability thresholds deliver dramatically better compounding.
Recommendation: Probability 0.50
Best for Most Traders
394.02% annual return with 88.66% win rate (global reference, gross of tax; your tradable market will be lower, US backtests at about 220% gross / 128-167% net)
Out of 238 annual trades, you win 211 and lose 27. That's one failure every ~9 trading days. Even with these occasional losses, the mathematics of daily frequency + dividend collection + frequent partial recoveries creates exceptional compounding.
A sub-90% win rate sounds risky. It's not. 88% accuracy is elite in trading. The difference between a loss and a win here is small: losses average only a few percentage points, while wins often hit 2-3%. The frequency and mathematics work in your favor.
Understanding Win vs Loss
WIN: Price Recovery + Dividend Profit
The stock recovered enough on the ex-date that: dividend collected + price movement (full or partial) = net positive. This includes both perfect recoveries (stock bounces back fully) and partial recoveries (stock doesn't fully recover, but dividend plus partial recovery = profit).
LOSS: Price Drops Below Dividend
The stock dropped more than the dividend amount. You collected 1.5% dividend, but the stock fell 3%, resulting in a -1.5% net loss. The returns shown above include all these losses: they're not excluded or smoothed.
The brutal math in your favor: even at 88% accuracy, with 238 opportunities per year, the small average loss is drowned out by the high frequency of small wins.
Data Distribution
Return vs Win Rate by Probability Threshold
Larger dots represent more trading opportunities. The pattern is clear: lower probability thresholds trade more frequently (larger dots) and deliver higher returns, despite slightly lower accuracy.
Top Probability Thresholds Ranked
Choosing Your Threshold
Conservative (0.65-0.75): Sleep soundly. Win rate exceeds 92%. Trade infrequently. Perfect if you want very few losses even though returns drop significantly.
Balanced (0.55-0.60): Good middle ground. Win rate reaches 91-92%, returns remain strong (320-365%). Suitable if you're new and want a safety net while building confidence.
Aggressive (0.50): Maximum daily opportunities. 88.66% win rate + 394% return. You accept more frequent losses but capture more of the available edge. Recommended for experienced traders who understand that occasional losses are part of the system.
Bottom line: Start at 0.55 if you're new. Move to 0.50 once you're comfortable with the occasional loss. Skip 0.70+ unless you specifically want minimal trading activity.
Summary
- One control: Probability threshold. That's it. No yield adjustments, no secondary levers.
- Lower probability = more opportunities: 0.50 trades 238 times/year at 88.66% accuracy. 0.75 trades only 45 times/year at 95.56% accuracy. Same signal, different frequency.
- 0.50 is optimal for returns: 394% annual return is nearly 14x better than 0.75. Even with 27 losses per year, the mathematics favor maximum frequency.
- Losses aren't disasters: They're typically 1-2%. Wins are typically 2-3%. At 238 trades/year, the math compounds strongly in your favor.
- This is not luck: the strategy validates back to 2000 across global markets, and the win rate has held up over more than two decades. The ~400% figure is a global ceiling, gross of tax, that assumes access to every world market. Your real, tradable return is lower: US, our most conservative market, backtests at about 220% gross, roughly 128 to 167% after tax. The signal is real and repeatable; the headline number is a reference.
TLDR: Pick your probability threshold. Walk away. The system finds the highest-yield opportunity that meets your criteria. 0.50 for aggressive, 0.55-0.60 for balanced, 0.65+ for conservative. Live results will be 75-90% of backtested due to execution timing and spreads, but the edge is real and repeatable.
Frequently asked questions
What probability threshold should I use?
Lower thresholds around 0.50 trade more often for higher returns at a slightly lower win rate; higher thresholds trade rarely at a higher win rate but much lower returns. Most traders sit near the lower end for the compounding.
Are the 394% figures realistic for me?
No. They assume you can trade every qualifying dividend on earth and always take the single best one, which no broker allows. Treat them as a global reference ceiling; your tradable benchmark is your own market, about 220% gross on US and 128 to 167% after tax.
