FAQs

Overview of model questions asked frequently

What is StrikeVision’s prediction model?
- At StrikeVision, we test multiple machine learning models and select the one that performs best based on accuracy rates. We’ve experimented with models such as Logistic Regression, Random Forest, Neural Networks, and XGBoost. Currently, our top performer is the XGBoost model, delivering a 70.97% accuracy rate on UFC fight predictions. We continuously review our models and update our process to ensure we’re always delivering the sharpest edge possible.
How does the model work?
- XGBoost works by building an ensemble of decision trees that “learn” from past UFC fights. The model looks at many variables—striking stats, grappling data, fighter age, and more—to estimate the probability of each fighter winning. We constantly update the model with new data to keep it sharp.
What factors does the model consider?
- The model analyzes the statistical differences between both fighters across a range of key UFC metrics—such as significant strikes landed per minute (SLpM), strikes absorbed per minute (SApM), takedown averages, and more. To give a fuller picture, we also factor in UFC experience (number of UFC fights), time since the last fight, age and reach differences, betting odds, and several additional variables. By comparing these factors side by side, the model aims to capture the most important edges in each matchup.
How accurate is the model?
- Our model is currently achieving a 70.97% accuracy rate, based on predictions from over 6,000 UFC fights. While no prediction system is perfect—especially in a sport as unpredictable as MMA—our data-driven approach consistently outperforms gut instincts and hot takes.
Is the model totally automated?
- Predictions are machine-generated, but we also review each pick for edge cases. Prop bets and specialty picks are further refined by tape study.
Can I trust these predictions for betting?
- Our model is built to find value and beat the market over the long run, but it’s important to remember that all betting involves risk—especially in a sport as unpredictable as MMA. We recommend using our picks as a smart part of your overall betting strategy, and always encourage responsible wagering.
What are XGBoost’s limitations?
- Like any model, XGBoost can’t predict random events (injuries, bad judging, etc.), and sometimes lacks context on motivation, late camp changes, or off-the-cage issues. We do our best to adjust, but there are always unpredictable factors in MMA.
How do you handle new fighters or limited data?
- For fighters with little UFC experience or limited data, the model may exclude the fight or lower its confidence in the prediction. We’ll always let you know when this happens.
Do you bet your own money on these picks?
- Yes, we put our own money behind many of our model’s picks, and we provide full transparency on wins, losses, and long-term ROI.
10. Can I see past results or your track record?
- Absolutely. We transparently track and share every bet, including results and ROI. Full history is available on our Patreon/website/socials. Our tracked picks are found here : https://www.betmma.tips/strikevision
What separates StrikeVision from other prediction services?
- StrikeVision blends advanced machine learning with full transparency. We openly track both wins and losses and are consistently working to update and perfect our model. Importantly, we place our own bets using these predictions and bet alongside our community, ensuring our interests are directly aligned with yours.
Do you guarantee wins or offer refunds?
- We don’t guarantee wins, and we don’t offer refunds. MMA is unpredictable by nature—no model or expert can see every upset coming. What we do guarantee is complete honesty: you’ll always see how each pick was made, what factors the model considered, and where our confidence stands. Our models have a strong track record, but results can and do vary from event to event. We focus on providing reliable insight, but managing your bets is always up to you.