From xG to xEV: The Evolution of Sports Metrics

The Analytics Revolution Nobody's Talking About

Expected Goals—xG—changed football forever. In 2017, when analytics teams could finally quantify what humans only intuited, the sport gained a new language. A team could dominate possession but lose decisively. A team could sneak through on efficiency alone. xG made invisible patterns visible.

But here's what most people in sports media miss: xG is just the beginning.

The real commercial opportunity isn't in knowing which team had better chances. It's in translating that knowledge into revenue streams, user engagement, and competitive advantage for the operators, platforms, and rights holders who control the infrastructure.

This is the gap between xG and xEV—Expected Value.

For rights holders, broadcasters, and operators, the question isn't "Is xG accurate?" It's "How do we monetise it?" And the answer requires understanding how analytics flows through three critical commercial layers: prediction, personalisation, and pricing.

This article explains how modern B2B sports tech infrastructure bridges that gap—and why the operators winning market share in 2026 aren't the ones using analytics best. They're the ones that monetise it fastest.

Why Analytics Alone Doesn't Generate Revenue

Before we talk about commercial value, let's be honest about what analytics isn't.

Analytics is not content. A chart showing xG distribution doesn't make anyone place a bet. A dashboard of shot quality metrics doesn't convince a viewer to subscribe. Analytics is inert—valuable in aggregate, invisible in execution.

The sports media industry has spent the last decade creating analyst roles, hiring data scientists, and building dashboards. Millions of dollars have poured into infrastructure to calculate better predictions. But relatively little has gone into the harder problem: scaling predictions into business outcomes.

This is where the disconnect between sports media companies and B2B technology operators becomes obvious.

A traditional sports broadcaster might use xG internally—to inform commentary, to build graphics, to analyse performance. But they're not extracting commercial value from that analysis. They're using it as an input to content production, which is a margin-destroying model.

An operator building on modern B2B infrastructure, by contrast, treats analytics as the product itself. They're not making xG predictions for internal use. They're making 1.1 billion predictions per year and monetising every single one.

That's the shift from xG to xEV. Expected Goals is descriptive. Expected Value is commercial.

The Three Layers of Monetisation

Layer 1: Prediction at Scale

The foundation of commercial analytics is pure throughput. Operators need predictions—not once per match, but continuously, across thousands of events, updating in real-time as conditions change.

FairPlay's FairPlay AI engine processes 1.1 billion AI predictions annually. That's not the most predictions in the industry; it's the sustained rate of production that matters. Here's why:

125 million daily price changes need to be informed by updated predictions. A betting exchange doesn't set odds once at kickoff. It adjusts them continuously based on:

Current game state (possession, passes, expected outcome)
Injury information and team news
Betting volume and market sentiment
Historical patterns from similar situations

Each of those signals requires a prediction. Each prediction requires an AI model trained on historical data, optimised for latency, and updated in milliseconds. And each update needs to be commercially justified—it must either prevent losses or generate spreads that are profitable.

This is the first layer of xEV: predictions are only valuable if they can be generated at the scale your business operates.

For rights holders considering partnerships, this is the critical question: Does your partner have the infrastructure to make predictions fast enough to matter? A prediction that takes 10 seconds to generate is already obsolete in modern betting markets. The operators winning in 2026 are making predictions in tens of milliseconds.

Layer 2: Personalisation and Engagement

Once you have predictions flowing through your system, the next question is simple: How do you use them to keep users engaged?

The data here is stunning. Not 18% more. 18 times more engagement.

What changed? Not the underlying data—a global broadcaster partner had sports data before. What changed was the application of that data.

Personalisation isn't about showing everyone the same prediction. It's about understanding what each user wants to see, then delivering that insight in the format they're most likely to act on.

Consider two users watching the same match:

User A is a casual viewer. They want simple, visual insights. Show them a player form chart or a simple "Team A has a 65% win probability" overlay.
User B is an experienced bettor. They want edge. Show them why that probability exists, where the sharp bettors are placing money, what changed since the previous match.

Both are watching the same game. Both benefit from the same underlying predictions. But the personalisation layer—the AI system that decides what to show whom—is what drives engagement.

This is where the commercial value compounds. Higher engagement means:

More time on platform (retention)
More transactions (betting volume)
Better targeting data (for advertisers)
Increased lifetime value (for operators)

The operators building this correctly are seeing significant engagement uplift because they're not just predicting outcomes. They're predicting what users care about and delivering it in real-time.

Layer 3: Pricing and Margin

The final layer is the one most stakeholders understand poorly: pricing.

When xG became standard in football analysis, bookmakers and exchange operators faced a problem. If everyone has access to the same xG model, how does anyone get an edge?

The answer: you don't. You need a better prediction model, applied faster, with more context, and updated more frequently.

This is where infrastructure becomes genuinely defensible. It's not defensible because it's complicated—it's defensible because it requires scale, speed, and continuous improvement.

Here's the mechanical insight: On a betting exchange, if your prediction of the outcome differs from the market's consensus price, you have a profit opportunity. If the market thinks a team has a 55% win probability (reflected in -120 odds), but your model thinks they have a 60% probability, you can lay the opposite side and expect profit over a large sample size.

But this only works if:

Your model is genuinely better (requires hundreds of millions of training examples)
You can execute your edge before the market corrects (requires millisecond speed)
You can repeat this across thousands of events (requires systematic, automated trading)

Rights holders licensing their data to operators understand this intellectually. But they often don't recognize that their competitive advantage, when monetised properly, can generate $5 million+ in annual revenue (as premium US sports publishers has achieved through BetTech partnerships).

Why? Because their data—combined with modern AI infrastructure—becomes an edge. And that edge can be automated, scaled, and repeated continuously.

The Operator's Perspective: Why This Matters

From an operator's viewpoint, the shift from xG to xEV is structural.

Traditional model: Build prediction models → Publish results to users → Hope they engage → Hope they bet

Modern model: Build prediction infrastructure → Feed predictions into real-time pricing → Personalise delivery → Monetise through volume and spread

The second model generates revenue in at least five ways:

Direct spread capture (exchange operators)
Margin on implied odds (bookmakers)
Engagement advertising (publishers)
Data licensing (to third-party platforms)
Subscription premium content (analysts and professionals)

And that's just the direct channels. Indirect value—improved retention, reduced player churn, better player lifetime value—often exceeds direct revenue.

The data backs this up. Operators serving 45+ regulated markets simultaneously are processing that volume specifically because they've figured out how to extract xEV from xG. They're not serving multiple markets for fun. They're doing it because the infrastructure that handles one market can handle fifty with minimal additional cost.

How to Assess Your Analytics Infrastructure

If you're a rights holder or operator evaluating whether your current analytics setup is generating genuine xEV, ask these questions:

1. Speed: How fast are your predictions generated? If the answer is "within a minute," you're too slow. Modern infrastructure operates in milliseconds. This matters because the faster you predict, the more pricing opportunities you capture before the market corrects.

2. Scale: How many predictions can you generate simultaneously? If you're calculating predictions for one match at a time, you're not scaling. The winning infrastructure makes thousands of simultaneous predictions across global markets.

3. Update frequency: Do your models train and retrain continuously? Analytics in sports is like weather forecasting: yesterday's data is less valuable than today's data. The best infrastructure retrains models daily, or in some cases hourly.

4. Integration: Are your predictions automatically flowing into pricing and personalisation systems? Or are you generating predictions and then manually deciding how to use them? Integration is where value gets trapped. If someone has to interpret your predictions before acting, you're losing speed and scale.

5. Monetisation: Are you capturing revenue from your predictions, or just publishing them? This is the crucial question. Many organizations have phenomenal analytics. Almost none monetise them systematically.

The $60B Opportunity

The US regulated sports betting market alone represents a $60 billion addressable opportunity. But the value isn't distributed evenly. It's concentrated among operators who have solved the xEV problem.

The operators who can:

Generate predictions faster than competitors
Personalise those predictions to drive engagement
Systematically capture profit margins from superior predictions
Scale these capabilities across markets

...are the ones capturing the majority of that $60 billion TAM.

For rights holders, this means your data is only as valuable as the infrastructure you partner with. A partnership with an operator that generates 1.1 billion predictions annually and has proven significant engagement uplift is fundamentally different from a partnership with someone who just wants to publish a weekly analytics report.

For publishers and platforms, the shift to xEV means analytics is no longer a content layer. It's infrastructure. And infrastructure generates revenue at a completely different scale than content.

From Theory to Practice: The Infrastructure Stack

Making the jump from xG calculation to xEV monetisation requires five core components:

1. Data ingestion: Real-time collection of game data, betting volume, news events, and injury reports. This needs to be automated and validated. At scale, data quality issues can destroy edge.

2. Model training: Continuous retraining of prediction models on historical data combined with recent outcomes. The model that works today won't work in six months unless you're constantly learning.

3. Prediction generation: The actual computation of predictions at scale. This usually means distributed computing, edge caching, and optimisation for latency rather than perfect accuracy.

4. Integration layer: Automatic feeding of predictions into pricing systems, personalisation engines, and user-facing applications. This is where many organizations fail—they have great predictions but don't have great systems to use them.

5. Monetisation tracking: Systems to measure whether predictions are actually generating revenue, which predictions generate the most revenue, and where to invest in model improvement.

Organizations that have all five of these components working together are the ones extracting xEV. Organizations that have some but not others are sitting on valuable intellectual property that isn't generating proportional returns.

Common Mistakes in Analytics Monetisation

Mistake 1: Confusing Accuracy With Value

The highest-accuracy prediction model doesn't automatically generate the most revenue. Sometimes a simpler, faster prediction that captures 80% of accuracy but trades 20 percentage points for 50x speed advantage is worth more.

Why? Because in markets where timing matters, speed often beats accuracy. A correct prediction two seconds too late is useless.

Mistake 2: Treating Analytics as Content

Sports media companies often make this error: they build great analytics, then publish them as content (articles, graphics, commentary). This is margin-destructive. Analytics should feed into products and pricing, not sit on top of them.

Mistake 3: Not Investing in Data Quality

Garbage in, garbage out. Operators who win are obsessive about data quality because a single bad data point can corrupt an entire model training cycle. This requires investment in validation pipelines, testing, and monitoring.

Mistake 4: Underestimating the Importance of Real-Time

Predictions that are accurate but stale are worse than useless—they're dangerous. They can lead you to make decisions (in pricing, in trading, in market making) that are directionally correct but tactically wrong.

Mistake 5: Not Measuring Commercial Impact

Many organizations generate predictions but don't measure whether those predictions actually drive revenue. This is the biggest miss. If you're not measuring xEV separately from xG, you have no idea whether your analytics investment is actually working.

The Competitive Landscape

In 2026, the operators winning market share aren't the ones with the best data. They're the ones with the best infrastructure for monetising data.

This is why traditional sports media companies, despite having incredible data assets (broadcast footage, historical records, fan relationship data), struggle against B2B tech operators. They have better data, but operators have better systems.

The operators who have figured out how to:

Process 125 million price changes daily
Generate 1.1 billion predictions annually
Serve markets across 45+ regulated markets
Deliver significant engagement uplift through personalisation
Generate $5M+ annual revenue per partner

...are building defensible moats. And that moat is built on infrastructure, not insight.

For rights holders considering partnerships, this should be the primary question: "Does this operator have the infrastructure to monetise my data?" Not "Do they understand sports?" Not "Are they credible?" But specifically: "Can they extract commercial value from analytics at scale?"

Building Your xEV Strategy

If you're evaluating your analytics monetisation, here's a framework:

Step 1: Assess current state. Where are your predictions generated? How fast? How often do you update them? Are they feeding into products, or are they sitting in dashboards?

Step 2: Identify the monetisation gap. If you're generating predictions but not extracting revenue, what's missing? Is it speed? Scale? Integration? Personalisation?

Step 3: Prioritize infrastructure over insights. Yes, better models matter. But better infrastructure to deploy existing models matters more. Speed, scale, and automation compound faster than raw model accuracy.

Step 4: Measure everything. Set up systems to track which predictions drive revenue, which personalisation strategies engage which user segments, and how your analytics infrastructure compares to competitors.

Step 5: Partner strategically. If you're a rights holder, partner with operators who have proven infrastructure. If you're an operator, invest in infrastructure before investing in model complexity.

Why This Matters for Investors

From an investor perspective, the shift from xG to xEV represents a fundamental change in how sports betting and gaming infrastructure creates value.

Companies that can demonstrate:

Predictive output at scale (1 billion+ predictions annually)
Conversion of predictions into revenue (measurable uplift in operator margins)
Deployment across multiple markets and use cases
Continuous improvement in model performance

...are building valuable, defensible businesses.

The key metric isn't "How accurate are our predictions?" It's "What revenue did our predictions generate, compared to the cost of generating them?"

Investors in sports betting infrastructure should be asking operators to show that ratio. If they can't measure it, they're not really monetising their analytics.

The Future: From xEV to Automated Value Creation

Looking forward, the next phase of monetisation moves beyond personalisation to automation. This is where agentic AI enters the picture—systems that don't just predict outcomes and recommend actions, but that execute actions autonomously within defined parameters.

An agentic AI might:

Automatically adjust odds across markets based on prediction updates
Optimise user segmentation and offer delivery in real-time
Manage hedging and exposure automatically
Route users to the products they're most likely to convert on

This represents the final frontier of xEV monetisation: not just predicting value, but automating the capture of it.

But that's a conversation for another article.

For now, the key insight is simple: Analytics alone doesn't generate revenue. Only analytics infrastructure, deployed at scale, feeding into integrated products, generates commercial value.

The operators winning in 2026 have figured this out. Rights holders and platforms that want to compete need to figure it out too.

Conclusion: The Infrastructure Imperative

The era of analytics in sports is over. The era of analytics infrastructure has begun.

What this means practically: the organizations extracting the most value from sports data in 2026 are not the ones with the best analysts or the smartest models. They're the ones with the best systems for deploying those models at scale, in real-time, integrated with business systems, continuously improving.

For rights holders, this is an opportunity to partner with operators who can genuinely monetise your data. For operators, this is a mandate to build infrastructure before building features. For investors, this is a signal to look for companies that can measure and defend their analytics edge through business outcomes, not just model quality.

The shift from xG to xEV isn't semantic. It's the difference between having valuable insights and having a valuable business.

The operators winning in 2026 know the difference. Do you?

Ready to explore how modern infrastructure can turn your analytics into revenue? FairPlay's infrastructure powers 1.1 billion AI predictions annually across 45+ regulated markets, generating proven revenue uplift for rights holders and operators. Contact FairPlay to discuss your analytics monetisation strategy.