What Is Edge?

What is edge?

Edge is the difference between the market's fair probability and your own estimated probability.

The basic formula is:

For example:

• Market fair probability: 48%
• Your estimated probability: 53%

The edge is:

This is a difference of five percentage points, not a five-percent increase.

Do not use raw market probability without adjustment

The probability derived directly from bookmaker odds includes the bookmaker margin.

For odds of 2.00:

But in markets with multiple selections, such as 1X2 or Over/Under, the implied probabilities usually add up to more than 100%.

Before calculating edge, we therefore need to remove the bookmaker margin and calculate fair probability.

Edge should generally be measured against this fair probability.

A basic example of edge

Consider the following bet:

• Offered odds: 2.10
• Market fair probability: 47%
• Your estimated probability: 52%

The edge is:

Expected value is:

Therefore:

Edge and EV are related, but they are not the same number.

Edge measures the difference in probability.

EV measures the profitability created by combining that probability difference with the offered odds.

The difference between edge and EV

Edge answers:

EV answers:

The same edge can produce different EV depending on the offered odds.

Suppose your estimated probability is 55% and the market fair probability is 50%.

The edge is five percentage points.

At odds of 2.00:

EV is +10%.

At odds of 1.80:

EV is −1%.

An edge does not guarantee positive EV if the offered price is too low.

Percentage-point edge and relative edge

There are two common ways to express edge.

1. Percentage-point difference

For example:

• Market: 40%
• Your estimate: 45%

The edge is:

2. Relative edge

This measures how much higher your estimate is relative to the market probability.

Using the same example:

The relative edge is:

A five-point gap has a different meaning when moving from 10% to 15% than when moving from 80% to 85%.

Relative differences can look large in low-probability markets

Suppose:

• Market fair probability: 10%
• Your estimated probability: 13%

The edge is three percentage points.

The relative edge is:

Now consider:

• Market fair probability: 70%
• Your estimated probability: 73%

The edge is still three percentage points.

The relative edge is:

Small percentage-point differences can appear large in relative terms when the base probability is low.

However, low-probability events are also harder to estimate accurately, so a large relative edge should not automatically be treated as strong evidence.

Edge does not guarantee profit

A positive edge does not mean the bet will win.

If your estimated probability is 55%, the bet still loses 45% of the time.

Edge is not a prediction of a single result.

It is an indication that your probability estimate may outperform the market over repeated decisions.

That is why edge must be evaluated over a large sample.

The hardest part is creating your own probability

The arithmetic behind edge is simple.

The difficult question is:

Probability estimates may use information such as:

• xG and xGA
• shot volume
• shots on target
• home and away splits
• strength of opposition
• player absences
• formation
• schedule
• importance of the match
• market-specific characteristics
• odds movement
• sample size
• data quality

If these inputs are handled poorly, the calculated edge becomes an illusion.

Be sceptical when the edge is unusually large

In a mature market, an extremely large edge should trigger a review of your own analysis.

For example:

• Market fair probability: 45%
• Your estimated probability: 65%

That is a 20-point difference.

There may be a genuine information advantage.

More often, however, the cause may be:

• incorrect input data
• missed injury or lineup information
• biased samples
• confusion between home and away data
• different market definitions
• confusion between 90-minute and extra-time markets
• timing differences in odds collection
• model overfitting

The larger the edge, the more carefully it should be verified before execution.

Does a small edge have value?

A small edge can still be valuable if the estimate is accurate, costs are low, and the opportunity can be repeated.

For example:

• Market fair probability: 50%
• Your estimated probability: 52%

The edge is two percentage points.

At odds of 2.05:

However, if the estimation error is ±4 points, a two-point edge may simply be noise.

The size of the edge is not enough on its own.

We also need to consider confidence intervals and stability.

Edge and estimation error

An estimate of 52% does not mean the true probability is exactly 52%.

It may actually be 48%.

Every probability estimate contains error.

In practice, we need to think about:

For example:

• Apparent edge: 5 points
• Estimated error: ±4 points

The real advantage may be very small.

By contrast:

• Apparent edge: 8 points
• Estimated error: ±2 points

This is a more credible candidate.

Edge reliability varies by market

Edge should not be treated equally across every market.

Highly liquid markets such as:

• 1X2
• Asian handicap
• main totals markets

tend to be more efficient.

Less liquid markets such as:

• corners
• cards
• offsides
• player props
• smaller leagues
• niche markets

may retain more pricing inefficiency.

However, lower-liquidity markets also tend to have weaker data, lower limits, and greater uncertainty.

A market being easier to beat does not necessarily mean it is easier to profit from consistently.

Odds movement and edge

If the odds shorten after you place a bet, the market has moved in your direction.

For example, if you bet at 2.10 and the market closes at 1.95, you may have captured a better price than the closing market.

This is commonly measured through Closing Line Value, or CLV.

CLV is an important supporting indicator when evaluating whether your edge estimates are directionally correct.

However, odds movement alone does not prove that the original decision was correct.

Markets can also overreact or move temporarily in the wrong direction.

Edge and the Kelly criterion

Edge also affects bet sizing.

The Kelly criterion uses your estimated probability and the offered odds to calculate a theoretical optimal fraction of bankroll to wager.

The basic formula is:

A larger edge generally produces a larger Kelly fraction.

Because probability estimates are uncertain, practical systems often use Fractional Kelly, such as:

• 0.25 Kelly
• 0.35 Kelly
• 0.50 Kelly

rather than full Kelly.

When a positive edge should still be ignored

Even when the calculated edge is positive, a bet may still be rejected if:

• the supporting data is limited
• lineup or injury information is unresolved
• the market definition is unclear
• the odds have already fallen significantly
• the model performs poorly in that market
• the thesis depends on several assumptions
• the bet is strongly correlated with other positions
• betting limits are low
• the sample is biased
• estimation error is larger than the edge

Edge is a condition for considering a bet, not an automatic instruction to place one.

How to validate edge

To test whether edge estimates are reliable, we need to record the process rather than only the result.

At minimum, record:

• odds at the time of the bet
• market fair probability
• your estimated probability
• edge
• EV
• confidence
• stake
• closing odds
• result
• reasoning
• review notes

Once enough data has been collected, evaluate:

• probability calibration
• performance by edge band
• performance by market
• CLV
• ROI
• maximum drawdown

What is calibration?

Calibration measures whether predicted probabilities match actual outcomes.

If bets estimated at 55% win approximately 55% of the time over a large sample, the model is well calibrated.

If bets estimated at 55% win only 45% of the time, the model is overestimating probability.

Edge calculations assume that your probability estimates are reasonably accurate.

Calibration is therefore fundamental to edge analysis.

How AI uses edge

AI can support data processing, probability estimation, market comparison, and simulation.

It can help automate:

• combining multiple data sources
• normalising market odds
• producing model probabilities
• calculating edge
• calculating EV
• adjusting confidence by market
• validating the model against historical results

But AI-generated edge still depends on the input data and model design.

A large edge produced by AI should not be accepted blindly.

We still need to ask:

• Why does the model disagree with the market?
• Is there an explainable reason for the difference?
• Is the information current?
• Has the model been validated in this market?

Edge in EV Bet Engine

EV Bet Engine does not evaluate edge in isolation.

It combines:

• market fair probability
• model probability
• edge
• EV
• data quality
• model confidence
• odds movement
• market characteristics
• correlation
• bet size

The key question is not only:

It is also:

Summary

Edge is the difference between the market's fair probability and your own estimated probability.

The basic formula is:

Edge indicates that your probability estimate may be more favourable than the market's.

However:

• edge does not guarantee profit
• market probability should be adjusted for margin
• estimation error must be considered
• unusually large edges require additional verification
• reliability varies by market
• edge should be validated alongside EV, CLV, and calibration

The goal is not to create predictions that are merely different from the market.

Edge is the metric used to measure that difference.