Most Elo rating systems give a team a single number. BetSignals gives every team three: an overall (composite) rating, a separate attacking rating, and a separate defending rating. That split is the unusual part, and it is what lets the model reason about a match as a set of matchups rather than a single strength gap.
What Elo is
Elo is a rating system originally built for chess. Each team carries a number; when they play, the result nudges both ratings — beat a stronger side and you gain more than beating a weaker one. Over a season the numbers settle into a stable picture of relative strength. In BetSignals most teams sit somewhere in the 1300–1600 range, with the strongest sides pulling clear at the top.
Unlike a league table, Elo is continuous, updates after every game, and accounts for the quality of the opponent — not just whether you won.
Why we split attacking from defending
A single strength number hides how a team is good. Two sides on the same composite rating can be completely different: one wins 3–2, the other wins 1–0. Collapse them into one figure and you lose that.
So BetSignals tracks two extra numbers per team:
- Attacking Elo — how well a team creates and scores.
- Defending Elo — how well a team suppresses and prevents.
The value of this shows up when two teams meet. What matters for the home side's goals is not the opponent's overall rating but the opponent's defending rating — your attack against their defence. We call that gap the attack edge. A strong attack facing a weak defence produces a very different expectation than the composite gap alone would suggest, and the split lets the model see it.
How the ratings move
BetSignals uses a shots-adjusted update. Rather than reacting to the scoreline alone — which is noisy, because one deflection can flip a result — the update also weighs the underlying shot balance. A team that loses 1–0 but heavily out-shoots its opponent is not punished as harshly as the scoreline suggests. This makes the ratings steadier and quicker to reflect real performance.
Bigger, rarer matches move the numbers more: European fixtures use a larger update factor than domestic-league games, because they carry more information about where a side really stands.
How Elo feeds your signals
The ratings are not just a display — they feed directly into the model's goal expectations through a component called v12elo. When two teams are closely matched, Elo barely changes anything. When there is a clear quality gap, it does real work:
- The stronger team's expected goals are nudged up.
- The weaker team's expected goals are nudged down.
- Win probability shifts toward the stronger side, and BTTS / over-under expectations adjust with the new goal balance.
The effect is deliberately measured, not dramatic — the adjustment is scaled so it sharpens the prediction without overwhelming the base model. Its value is greatest on mismatched fixtures (a large Elo gap), where the plain model tends to under-rate the favourite, and smallest on balanced games, where it should be. The same attack-edge idea also tilts the player-bet Anchor Legs.
What the Elo numbers do not do
- They do not know team news. Elo is built from team performance data. A key striker or centre-back missing is not reflected in the rating.
- They need games to settle. A team with only a handful of matches on record carries a less reliable rating; the model flags reliability once enough games have accrued.
- They are not a price. A big Elo gap tells you who is stronger, not whether the odds on offer are value. Always weigh model probability against implied probability.
---
Next reads
- How the BetSignals Model Works: the full picture behind every signal.
- Understanding Signal Ratings: what ★, ★★ and ★★★ mean.
- Reading Model Confidence Levels: interpreting the probabilities.
---
18+ | Gambling should be enjoyable. If it stops being fun, take a break. For support and advice visit BeGambleAware.org. See our full responsible gambling guide.