Particle physics
9 min read
Are flavour anomalies new physics?
What is a flavour anomaly?
“Flavour anomalies” are measurements in heavy-flavour decays (usually of B mesons) that sit awkwardly with Standard Model (SM) expectations. They’re interesting physically because flavour-changing processes are rare and often loop-suppressed, so new heavy particles can leave indirect signatures even if they cannot be produced directly. But they are also a great teaching example for statistical literacy: how do we decide whether a deviation is a real signal or just a fluctuation?
In practice, an “anomaly” usually means a result differs from the SM prediction by a few standard deviations. That is not yet discovery-level. The key word is tension: something worth investigating, but not yet conclusive.
Where do the anomalies appear?
Most attention has focused on semileptonic B decays. Two main categories appear repeatedly:
Rare neutral-current decays (b → s l⁺ l⁻)
Examples include B → K l⁺ l⁻ and B → K* l⁺ l⁻. These are strongly suppressed in the SM, so they are sensitive probes of new physics. A clean way to test lepton flavour universality (LFU) is to compare muons and electrons using ratios such as
R_K = Br(B → K μ⁺ μ⁻) / Br(B → K e⁺ e⁻)
These ratios are designed so many hadronic uncertainties cancel. Importantly, as new data arrive and systematics improve, apparent deviations can weaken or disappear. This is a normal and healthy part of the scientific process.
Charged-current decays (b → c τ ν)
Examples include B → D τ ν and B → D* τ ν. Here, ratios like R(D) and R(D*) compare tau decays to lighter leptons. These measurements are experimentally challenging because taus decay quickly and involve neutrinos, so progress often comes from improved reconstruction and larger datasets rather than a single decisive measurement.
Why global fits are essential
Flavour physics rarely hinges on one observable. Instead, researchers perform global fits: combine many measurements (branching fractions, angular observables, LFU ratios, different kinematic bins) and analyse them within a common framework.
For rare decays, this is often done using an effective description in which potential new physics shifts a small set of parameters (often called Wilson coefficients). The question becomes: is there a consistent pattern of shifts that explains the whole dataset better than the SM?
Global fits are powerful, but they can also mislead. A pattern of mild deviations can look compelling when combined. Alternatively, giving the fit too many free parameters can artificially increase the apparent significance. Understanding the assumptions behind a fit is just as important as looking at the headline result.
Correlations: the hidden complication
Many measurements are not independent. Correlations arise because:
the same detector systematics affect multiple observables,
several measurements depend on the same hadronic form factors,
neighbouring kinematic bins share uncertainties from unfolding and background modelling.
If correlations are ignored, you can double-count information and overestimate significance. If they are handled poorly, you can also dilute a real effect. For students, the key lesson is that covariance matrices are not optional extras — they are part of the measurement.
Theory uncertainties and nuisance parameters
Flavour anomalies sit at the boundary between precise electroweak physics and complicated QCD. Theory uncertainties are therefore central. These include uncertainties in form factors, long-distance hadronic effects, and approximations used in calculations.
In statistical analyses, these are usually treated as nuisance parameters. The way they are modelled (for example, how conservative or data-driven they are) can noticeably affect the outcome of a global fit. Interpreting an anomaly therefore requires understanding not just the experimental error bars, but the theory assumptions behind them.
The look-elsewhere effect
If you search across many observables, bins, and new-physics scenarios, you should expect some deviation to appear just by chance. This is the look-elsewhere effect. In flavour physics, it can arise when:
trying many possible new-physics operator patterns,
focusing on whichever observable currently shows the largest discrepancy,
slicing data into many kinematic regions and highlighting the most unusual one.
A useful distinction is between local significance (how unusual one result is) and global significance (how unusual it is after accounting for all the ways you searched).