A Gaussian, statistically isotropic universe is generous with large-angle correlations: the CMB temperature two-point function should carry substantial power at all angular separations, because inflation stretches fluctuations to every observable scale without preference (Spergel et al. 2003; Planck Collaboration 2016). Measure the temperature at two points 70 or 120 degrees apart, and the standard model says their correlation should be modest but unmistakably nonzero.
Instead the correlation simply stops. COBE, WMAP, and Planck all find the temperature correlation above about 60 degrees anomalously close to zero, with the S_1/2 statistic that integrates the large-angle correlation far below the ΛCDM expectation, at a level realized in well under a percent of standard-model simulations (Copi et al. 2010; Schwarz et al. 2016). The feature is robust to masks, estimators, and instrument swaps across three generations of satellites, which makes the default explanations uncomfortable: a chance fluctuation this deep is rare, foregrounds do not produce it, and no simple parameter shift within the model removes it, because the model's initial conditions guarantee large-angle power by construction.
The standing is a quiet, stubborn anomaly that pairs with the low-multipole power deficit and the aligned low-ell anomalies: the largest observable scales carry less structure, and less random structure, than the inflationary mechanism mandates. The largest angles are also where cosmic variance is permanent, so the verdict will come from polarization and cross-statistics rather than more temperature data.