Based on the ERA 0.25°×0.25° reanalysis data with 1 h interval, the physical quantities of four times per day from May to September in 1990—2019 are calculated, and the anomaly degree of physical quantities of extreme heavy precipitation cases is statistically analyzed. Four testing methods are used for comparative analysis with the objective of establishing the optimal probability prediction equation of extreme heavy precipitation in northeast of Sichuan Province. The following conclusions are drawn from the test results: (1) The more obvious anomaly factors are specific humidity at 700 hPa and 850 hPa, water vapor flux divergence at 700 hPa and 850 hPa, and vertical velocity at 500 hPa, 700 hPa and 850 hPa. The absolute values of most of the mean anomalies are above 1.5, the K index is equivalent to the pseudo-equivalent potential temperature, and the overall anomaly at 08∶00 is more obvious than that at other times. (2) The probability prediction equation of extreme heavy precipitation has a positive forecasting effect. In sample test, the mean forecast probability is 0.652, and the percentile value of 60% is 0.623, and the max value of forecast probability is 0.999. In out-of-sample test, when the prediction index threshold is 0.6, the average accuracy reaches 90%, the false prediction ratio is 9.3%, and the average missing report rate is 0.7%, among which 13 meteorological stations have the missing report rate of 0%. When threshold is 0.65, the average accuracy reaches 92.5%, the false prediction ratio is 6.9%, and the average missing report rate is 0.7%, including the 0% missing report rate of 13 meteorological stations, which shows an even better prediction effect.