A report by the Earthquake Research Committee of the Japanese government released on 26 March 2021 puts the probability of a magnitude 8-9 quake along the Nankai Trough off the coast of central Japan at 70 to 80 percent:

The results stem from high probabilities of two huge offshore quakes over the next 30 years.

One is a temblor with a magnitude of around 8 forecast to occur with an 80% probability along the Chishima Trench off Nemuro, Hokkaido.

The other one, seen occurring along the Nankai Trough off the country’s central to southwestern coast with a magnitude of 8 to 9, has a probability of 70% to 80%.

As an investigative report by Keiichi Ozawa in the Chunichi Shimbun exposed last year, this figure is based on manipulation. The estimate for the Nankai Trough uses a different methodology from the one used for all other areas of Japan. If the same methodology were used, the estimated probability would be slashed to about 20 percent.

When a government panel of earthquake prediction experts revised the probability of the Nankai Trough earthquake scenario from “about 70 percent” to “between 70 and 80 percent” in February 2018, some pointed out that the figure was inflated. The minutes of previous meetings in 2012 and 2013, obtained by the Chunichi Shimbun, revealed the figure had been slammed as “unscientific” at the time.

The seismologists were of the opinion that the probability was not being presented fairly and called for the release of not only the highest figure but a low one as well.

But their argument was swiftly shot down by scholars in the field of disaster prevention, who, the minutes of the preliminary meetings showed, made no secret of their desire to prioritize securing themselves budgets over respecting science.
(…)
During joint sessions the committee held in December 2012 and February 2013, seismologists and disaster prevention experts clashed over whether to mention the low figure for the Nankai scenario in the “main text,” or summary, of the government report.

In the draft version, there was no mention of the fact that the time-predictable model — the measurement of land movements from previous quakes taken at Murotsu Port, northwest of Cape Muroto in Kochi Prefecture — was applicable only to the Nankai scenario, and that applying the model used for all other earthquakes would slash the probability to about 20 percent.
((Japan Times, 2020-07-10)

I am wondering how long these questionable figures will remain unchallenged year after year.