Jiang Xueqin's Wrong Predictions: One Miss, Fully Documented

As of 2026-10-24, jiangpredictions.com has marked five of Professor Jiang Xueqin's predictions wrong. Of the 22 that have resolved, 17 were confirmed, which works out to 77% accuracy. This post covers the one miss we can describe in detail, and it is honest about what that number does and does not mean.

The miss we can document: Nikki Haley as VP

The clearest example of a Jiang Xueqin wrong prediction is P002a. In May 2024, he predicted Donald Trump would pick Nikki Haley as his vice president, with the call due to resolve at the August 2024 Republican Convention. It was marked wrong when the convention nominated a different running mate.

The same lecture period produced a related call that held up. P002b held that JD Vance was a strong alternative VP pick for Trump. It was confirmed. So the two claims split: the primary call missed and the hedge landed.

The other four

The tracker currently lists five predictions as wrong. Only P002a was among the predictions supplied for this post, so we won't describe the other four here, and we won't guess at their details. Each one is listed, with its category, timeline and status, on the wrong-predictions page. That page is the full record of every Jiang Xueqin failed prediction we have marked.

How to read 77%

The headline figure is confirmed predictions divided by confirmed plus wrong: 17 of 22, as of 2026-10-24. Treat it with caution for three reasons.

• The denominator is small. Twenty-two resolved predictions is not a large sample. One or two more misses would move the percentage noticeably.
• Most predictions are unresolved. The system tracks 450 predictions, and 403 are still pending, meaning either not yet due or unverifiable. Resolved calls are a small slice of the whole.
• Resolved calls are not a random sample. Predictions resolve when events happen. Near-term, concrete events resolve first, while long-horizon calls stay pending. The 77% describes the predictions that have come due so far, not everything he has said.

Clustering matters too. Many of the confirmed calls belong to one theme. Eleven of the 17 fall under the US-Iran war or the wider Middle East or Iran categories, for example P011 (US war with Iran), P012 (bombing of Iranian nuclear facilities), P026 (closure of the Strait of Hormuz) and P041 (Mojtaba Khamenei becoming Supreme Leader). Several confirmed items are tightly linked, so they are not fully independent hits. If the underlying thesis had been wrong, many of them would likely have failed together.

What counts as wrong, and what doesn't

We mark a prediction wrong only when a reported event clearly contradicts it, as with the convention outcome in P002a. We do not mark it wrong because a timeline is running long. Those stay pending. A prediction such as P011, with a window of two to four years from April 2024, was confirmed once the war began. Had we counted it a miss in 2025, we would have been wrong to.

The reverse also applies. Some confirmed entries carry a "predicted: N/A" date, such as P088 (the decapitation strike on Khamenei) and P105 (destruction of the Iranian navy). Those were confirmed by events, but the record doesn't give a clear date for when the forecast was made. Readers should weigh them accordingly. Our predictive history misses and hits are only as good as that dating.

Why we publish the misses

A tracker that showed only hits would be marketing. The source material is 183 lectures, 182 of them transcribed, and the predictions are pulled from them. Resolution depends on news reports, not on the tracker's own judgment of what Jiang meant. When a call fails, it goes on the record with the same prominence as one that lands.

This page regenerates every three weeks as predictions resolve. The numbers above will change, and the wrong count may rise. The full, current list of misses is always on the wrong-predictions page.

Browse all tracked predictions · Accuracy analysis · All blog posts