Jiang Xueqin Wrong Predictions: The Misses, Listed Plainly
As of 2026-11-15, Jiang Xueqin has five logged wrong predictions in this independent tracker. That is the direct answer, and it sits beside the wins rather than behind them. The tracker currently records 482 predictions: 17 confirmed, 5 wrong, 433 still pending (not_yet plus unverifiable), and 22 resolved. Headline accuracy is 77% of confirmed plus wrong.
The caveat comes before the celebration or the autopsy. Twenty-two resolved calls is a small denominator. The 433 pending entries are neither hits nor misses yet. The wrong column is a ledger, not a verdict on every lecture in the corpus.
The Wrong-Prediction Ledger, in Plain Numbers
The Jiang Xueqin wrong predictions list is public. The wrong-prediction ledger returns every logged miss; at this snapshot it contains five rows. The corpus behind the scoring is 185 source lectures, 184 of them transcribed. The scoring rule is simple: a prediction has to be specific enough to fail.
Accuracy is confirmed divided by confirmed plus wrong: 17 of 22, or 77%. That figure does not mean 77% of all 482 calls are right. It means 77% of the 22 calls that have resolved. A large pending pile is not evidence of accuracy, and it is not evidence of error. It is unscored. This page does not launder pending calls into either column.
The Clean Miss: P002a and the Vice-Presidential Call
P002a is the clearest citable miss in this snapshot. In May 2024, Jiang predicted that Trump would pick Nikki Haley as his vice president, with the Republican Convention in August 2024 as the timeline. It was a named, dated, checkable call. It failed. The Republican ticket was set with JD Vance, not Haley. The companion call, P002b—that Vance was a strong alternative VP pick—reads confirmed in the ledger. The pair is useful because it shows a hit and a miss at the same event without blurring them.
This is the kind of clean Jiang Xueqin failed predictions entry that makes a wrong-prediction page worth reading: a specific name, a specific event, and a status that cannot be explained away after the fact. P002a did not say “a conventional pick” or “a surprise.” It named Haley. Once the ticket was set, the entry was marked wrong.
What Five Wrong Calls Do—and Do Not—Prove
Five misses do not erase 17 confirmed calls, and they do not get diluted by 433 pending ones. Good tracking keeps the categories separate. The confirmed side includes high-salience calls: P001 on Trump’s 2024 election win, P011 and P012 on a US-Iran war and bombing of Iranian nuclear facilities, P026 on the Strait of Hormuz, P082 on an attack on Venezuela, and P088 on the decapitation strike against Khamenei. Those entries are context for why the wrong column has to stay visible. They are not being relitigated here.
The remaining wrong entries are in the live filter, not in a separate private file. This page regenerates every three weeks, so the filter—not a frozen paragraph—is the canonical list of every miss. At this writing the filter has five rows. P002a is the clean election example; readers who want the full set should use the public ledger rather than trust a headline that says “mostly right” without showing the failures.
Why Predictive History Misses Stay Public
Publishing predictive history misses is counter-intuitive for a prediction site. It is also the only way the headline accuracy means anything. A tracker that lists only confirmed calls is a highlight reel; a tracker that lists wrong calls, pending calls, and unverifiable calls is an audit trail. The full wrong-prediction ledger keeps the current five misses in view. As predictions resolve, the numbers will move. The standard will not: score the named, dated calls, flag the small resolved denominator, and leave the 433 pending entries pending until they can be checked.
Browse all tracked predictions · Accuracy analysis · All blog posts