Predictive History: Jiang Xueqin's Framework for Reading Geopolitics
Predictive history is a method for forecasting geopolitical events by treating history as a set of recurring structural patterns — imperial decline, elite fracture, resource competition — rather than a sequence of unique, personality-driven accidents. Professor Jiang Xueqin, the Chinese-Canadian educator and geopolitical analyst behind the YouTube channel @PredictiveHistory, built the framework and applies it lecture by lecture to call specific, falsifiable outcomes months or years in advance. This site tracks every one of those calls against what actually happens.
History rhymes; structure outlasts personalities
The claim runs through dozens of Jiang Xueqin lectures cataloged on the by-video index: actors change, but the underlying structure — declining hegemons, rising challengers, brittle regimes, resource chokepoints — repeats closely enough to forecast from. A president, a supreme leader, or a general is a variable operating inside a much slower-moving system; predictive history bets on the system, not the individual, and treats elections, wars, and successions as expressions of pressure that was already building.
That bet paid off clearly in Jiang's April 2024 call that Donald Trump would win the 2024 election (P001), made seven months out, and in his companion read that JD Vance was a credible VP pick (P002b) — both structural calls about where the Republican coalition was heading, not insider tips. The same lecture cycle produced a miss: Jiang's guess that Nikki Haley would be the VP pick (P002a) was wrong once Trump chose Vance instead. Structure predicted the winning ticket; a specific personnel guess inside it did not land. That pairing — right on the macro shape, wrong on a granular personality call — shows up again and again in the record and is the clearest limit of the method.
The Iran war as a stress test
Jiang's most-tested body of predictions concerns the US-Iran war, and it doubles as the best illustration of predictive history in practice. As early as April 2024, he called that the United States would go to war with Iran within two to four years (P011) and that Washington would bomb Iranian nuclear facilities (P012) — both confirmed once the conflict materialized on roughly that timeline. He then predicted the shape the war would take rather than just its existence: that "shock and awe" would fail to produce a quick collapse (P015), that the war would drag on for weeks or potentially years (P018), that Iran would close the Strait of Hormuz (P026) and strike Gulf states including Dubai, Bahrain, and Qatar once fighting started (P027), and that civilian infrastructure — water, dams, power plants — would come under escalating attack (P020). All five held up as the war unfolded through early 2026, including the destruction of Iran's navy (P105) and a decapitation strike on the Supreme Leader (P088). Jiang also called succession years ahead of time, naming Mojtaba Khamenei as the likely next Supreme Leader back in May 2024 (P041).
Not every adjacent call held up. Jiang's prediction that Vladimir Putin would extend a nuclear umbrella over Iran to deter US strikes (P017) was wrong — no such umbrella materialized, and the war proceeded without nuclear weapons on either side, a separate outcome Jiang had also predicted correctly (P016). He was right that China would stay out of direct participation (P071). The pattern across this cluster: structural, systems-level calls about escalation dynamics and regional spillover held up consistently; a call resting on another leader's specific diplomatic choice did not.
What the numbers actually say
As of 2026-09-17, this site tracks 400 predictions drawn from 180 source lectures (179 already transcribed). Of those, 22 have resolved one way or the other — 16 confirmed, 6 wrong — for a headline accuracy of 73%. That figure is worth reading carefully: it's a 22-prediction sample against a pool of 361 still pending or unverifiable, so it will move, sometimes sharply, as individual claims resolve. Nothing here is scored until it's either proven true or proven false; forward-looking claims sit in a separate, unscored "leaning" state until the evidence is conclusive.
Checking the method against the record
Predictive history is falsifiable by design — every call is dated, sourced to a specific lecture, and re-checked against live news rather than retrofitted after the fact. Readers can trace any prediction back to the lecture it came from through the by-video index, or read the background on Jiang's career and the case for treating his forecasts as a serious dataset on the about page. Searching "predictive history youtube" turns up the source lectures directly; this site exists to show, claim by claim, how often the pattern-reading actually holds.
Browse all tracked predictions · Accuracy analysis · All blog posts