Predictive History: How Jiang Xueqin Turns the Past Into Forecasts
Predictive history is the forecasting method Professor Jiang Xueqin teaches on the Predictive History YouTube channel: read civilizational patterns and structural forces, lean on the tendency of history to rhyme, and then make specific, timestamped claims about what comes next — claims that can be checked against events as they unfold. It is not punditry or gut instinct. It treats the future the way a weather service treats a storm system: as something a long enough record can constrain.
More than 160 of the Jiang Xueqin lectures are now transcribed in this tracker, organized by video. The project is narrow and unforgiving: extract every claim, attach a date and a category, and re-check each one against live news until it is either confirmed or proved wrong.
Three ideas that anchor the method
History rhymes. The recurring premise is that geopolitical situations are re-runs of earlier ones — the same structural pressures producing the same moves under different names. A leader's choices are constrained by the position their state occupies, not by personal temperament. That is why the map, the trade balance, and the demographics get weighed ahead of whoever happens to hold office.
Civilizations decline on a schedule. A running theme in the lectures is imperial overstretch — the idea that hegemonic powers follow recognizable phases of overreach abroad, fiscal strain, and a rival's ascent before they recede. The United States, in this reading, sits in a late phase of that arc, which is what licenses predictions about retreat, overextension, and new wars.
Structure beats personality. Where most commentary fixates on leaders and elections, the framework treats them as secondary. The 2024 cycle is the cleanest example: Jiang called Trump's re-election (P001) in April 2024, and split the vice-presidential picks — JD Vance as a strong alternative (P002b), confirmed, against Nikki Haley as the nominee (P002a), wrong. The structural call about who the base would reward landed; the specific name missed. The framework explicitly expects some of both.
How the claims get tested
This is where the statistical half of the project matters. Every prediction extracted from the lectures carries a category, a timeline, and a predicted date, and each is re-checked daily against fresh reporting. A claim resolves as confirmed when the event it described is reported by a news source, and wrong when its deadline passes or the situation conclusively ends otherwise. There is no partial credit; the headline accuracy is strictly binary.
As of 2026-06-24, the system tracks 339 predictions drawn from 167 source lectures (166 transcribed). Of those, 20 have resolved — 15 confirmed and 5 wrong — for a headline accuracy of 75%. The remaining 304 are still pending, either waiting on their timelines or, in a small share, judged unverifiable.
That denominator is the honest caveat. Twenty resolutions is a thin sample, and 75% across 20 calls is wide enough that it could settle higher or lower as the 304 open predictions — especially the multi-year ones on Iran, China, and the global order — come due. Treat the figure as a status report, not a verdict.
The Iran cluster shows the method under stress. Jiang predicted a US-Iran war (P011), the bombing of Iranian nuclear facilities (P012), the failure of Shock and Awe (P015), an open-ended conflict measured in weeks to years (P018), and the targeting of civilian infrastructure (P020) — all confirmed as the 2026 escalation played out much the way the structural argument anticipated. The framework also flagged the limits of restraint: nuclear weapons would not be used (P016), confirmed, and China would not meaningfully join the fight (P071), confirmed. But it misfired, too. Putin's promised nuclear umbrella over Iran (P017) never materialized, and the early read on Iran's 2024 presidential election (P040) picked the wrong candidate. Structure constrains; it does not dictate.
Where to watch it work
The method only means something because every call sits in public, dated, and revisable. Browse the full set of claims and their sources by lecture, or read more about the professor behind them. The predictions still on the clock are the real test — and most of them have not come due yet.
Browse all tracked predictions · Accuracy analysis · All blog posts