Baseline of historically relevant events based on Wikipedia.
Date
Event
08/04/22
August 2022 Military Exercises (Post-Pelosi visit)
04/08/23
Joint Sword
08/19/23
August 2023 Drills (Post-William Lai US transit)
05/23/24
Joint Sword–2024A
10/14/24
Joint Sword–2024B
12/09/24
Unnamed Winter Drills
02/26/25
Unannounced Surprise Attack Simulation
01/04/25
Strait Thunder–2025A
12/29/25
Justice Mission 2025
9 events over 3.5 years translates to a base rate 72% chance that a new round of military drills will be initiated in the 6 months between the beginning of February and the end of July.
The above approach is rather naive, so I decided to look at the average time distance between consecutive drills to account for the potential existence of a "cooldown" effect that makes back-to-back actions unlikely. Using inter-arrival time modeling, we get a very similar base rate estimate of 69%. There's no reason to suspect the average distance between consecutive events is shortening/lengthening, as it seems that distances are almost random.
To this point, it's worth considering that at least some of the drills are not planned with a large advance for a specific date, but are rather triggered by key strategic events (Pelosi visit, Lai US transit).
So what are the key events of the next 6 months that may temporarily higher/lower the probability of a Chinese exercise?
April 2026 – Trump-Xi Summit in Beijing: it seems plausible that both countries will not want to escalate tensions as we approach the highly anticipated bilateral. A deal with China is Trump's White Whale, and Xi might also prefer a softer approach to extract maximum value from any negotiation. If the bilateral is successful, further normalization of the relationship could occur during later rounds of dialogue and negotiations, reducing the chance of an antagonistic action from China.
For these reasons, I adjust my prediction slightly downward.
Files
Why might you be wrong?
China might want to set the stage before the negotiations by increasing pressure on Taiwan, under the impression that Trump would not jeopardize the possibility of a deal with China
The most recent military drills were a consequence of the US supplying arms to Taiwan. It's entirely plausible that both countries will stay on course to their strategic goals, independent of the possibility of normalization during bilateral talks.
A failed Trump-Xi meeting could further inflame tensions, leading to a Chinese escalation
Why do you think you're right?
Baseline of historically relevant events based on Wikipedia.
9 events over 3.5 years translates to a base rate 72% chance that a new round of military drills will be initiated in the 6 months between the beginning of February and the end of July.
The above approach is rather naive, so I decided to look at the average time distance between consecutive drills to account for the potential existence of a "cooldown" effect that makes back-to-back actions unlikely. Using inter-arrival time modeling, we get a very similar base rate estimate of 69%. There's no reason to suspect the average distance between consecutive events is shortening/lengthening, as it seems that distances are almost random.
To this point, it's worth considering that at least some of the drills are not planned with a large advance for a specific date, but are rather triggered by key strategic events (Pelosi visit, Lai US transit).
So what are the key events of the next 6 months that may temporarily higher/lower the probability of a Chinese exercise?
April 2026 – Trump-Xi Summit in Beijing: it seems plausible that both countries will not want to escalate tensions as we approach the highly anticipated bilateral. A deal with China is Trump's White Whale, and Xi might also prefer a softer approach to extract maximum value from any negotiation. If the bilateral is successful, further normalization of the relationship could occur during later rounds of dialogue and negotiations, reducing the chance of an antagonistic action from China.
For these reasons, I adjust my prediction slightly downward.
Why might you be wrong?