Every single commodity and currency moving like a drum. We treat a presidential Truth Social post as a delta function in geopolitical risk space and measure each asset's transfer function. The overshoot is tradeable.
On March 26, 2026 at 20:11 UTC, a single social media post moved every futures market simultaneously. The response wasn't random—it was a damped impulse response, and each market is a different oscillator. We fitted parametric models, benchmarked trading strategies, and built a gym to test them live.
A Trump Announcement Creates Opportunity—TACO. The post hit at 20:11 UTC. Within 60 seconds, ES gapped +0.94%, Brent crude dropped −3.37%, and Treasury futures spiked +0.24%. By minute 10, the algo overshoot was already reverting. By minute 40, 70–90% of the initial move had dissipated in equity and bond markets.
Three distinct events in 48 hours revealed the fundamental asymmetry: de-escalation overshoots and reverts (fadeable), escalation cascades and amplifies (follow). The physics is different. The strategy must be different.
The fundamental finding: the response is nonlinear in exactly one way. De-escalation events (relief) overshoot because algos front-run the move, then revert as the information is digested. Escalation events (fear) cascade because leverage + loss aversion creates a positive feedback loop.
Each market responds as a different type of damped oscillator. Algo-dominated futures (ES, NQ, RTY) are underdamped—they overshoot and ring. Oil is the anti-correlated factor. Gold is driven, peaking at t=29 minutes. Silver has a 2-minute dead time—human traders placing phone orders.
A live 165fps synthetic market feed. Fire a TACO event and watch the impulse propagate through all instruments simultaneously. The parametric models drive realistic price responses. Toggle strategies to see entry, exit, and P&L in real time.
Closed-loop evaluation: detect TACO events, predict the response curve, simulate fade trades. The parametric model beats TimesFM 7× on de-escalation events (0.09% vs 0.65% MAE) but fails on escalation. The optimal system is an ensemble.
| Event | Direction | Detected | Confidence | Correct |
|---|---|---|---|---|
| TACO | de-escalation | Yes | 0.90 | Yes |
| AntiTACO-1 | escalation | Yes | 0.40 | Yes |
| AntiTACO-2 | de-escalation | Yes | 0.65 | Yes |
| non-event 18:00 | none | Yes (FP) | 0.80 | No |
| non-event 19:00 | none | Yes (FP) | 0.70 | No |
| non-event 20:30 | none | No | 0.00 | Yes |
| non-event 08:00 | none | No | 0.15 | Yes |
| Event | Symbol | Side | Entry | Exit | P&L |
|---|---|---|---|---|---|
| TACO | ES | Short | 5935.0 | 5917.5 | +$350 |
| TACO | NQ | Short | 20337 | 20277 | +$240 |
| TACO | BZ | Long | 72.18 | 72.91 | +$730 |
| TACO | ZN | Short | 110.56 | 110.48 | +$125 |
| AntiTACO-2 | ES | Short | 5890.5 | 5894.0 | -$70 |
| AntiTACO-2 | NQ | Short | 20150 | 20188 | -$152 |
| AntiTACO-2 | BZ | Long | 71.80 | 71.15 | -$650 |
| AntiTACO-2 | RTY | Short | 2100.2 | 2104.8 | -$230 |
| Net | +$526 | ||||