Our Track Record
We show everything — the wins and the losses. Every signal TOWNGRID has issued is scored against the real-world outcome within the predicted window.
Total predictions
125
Correct
108
17 missed
Accuracy
86.4%
Avg lead time
12.4 days
on confirmed events
Accuracy over time
Last 12 monthsEach point is a calendar monthAccuracy % of resolved predictions
Filter125 resolved · 31 pending
Signal performance
| Signal | Market | Predictions | Correct | Accuracy | Avg lead |
|---|---|---|---|---|---|
| ERCOT Grid Stress | ERCOT | 20 | 19 | 95.0% | 2.0 days |
| Renewable Output Variability | CAISO | 19 | 18 | 94.7% | 3.0 days |
| AI Datacenter Load Surge | Multi-ISO | 18 | 14 | 77.8% | 13.1 days |
| Price Breakout — ERCOT | ERCOT | 18 | 15 | 83.3% | 4.1 days |
| Forward Curve Inversion | PJM | 18 | 14 | 77.8% | 13.3 days |
| Natural Gas Correlation Shift | Multi-ISO | 18 | 16 | 88.9% | 9.4 days |
| Capacity Breach Risk | ERCOT | 8 | 7 | 87.5% | 59.5 days |
| Interconnection Queue Acceleration | MISO | 6 | 5 | 83.3% | 50.6 days |
Recent predictions(last 20 resolved)
| Date | Signal | Probability | Window | Outcome | Result |
|---|---|---|---|---|---|
| May 29 | ERCOT Grid Stress ERCOT | 92% | 24-48h | Occurred | Correct |
| May 20 | Renewable Output Variability CAISO | 74% | 3-5 days | Occurred | Correct |
| May 20 | Forward Curve Inversion PJM | 58% | 1-2 weeks | Did not occur | Miss |
| May 19 | Natural Gas Correlation Shift Multi-ISO | 59% | 1-2 weeks | Occurred | Correct |
| May 18 | Price Breakout — ERCOT ERCOT | 88% | 3-5 days | Occurred | Correct |
| May 16 | AI Datacenter Load Surge Multi-ISO | 70% | 1-2 weeks | Occurred | Correct |
| May 15 | Price Breakout — ERCOT ERCOT | 80% | 1 week | Occurred | Correct |
| May 14 | ERCOT Grid Stress ERCOT | 88% | 24-48h | Occurred | Correct |
| May 13 | ERCOT Grid Stress ERCOT | 91% | 3-5 days | Occurred | Correct |
| May 12 | Natural Gas Correlation Shift Multi-ISO | 59% | 1-2 weeks | Occurred | Correct |
| May 9 | AI Datacenter Load Surge Multi-ISO | 80% | 1-2 weeks | Did not occur | Miss |
| May 9 | Natural Gas Correlation Shift Multi-ISO | 71% | 1-2 weeks | Occurred | Correct |
| May 4 | Forward Curve Inversion PJM | 75% | 2-4 weeks | Occurred | Correct |
| May 3 | Renewable Output Variability CAISO | 67% | 3-5 days | Occurred | Correct |
| May 3 | Natural Gas Correlation Shift Multi-ISO | 64% | 1-2 weeks | Occurred | Correct |
| May 3 | ERCOT Grid Stress ERCOT | 92% | 48-72h | Occurred | Correct |
| May 1 | ERCOT Grid Stress ERCOT | 66% | 3-5 days | Occurred | Correct |
| Apr 28 | Natural Gas Correlation Shift Multi-ISO | 74% | 1-2 weeks | Occurred | Correct |
| Apr 27 | Price Breakout — ERCOT ERCOT | 73% | 3-5 days | Occurred | Correct |
| Apr 26 | Price Breakout — ERCOT ERCOT | 88% | 1 week | Occurred | Correct |
How we score predictions
Every signal carries a stated probability and a resolution window (e.g. 3-5 days). We mark an outcome as Correct only when the predicted event is materially observed inside that window — no retroactive window extensions, no partial credit.
- What counts as correct: the predicted event (grid stress breach, price breakout, load surge, etc.) occurs within the stated window at or above the threshold materially signaled.
- What counts as a miss: the window expires without confirmation, or the magnitude falls materially short of the signaled threshold.
- No cherry-picking: every signal that fires is recorded the moment it activates. Outcomes are resolved on a schedule — not on demand when they look favourable.
- Lead time: the interval between issuing the signal and the first confirming observation, for correct predictions only.
If you see a prediction here you believe was scored incorrectly, write to us — we publish corrections.
Want these signals before the event?
TOWNGRID delivers live signal intelligence with multi-timeframe outlooks, webhooks, and real-time alerts.
See plans