ENTSO-E activated balancing price mean reversion Tier 1

Source: ENTSO-E, Germany (50Hertz control area) · entsoe_activated_balancing_price

Germany's real-time activated balancing-energy price, both real confirmed flow directions, each tested against its own forward mean reversion - a core factor at full walk-forward confidence, independently confirmed not redundant with three other real balancing-market document types.

Economic effect

-0.43x (A01) and -0.72x (A02) realized volatility at 5 trading days - the horizon was deliberately narrowed from the usual (1, 5, 10, 20) set after a dedicated statistical-significance check (see below).

Servable directions

Direction	Control area EIC	Sign consistency	Economic effect (h=5)
`A01`	`10YDE-VE-------2`	95%	-0.43x
`A02`	`10YDE-VE-------2`	100%	-0.72x

Why the horizon is narrower than our other mean-reversion factors. A single train/test split initially passed cleanly at every tested horizon (1, 5, 10, 20 trading days) - but before trusting that, we ran a dedicated permutation-significance test and found that a pure random walk, with no real mean reversion at all, can spuriously produce the same pattern at longer horizons. The check confirmed real, statistically distinguishable signal at 1 and 5 trading days for this factor, and correctly flagged 10 and 20 days as indistinguishable from that artifact - so we validated and published only the horizons the evidence actually supports, not the full set the raw correlation alone would have suggested. Both directions then cleared a clean, full-confidence (11 of 12 quarterly walk-forward folds) core factor verdict at the narrowed horizons.

Confirmed not redundant with the rest of the balancing market. We tested this factor against three other real ENTSO-E balancing-market document types - contracted reserve capacity, procured balancing capacity, and aggregated balancing bids - across both flow directions. All 6 (direction × baseline) combinations retained roughly 100-103% of their own raw correlation's magnitude after controlling for each baseline, at a 100% walk-forward hit rate. This factor's own mean reversion is not simply an echo of a more basic balancing-market signal this project already had.

One open, honestly-flagged detail: A01/A02 are the real confirmed codes ENTSO-E's own API returns for this factor's `flowDirection.direction` field - we have not independently confirmed which real-world direction (up-regulation vs. down-regulation activation) either code maps to, so the subject carries the raw confirmed code rather than a guessed label.

Query it

GET /v1/factors?factor=entsoe_activated_balancing_price&subject=10YDE-VE-------2/A01

Swap the subject's trailing code for A02 for the other real direction. Also wired into /v1/factors/latest, /v1/rankings, /v1/screen, and /v1/factors/metadata, the same generic percentile/momentum/volatility machinery our other Tier 1 factors use.

Raw source: ENTSO-E Transparency Platform. Related factors: EU imbalance price (another European balancing-market price tested against its own mean reversion).

Full validation methodology: how we score factors and every real number behind these verdicts in FACTOR_CATALOGUE.md.

Free early access to the API: Get early access →