Researchers examined whether electric battery operators that use the same third‑party autobidding software end up bidding like one firm. They used detailed data from the Australian National Electricity Market where every battery’s bids are recorded every five minutes and can be linked to the specific autobidding provider that produced them. The main finding is that bids made by the same provider move together, and this synchronization strengthened after a rule change made it easier to see when electricity was scarce.

To separate two possible causes of that co‑movement — common information versus joint profit‑seeking — the authors estimated each battery’s dynamic value of stored energy. In plain terms, they worked out what each battery would optimally do over time if it were only trying to maximise its own profit. They then simulated (“recleared”) the market using counterfactual bids based on those values to see whether the observed bids could be explained by normal profit‑maximising behaviour.

Those simulations showed that owner‑level profit motives do not explain the observed patterns. In many cases batteries avoided dispatch that would have been directly profitable for the device’s owner because dispatch would lower prices earned by other batteries that shared the same provider but were owned by rival firms. Quantitatively, the behavior is consistent with the algorithms acting as if they put almost as much weight on the rivals’ profits as on their own.

The paper finds this conduct only when a provider controls a substantial share of the capacity that is near the market margin — roughly a 30% share of near‑margin battery capacity, which corresponds to about a 20% share of installed battery capacity. On the current fleet, the authors estimate the effect costs consumers about US$5.5 million per year. Importantly, the pattern appears to come from the autobidding provider itself, not from the asset owners, so standard screens that look only at ownership concentration could miss it.