What the market asked
The market concerned the Bank of Russia’s scheduled key-rate decision at its 24 July 2026 Board meeting. Resolution turned on a binary, mechanical outcome: whether the Board would leave the key rate unchanged, decrease it, or (less commonly framed) raise it, with the official press release expected around 13:30 Moscow time. The graded question the council treated as most likely was: “Will the Bank of Russia make no change to the key rate after the July Meeting?”
What the council concluded
All four models—Gemini, Claude, GPT, and Grok—participated, with Grok acting as chairman. The council’s formal action was BUY_YES at a 63% probability and 72% confidence. Its stated most-likely outcome was “no change,” assigned 64%. The published reasoning was brief and procedural:
“Council recommends BUY_YES (YES). Probability: 63%, Confidence: 72%. Resolution is binary and mechanical: the Bank of Russia Board decides the key rate at its scheduled 24 July 2026 meeting (press release ~13:30 Moscow time).”
In short, the council treated the hold as the base case and sized the position accordingly. It did not surface, in the recorded rationale, a detailed inflation-path or FX-pressure argument that might have elevated the probability of a cut.
What the market itself was pricing
The council’s own 63–64% assessment on “no change” was the figure it used for the graded pick. No richer external order-book snapshot is preserved in the report, so the operative signal is simply that the council viewed a hold as more likely than not, yet far from a lock. That left meaningful room—roughly one-in-three—for a move in either direction, with a decrease being the eventual winner.
What actually happened
The Bank of Russia decreased the key rate. The winning outcome was therefore “Will the Bank of Russia decrease the key rate after the July Meeting?” The council’s graded pick (“no change” at 64%) was incorrect. The Brier score on the call was 0.4096—materially worse than a well-calibrated forecast at that probability would have produced. Because grade basis was “most_likely,” the miss is unambiguous: the outcome the council ranked highest did not occur.
Takeaway
A 64% hold call on a central-bank decision is not reckless, but it is only modestly informative. When the sole recorded rationale emphasizes meeting mechanics rather than the macro data the Board actually weighs, the probability can drift away from the real decision drivers—inflation prints, ruble dynamics, and forward guidance. The council was wrong; the cut was the outcome that paid. Future rate-decision reports will need tighter linkage between the stated probability and the concrete indicators the Bank of Russia has historically cited, or the same modest edge will keep producing the same Brier penalty.
AI-generated analysis for informational purposes only. Not financial advice.