Bank of Russia key rateHow key rate expectations are calculated

Moscow Exchange lists a futures contract on the Bank of Russia key rate. Each contract is tied to a specific meeting and settles on the day of that meeting at the rate set by the decision. Its price before the meeting is therefore the rate that traders on average expect after the decision: those who think the price is too low buy the contract, those who think it is too high sell it, and the price converges on a shared estimate. We take the contract price for the latest trading day and show it as the expected rate, together with the date it refers to.

The expected rate almost always falls between two neighbouring levels the rate could take after the meeting. The probability of each is measured by distance: the closer the expected rate is to a level, the more likely that level is. The probability of the upper level equals the gap between the expected rate and the lower level divided by the step between the levels; the probability of the lower level is the remainder up to one hundred percent.

An example. The rate is 14.00%, the contract price is 13.91%. The neighbouring levels are 14.00% and 13.75%. The price sits 16 basis points above the lower level out of 25, so 64% goes to a hold and 36% to a cut to 13.75%.

So far the Bank of Russia has moved the key rate in steps that are multiples of 25 basis points, that is, a quarter of a percentage point. The grid of possible levels is therefore built from the current rate with the same step, and the expectation is split between the two grid levels closest to it.

Two levels are a simplification, not a claim that no other outcome is possible: the decision may land outside the neighbouring pair, for example a cut of 50 basis points at once. The contract price includes a risk premium, so it is not a pure market opinion about the decision. And the fewer trades there are in the contract, the coarser the estimate: one large trade can move it more than new information does.

Probabilities are shown only for meetings that have a traded contract. Where there is no contract for a date, the calendar shows a dash: we chose not to fill the gap from neighbouring meetings, because that would be a guess rather than an observation.

After each meeting we compare the rate expected the day before with the decision taken and show the difference in basis points, plus the mean absolute error across all comparisons. There are few observations so far: the contracts have been trading since 29 September 2026, and each meeting adds one comparison. Two or three points are too few to judge the method, which is why the mean error always comes with the number of comparisons.