If you look at the RBA page, you will notice two different sets of numbers that both claim to describe what markets expect for rates:

• The Probability Summary table shows a headline hike probability — say, 36%.
• The Granular Rate Change Probabilities chart shows a collection of bars that, if you add them up, give a smaller total hike probability — say, 31%.

This is not a calculation error. The two figures come from two genuinely different methodologies that slice the same market expectation in different ways. This page explains what each method does, why they diverge, and which one to use for which purpose.

The RBA page displays two distinct probability measures drawn from the same underlying market data but computed under different structural assumptions:

1. ASX single-step binary method — used for the Probability Summary table. Strictly per-meeting, binary outcomes only. Derived from ASX 30-Day Interbank Cash Rate futures using the official ASX RBA Rate Tracker formula.
2. CME expanding probability tree — used for the Granular Rate Change Probabilities chart. Produces a full discrete distribution of the cumulative rate change from today, convolved across all meetings up to the target horizon. Identical in structure to the method applied for the Fed and ECB pages.

Both are internally consistent, both encode the same first moment (expected rate change), and their divergence in aggregate hike probability is a direct consequence of differing distributional assumptions — not a discrepancy in source data or implementation.

The ASX method is the official approach used by the Australian Securities Exchange in its RBA Rate Tracker. For each upcoming RBA meeting, it assumes that exactly two things can happen:

• The cash rate holds unchanged, or
• The cash rate makes exactly one 25bp move (a cut or a hike — whichever direction markets are leaning).

The probability of that move is calculated from the futures price for the month of the meeting, adjusted for exactly how many days in the month the new rate would be in effect.

The Probability Summary table on the RBA page uses this method. It is the number the ASX itself publishes and the figure most Australian financial media report.

Formal Definition

Let the November futures contract settle at price \(F\), so the implied monthly-average cash rate is \(X = 100 - F\). Let \(r_t\) be the current (prevailing) cash rate, \(N\) the total number of calendar days in the contract's expiry month, and \(n_a\) the number of days in that month from the meeting date onward (inclusive of the meeting day itself). The ASX day-weighted probability formula is:

where \(0.25\) represents one 25bp step in percentage points.

Derivation: The implied average rate \(X\) is a day-weighted blend of the rate before the meeting (assumed equal to \(r_t\)) and the rate after the meeting (either \(r_t\) — hold — or \(r_t + 0.25\) — one hike):

Solving for \(p\) yields the formula above.

Common simplification: When the meeting falls in the month before the contract's expiry month (so the entire expiry month is post-meeting, \(n_a = N\)), the formula reduces to \(p = (X - r_t) / 0.25\). This is the case the ASX publishes for most look-ahead horizons. The full day-weighted formula is necessary when the meeting and contract expiry fall in the same month.

Binary constraint: The method forces exactly two outcomes. It cannot decompose a situation where a 50bp hike has material probability. In that case the formula still returns a single \(p \in [0, 1]\) that preserves the mean implied change, but the binary structure misrepresents the true distribution.

The CME method takes a different approach. Instead of asking "hold or one move?" at a single meeting, it builds a full probability tree that spans all upcoming meetings and tracks every possible cumulative outcome — hold, +25bp, +50bp, +75bp, −25bp, and so on.

The result, at any given horizon, is a bar chart of cumulative outcomes: the market-implied probability that rates will be exactly X bp higher (or lower) than today by that meeting.

Summing all the positive bars gives the probability that rates will be higher in any amount by that meeting — the "any hike" probability. This is what the Granular Rate Change Probabilities chart displays, and it is the same method used for the Fed and ECB pages on this site.

Formal Definition

For each pair of consecutive meetings \(i\) and \(i+1\), extract the incremental implied change \(\delta_i\) in basis points from the corresponding futures contracts. At meeting \(i\), compute:

This produces a two-point distribution at meeting \(i\) with mean exactly \(\delta_i\):

The cumulative distribution at meeting \(k\) is the discrete convolution of all per-meeting distributions from meeting 1 through meeting \(k\):

where \(*\) denotes discrete convolution and each \(\mathbf{P}_i\) is the two-point distribution defined above. The final distribution \(\mathbf{P}_k\) gives the probability of every possible total rate change from today's rate to the rate in effect at meeting \(k\).

For display, outcomes are sorted by probability, the top 9 are retained, and probabilities are renormalized to sum to 1. The aggregate "hike probability" is \(\sum_{j: c_j > 0} P_k(c_j)\), summing over all positive cumulative changes \(c_j\).

Relationship to the detailed CME FedWatch algorithm: The incremental decomposition above is equivalent to the within-month rate extraction described on the CME Expanding Tree Methodology page, applied iteratively. See that page for derivation of \(\delta_i\) from futures settlement prices and the continuity constraint across anchor months.

Here is the key insight: both methods encode the same expected rate change. They disagree on what the distribution around that expectation looks like.

The ASX method says: "I will represent all the uncertainty as a single 25bp move with probability \(p\)." That forces the full expected-change mass into one probability number.

The CME tree says: "I will let the distribution spread out. There is a chance of a +50bp cumulative move, and that +50bp outcome contributes twice the rate change per unit of probability." Because large outcomes are more rate-change-efficient, the tree can reach the same mean rate change with a lower total probability of any positive outcome — since some of the heavy lifting is done by the tails.

The gap grows with each additional meeting in the horizon, because the convolution adds more mass to the tails. For the very first meeting (only one step away, incremental move at most 25bp), the two methods nearly coincide.

Formal Analysis of the Divergence

Let \(\mu\) be the common mean implied change (in basis points) at a given meeting horizon. Both methods preserve \(\mu\) by construction.

Under the ASX binary method (hike direction), the single-step probability is:

(using the simplified form where \(n_a / N = 1\); the day-weighted version modifies the denominator but the principle is identical).

Under the CME tree, the aggregate hike probability at the same horizon is:

where the distribution \(\mathbf{P}_k\) is a convolution that places mass at outcomes \(c_j \in \{0, 25, 50, 75, \ldots\} \cup \{-25, -50, \ldots\}\).

The mean constraint requires:

Rearranging and comparing with \(\mu = 25 \cdot p_{\text{ASX}}\):

Since every term \((c_j - 25) \cdot P_k(c_j) \geq 0\) for \(c_j \geq 50\), we have:

with equality only when all probability mass in the CME tree falls at exactly 0bp or exactly 25bp (i.e., the first meeting, before convolution distributes mass to larger outcomes). The divergence grows monotonically with convolution depth — i.e., with the number of meetings in the horizon — as more mass accumulates at \(c_j \geq 50\).

Here is a concrete example using the RBA November 3, 2026 meeting, with a current cash rate of 4.35% and an ASX futures-implied average rate of 4.435% for the November contract. The implied expected change from today is +8.5bp.

ASX Single-Step Result

The November contract expires at the end of November. The meeting falls on November 3, so the new rate (if changed) would be in effect for 28 of the 30 days in the month (na = 28, N = 30). The day-weighted formula gives:

CME Expanding Tree Result

The CME tree, computed cumulatively from today through the November meeting, spreads the same +8.5bp mean across a distribution:

Side-by-Side Comparison

Both methods agree on the expected change (+8.5bp). The ASX method concentrates it as a clean 36.4% chance of one 25bp hike. The CME tree spreads it, yielding a lower summed-hike total (31.3%) but allowing for non-zero probability of larger cumulative moves. Neither is wrong — they are answering slightly different questions.

Numerical Verification: November 2026

Parameters: \(r_t = 4.35\%\), \(X = 4.435\%\), \(N = 30\), \(n_a = 28\) (meeting Nov 3, days Nov 3–30).

ASX Calculation

Comparison: the simplified formula \(\mu / 25 = 8.5 / 25 = 0.340 = 34.0\%\) omits the factor \(n_a / N = 28/30 < 1\) and understates the probability by 2.4 percentage points.

CME Tree Calculation

The tree at this horizon produces the following distribution (renormalized top outcomes):

Mean check: \(-0.35 + 0 + 6.875 + 1.85 + 0.15 = 8.525 \approx 8.5\text{bp}\) ✓

Aggregate hike probability: \(27.5 + 3.7 + 0.2 = 31.3\%\)

Verification of the Divergence Formula

So \(p_{\text{ASX}} - p_{\text{CME}} \approx 4.1\%\), which matches \(36.4\% - 31.3\% = 5.1\%\) to within rounding on the displayed probabilities.

This site uses the two methods in complementary roles:

• Probability Summary (ASX method): The headline figure for the RBA. It matches the number the ASX RBA Rate Tracker publishes, the number cited in Australian financial press, and the number most appropriate for comparing directly to official central bank communications. Use this when you want a single, simple probability for a given meeting.
• Granular Rate Change Probabilities (CME tree): The full distribution view. Use this when you want to see the tail risks — the probability of a 50bp move, the skew of the distribution, or how uncertainty compounds across multiple meetings. This chart is built with the same method used for the Fed and ECB pages, making cross-central-bank comparisons consistent.

For further detail on the CME expanding tree algorithm, see the CME Expanding Tree Methodology page. For the full RBA probability dashboard, return to the Reserve Bank of Australia page.