Of course, we are aware that there will be no CPI release this month. Nevertheless, we continue to update and run our models using the data currently available and a set of assumptions outlined below.
We project the seasonally adjusted month-over-month (MoM) increase in core Consumer Price Index (CPI) for October to be 0.3% (25 basis points), with headline CPI expected to rise by 0.2% (16 basis points). This month, we view the risks as broadly balanced.
Our machine learning (ML) model—described in more detail below—yields projections in line with our bottom-up forecast. While we typically place less weight on the ML model, we give it greater consideration in this instance, as its reliance on a broad set of variables makes it less sensitive to the absence of any specific data series.
From a broader perspective, underlying inflationary pressures remain elevated, even after adjusting for tariff-related effects. The distribution of price changes within core CPI continues to diverge from levels consistent with the inflation target. The persistence of fat-tailed distributions further highlights the prevailing uncertainty surrounding the inflation outlook.
Our medium-term baseline model has been revised downward; further details are provided below.
The FOMC appears divided and is showing signs of internal uncertainty. It is likely that the October CPI report, even if available, would have offered limited clarity in resolving these divisions.
Important note: Our bottom-up forecast has been revised down slightly, reflecting updated assumptions regarding tariffs. We will circulate an updated table with all model forecasts later this week. Additionally, we will include a Q&A section addressing how the models are currently being affected—and may continue to be affected—by the absence of key data.
Our machine-learning model. We have not introduced our machine-learning model yet (documentation is here). However, we have been monitoring this model for months and the out-of-sample performance is accurate. For the record, the model (which includes more than 100 variables) is forecasting 28bps MoM sa for core CPI in October. The model estimates 39% probability of having a 0.2% (or lower) MoM in core CPI, 36% probablity of having a 0.3% in core CPI, and 25% probability of having a 0.4% (or higher) MoM in core CPI. The cumulative densitive function of the MoM in October can be seen here.
Our forecast
We expect headline and core CPI to increase by 16 basis points and 25 basis points, respectively, in October, with non-seasonally adjusted (NSA) levels projected at 324.924 and 331.161. Based on our forecast, year-over-year (YoY) headline and core CPI are expected to print at 3.0% and 3.0%, respectively. Table 1 provides a sectoral breakdown of our seasonally adjusted MoM forecast, with core goods and core services anticipated to rise by 34 basis points and 22 basis points, respectively. Given our MoM forecast for core CPI we expect core PCE prices to grow 24bps MoM in October.
As always, we refrain from making sectoral decomposition comments, as our focus remains on the ex-post distribution of price changes. Table 2 outlines our real-time MoM forecast errors: over the past six months, our core CPI forecast—our primary focus—has been very competitive, with a standard error of 3 basis points and a standard deviation of 8 basis points.
Table 1. MoM sa CPI forecast – details
Table 2. Recent real-time Underlying Inflation MoM (sa) CPI forecast errors
Implications for the “main” model
Revised down.
This is the first time we include Q4 in-sample. We are working under the assumption that core PCE prices will expand 2.8% QoQ saar in Q4. This nowcast is marginally lower (1 tenth) than last week, given lower than expected (Adobe DPI) data. Given our nowcast, the medium-term model forecast for core PCE price inflation is revised down (2 tenths to rounding), as the data are lower than the model own forecast. The model’s latest Q4/Q4 forecasts are as follows: 2.9% in 2025, 2.9% in 2026, and 2.7% in 2027. The forecast remains well above target going forward.
Figure 1. Latest forecast of our “main” model for core PCE price inflation (YoY).