Most of this site is a count of something that happened. Counting is the easy half. Every one of the patterns below is in the data clearly enough that I trust the measurement, and for none of them do I have an explanation I would defend. I am writing them down because an unexplained result that is stated as unexplained is worth more than one dressed up as insight, and because if you know the answer to any of these I would like to hear it.
Every other release here is quieter in the thirty minutes before the number than in the thirty minutes after. Payrolls is 3.4 times busier afterwards, CPI 3.2 times, PPI 2.3 times. JOLTS is 0.83, which is the wrong side of one.
My working guess is that whatever moves price at 10:00 on a JOLTS day is mostly not JOLTS. Other things land at the top of that hour and the release is getting credit for a move it did not cause. To test that I would need to measure the same window on days when JOLTS is not scheduled at all, and compare. I have not done it, and until I do the number stands as a curiosity rather than a finding.
Seven of the eight releases here move the index less than they used to. That is the measurement, and it is consistent enough across releases that I do not think it is noise.
What I cannot separate is the cause. Either the market has become less reactive to the same information, or the numbers themselves have been landing closer to expectations, which would produce exactly the same chart for a completely different reason. Splitting the record by the size of the surprise rather than by date would begin to separate those, and that is a piece of work I have not built.
While seven releases shrank, PPI grew, from a median of 0.41% in the earlier half of its record to 0.51% in the later half. It is a small sample and one violent print near the end would do this, so the honest answer may simply be that the pattern is not real.
But if it is real, it is interesting, because PPI is the one release on this list that has arguably become more informative rather than less over the period. I have no way to test that with what I collect.
FOMC flips direction between the print and the thirty minute mark 30.4% of the time, roughly twice as often as CPI. The obvious reading is that the press conference contradicts the statement.
The competing reading is that nothing reversed and the move had not finished. The FOMC median at two hours is 0.53% against 0.20% at thirty minutes, so the half hour window is cutting that release in the middle in a way it does not cut the others. Those two explanations predict different things about where price ends up at four hours, which is a measurement I could make and have not.
When two reports print in the same minute, I expected both to look inflated. They do not. GDP looks about 1.5 times larger on the days it shares a minute with Core PCE, while Core PCE looks the same whether GDP is there or not. PPI actually looks smaller on the days it shares with retail sales, while retail sales looks larger.
The pattern is consistent with the quieter release borrowing the louder one’s range, which would mean the market is reacting to one of the two and mostly ignoring the other. That is a satisfying story and I want to be careful with it, because there are only six to nine shared days per pair and satisfying stories are what small samples are best at producing.
The agencies publish the actual numbers. None of them publish the forecast, so consensus here is read from a calendar, and it is the weakest input on the site by some distance. Everything on the surprise pages inherits that weakness.
I do not know how a consensus figure is assembled by the source I read, how often it is revised in the hours before a release, or whether the version I capture is the version the market was actually positioned against. Until I do, I treat the direction results as suggestive and the release-by-release counts as the part worth trusting.
Twenty three to thirty six observations per release, and fewer once any of them is split into groups. Prices are Nasdaq 100 and S&P 500 index CFDs on one minute bars rather than the CME contracts. Every figure above is a count of what happened and not a forecast of what will, and nothing on this site is investment advice.
If you have measured any of this independently and reached a different answer, that is the most useful email I could get. [email protected]
Releases: Core PCE Price Index Consumer Price Index FOMC Rate Decision GDP JOLTS Job Openings Nonfarm Payrolls Producer Price Index Retail Sales