Many central bank decisions are fully expected; markets price the outcome days or weeks in advance via interest rate futures and swaps. The decision itself then matters less than the details.
What tends to move the market on an expected decision:
Changes in the statement wording about future meetings.
Updated forecasts for growth, inflation and rates.
Vote splits or dissenting views.
The tone of the press conference.
Common patterns:
"Buy the rumour, sell the fact": a currency that rallied into an expected hike weakens after it.
A muted first reaction followed by a larger move during the press conference.
Checking market-implied probabilities before the meeting (from rate futures or tools provided by exchanges) shows how much is priced in. A 95% priced hike leaves little room for surprise in the decision itself.
Do you check market pricing before rate decisions, and how do you trade fully expected outcomes?
FXS Prop Desk wrote: A 95% priced hike leaves little room for surprise in the decision itself.
True for the decision, though the other 5% can still matter if the alternative isn't a hold. Sometimes the leftover probability is for a bigger move, say 50bp instead of 25, and then delivering the expected 25 actually reads as dovish. So it's worth checking what the remaining pricing is for, not just the headline odds. The other piece is pricing for the next two or three meetings. If the market has three more hikes priced after this one, the statement only needs to sound slightly less committed for the currency to drop even on a hike. Comparing the implied path before and after the press conference is a decent way to see whether the reaction you're trading matches what rates did.
Fairman, checking what the leftover 5 percent prices is the habit that saved me on the last Fed day. Markets had 25 bp nearly locked, but the residual was skewed to 50, so delivering 25 traded like a dovish miss and USD sold off into the press conference.
I now screenshot the implied path for the next three meetings before the statement, then again after the first 15 minutes of Q and A. If the path lost a full hike while the spot move was tiny, I wait rather than chase the first spike.
Buy the rumour sell the fact still shows up when the currency rallied hard into a fully priced hike. I prefer to be flat into those and trade the statement wording instead of the decision tick.
Vote splits and forecast dots move my bias more than the binary hike or hold once odds sit above 90 percent. That is when I stop treating the decision itself as the trade.
Do you size smaller when headline odds are above 90, or stand aside until the press conference tone is clear? I usually stand aside and trade the path shift.
One edge case for the priced in logic is the Bank of Japan, because it has no fixed release minute. The decision usually lands somewhere around midday Tokyo, so even a fully expected hold leaves you sitting in a window of an hour or more with no idea when the headline drops.
On those days I keep yen exposure to half size from the morning fix until the statement is out. A 20 pip spike in USDJPY on a quiet hold has happened often enough that I no longer treat the timing risk as zero.
The other thing I measure is how far the statement move gets before the press conference starts. For the Fed that gap is 30 minutes and for the ECB it is the half hour between 14:15 and 14:45 Frankfurt, and the first five minute range is often broken in the opposite direction once the questions begin.
That is why I take nothing in the gap unless the statement changed the forward guidance wording outright. A priced hike with unchanged wording is, for me, just a spread widening event followed by a wait.
Fairman, how many basis points of change in the next meeting pricing do you treat as meaningful? I use 5 as my threshold, which might be too tight.
A trade that follows every rule and loses is a good trade. A trade that breaks every rule and wins is a bad one. Outcome bias is the habit of reversing those judgments, scoring a decision by how it ended instead of by the quality of reasoning available when it was made.
In a high-frequency activity like scalping, the damage compounds. Sloppy entries that happen to work get reinforced, and disciplined entries that hit a stop get doubted. Over time, the strategy is edited by random noise rather than by evidence.
One countermeasure is to grade process separately from result. A journal can record two independent marks for each trade: whether the setup met the written criteria, and whether it made money. Four combinations appear, and the lucky violation is the one most worth examining.
Judging a method also requires a sample of trades large enough that variance averages out. A handful of results says very little about an edge, so changes to a strategy are better driven by patterns across many trades than by the last few outcomes.