Historical test result
Fed reversal retest 2017-2023: 0.16% a year on 11 pairs
My 2017-2023 retest on 11 FX pairs showed 0.16% a year. You get the full seven-year picture below in plain words, and FX means foreign exchange.
What you may have heard
The claim is that foreign currencies reverse against the dollar after a Fed announcement. The Fed is the US central bank that puts out rate news. That picture feels clear after a strong Fed day when prices jump and then pull back the next morning. You open the chart and the move looks obvious after the fact. Those mornings stand out on the chart and in chat. The quiet days get less attention and fade fast. A clear story plus strong memory can feel like proof. A feeling is still a starting point for a check, and a check needs all days in the period, including the quiet ones you forget.
What my retest covered
My 11-pair 2017-2023 retest showed 0.16% a year and lost in 2019-2021. A retest is a historical simulation on later prices. Wang’s 2018 dissertation reports minus 50.056% a year in foreign-currency excess return for 1999-2015 in its G14 sample. I keep that background apart from what follows, which is only my 2017-2023 run.
I checked 11 FX pairs from 2017 to 2023 across 55 Fed dates. The pairs were AUDUSD, USDCAD, EURUSD, USDJPY, USDNOK, NZDUSD, USDSGD, USDZAR, USDSEK, USDCHF and GBPUSD. A pair is one currency priced in another, like EURUSD for euro in dollars. This was a long-dollar run on those 11 pairs. The announcement anchor was 14:00 New York time, with entry twelve hours later and exit twenty-four hours after the announcement. Each pair used exposure equal to 0.5 divided by 11 times the account value. Total exposure was half the account value. Each position had a 1.99% stop, so the full 11-pair set risked 0.995% at the start. The stop set a modelled exit level in the test. Live fills can differ from that price. Both sides of costs were included.
What the numbers showed
From 2017 to 2023 it showed 0.16% a year, 1.12% in total over seven years, with a largest dip of 1.03%. A dip, also called drawdown, is a fall from a prior peak. The first part from 2017 to 2018 showed plus 0.51% a year. The middle part from 2019 to 2021 showed minus 0.14% a year. The last part from 2022 to 2023 showed plus 0.26% a year. Four of seven years and 54.55% of the 55 events were positive. Only 3 of the 11 pairs added gains on average in both later parts. That mix is why the run stays rejected. I keep this rule off my list because the result stayed tiny across seven years and passed only 4 of 17 checks.
What the start risk means in money
The test risked 0.995% at the start for the full set. On a purely hypothetical 25,000 dollar account, 0.995% equals 248.75 dollars, to see next to a clearly named hypothetical 1,250 dollar loss limit. This is initial-stop arithmetic, never a forecast of loss or return, and not evidence of intraday drawdown, broker compliance, or challenge safety.
How you check your own report in five minutes
Open your report, split it by year, and check whether gains came from many years or from one short spell. Pull the yearly split in your own report and look at each year on its own. If many years added a little, the yearly number rests on more ground. If one short spell did most of the work and the rest sat flat or down, you treat the single yearly number as weak. Ask for position size and both sides of costs next to the yearly percent when they are missing, so you see what the percent rests on.
What you do next
You write down the idea, the period you checked and the yearly percent you saw. You add the position size and both sides of costs you used. You file the note where you keep past checks. When a similar claim comes up, you open the note first. You stay with plain checks like this each time you face a new claim. Clear notes keep your work light and your list clean.
Limits
These are historical test results, not a live account. If you want to guard your challenge, try my free Prop Rule Guard at https://stratcorealpha.com/rule-guard?ref=website.