Skip to content
Start free
Get App

The Market Lies at 8:30 AM. I Tested 210 News Releases.

Analysis of 210 scheduled US macro releases (Fed, inflation, jobs) since 2020 using minute-level futures data, testing whether news-trading intuitions hold up.

Key Takeaways

  • Scheduled macro news concentrates a disproportionate share of market movement into very few minutes, but those minutes are mostly won by speed, not by reading headlines.
  • What markets react to changes over time with prevailing fears, so backtests of news strategies must account for regime shifts.
  • First-move direction is a weak predictor of the rest of the day, and early moves often partially reverse, especially around Fed events.
  • Strategy performance can differ sharply on news days versus ordinary days, so filtering or inverting strategies on event days can matter more than finding new signals.
  • Results are suggestive rather than conclusive given small samples, and most commercial news-trading content lacks this kind of disclosed testing.

What the video covers

  • Opens with the October 13, 2022 inflation report, where NASDAQ 100 futures fell sharply in the first minute before reversing and closing higher.
  • Builds a verified calendar of 210 events (52 Fed decisions, 80 inflation reports, 78 jobs reports), cross-checked against market volume since official sources were not automatically downloadable.
  • Finds that the first 5 minutes after releases account for only about 0.04% of sample minutes but roughly 2.3% of NASDAQ price movement, with post-release minutes carrying about 50 times normal movement.
  • Shows market sensitivity changed over time: the reaction to inflation reports rose more than 11-fold from 2020 to 2022, then shrank back to 0.3-0.4%.
  • Compares markets: stocks react most to inflation reports, while currencies and Treasuries react most to Fed decisions; oil barely reacts.
  • Tests a read-the-headline momentum approach and finds it loses money on every cost assumption, since most of the first move happens within the first minute.
  • Finds the first 5 minutes predict the following hour's direction no better than a coin flip, with a tendency for early moves to partly reverse, especially around the Fed press conference.
  • Reports that mean-reversion strategies performed better on news days while breakout strategies performed worse, and skipping news days would have improved a breakout strategy's results substantially.
  • A fade of the first 5 minutes around jobs reports, held until the 9:30 cash open, was profitable overall and beat randomized controls, though the effect is inconsistent across years.
  • Acknowledges limitations: sample sizes of 39 to 208 events, unverified causal explanations, and the risk that any edge has already been arbitraged away.

Chapters

  1. 00:00Opening: The October 2022 Inflation Report Selloff
  2. 00:57Building the Event Calendar and Market Dataset
  3. 01:42How Much Do Scheduled News Events Actually Matter?
  4. 02:58Which Markets React Most to Fed, Inflation, and Jobs News
  5. 04:35Testing the Read-the-Headline Momentum Approach
  6. 06:06Do the First Five Minutes Predict the Day?
  7. 06:51Existing Strategies on News Days vs. Ordinary Days
  8. 08:36Fading the Jobs Report Opening Move and Robustness Checks
  9. 10:38Limitations and Conclusions

Answers

Questions about this video

How much of the market's movement happens during scheduled news releases?

The first five minutes after the 210 releases total about 17 hours, roughly 0.04% of sampled minutes, yet they contain about 2.3% of NASDAQ price movement. Each minute right after a release carries around 50 times the movement of an ordinary minute.

Does reading the news headline and trading in its direction work?

No. In tests, entering in the first minute's direction and holding 15 minutes lost money under every slippage assumption tested, because most of the first move is already complete within the first minute.

Did the jobs report reversal strategy hold up in the analysis?

It was profitable overall and beat randomized controls, but the effect was inconsistent across years, with some years positive and others negative. The author explicitly notes the sample size is too small to be certain of any edge.

Full Transcript — Download SRT & Markdown

00:00
Speaker A
October 13th, 2022, 8:30 a.m. Eastern. The inflation report comes in above expectations. In the first minute, NASDAQ 100 futures fall $379 points. On one contract, that is $7,585 in 60 seconds. The low comes at 933, 3 minutes after the stock market opens.
00:25
Speaker A
Then the index rallies 596 points and closes the day above where it was before the release. The news was bad. The reaction was correct. And anyone who traded it lost money. Being right and losing anyway. Married men will know the
00:42
Speaker A
feeling. Somewhere someone sold that exact low with total conviction. He sells a course now. I wanted to know whether that day was an accident or a pattern. So, I took every Fed decision, every inflation report, and every jobs
00:57
Speaker A
report since 2020, and measured 10 markets minuteby minute around each one. The calendar comes from two official sources. Fed decision dates from the Federal Reserve, inflation, and jobs report dates from the Bureau of Labor Statistics. The Bureau of Labor
01:14
Speaker A
Statistics blocks automated downloads, a statistics agency protecting its statistics from people who want to read them. So every date was verified against the market itself. On a genuine release, volume in 10-year Treasury futures in the first 5 minutes is typically around
01:32
Speaker A
10 times the level of the half hour before. No spike, no release. That caught one wrong date, one market holiday, and several missing months.
01:42
Speaker A
Final calendar, 52 Fed decisions, 80 inflation reports, 78 jobs reports, 210 events. The market data, one minute bars for 10 futures contracts from the NASDAQ and the S&P 500 to treasuries, oil, gold, the euro, and the yen. Days where
02:01
Speaker A
a futures contract rolls to the next one are excluded wherever they would distort a return. First question, how much does news actually matter? The first 5 minutes after these 210 releases add up to about 17 hours. That is 0.04%
02:19
Speaker A
of all the minutes in the sample. Those 17 hours contain 2.3% of all the NASDAQ's price variants since 2020. So a minute right after one of these releases carries around 50 times as much movement as an ordinary minute. 17 hours. That's
02:38
Speaker A
less time than the average day trader spends per week staring at a one minute chart. The difference is these 17 actually mattered. And it isn't only the minutes. On Fed days and jobs report days, the NASDAQ's typical daily move is
02:53
Speaker A
about 55% larger than on a day without a release. On inflation days, about 10% larger. But what moves the market changes with what it's afraid of. In 2020, the first 15 minutes after an inflation report moved the NASDAQ by a
03:09
Speaker A
typical 0.16%. Nobody cared. In 2022, the same 15 minutes moved it by 1.78%. More than 11 times as much. A generation that learned trading on stimulus checks and cheap leverage met a central bank that had sobered up. The traders hadn't.
03:31
Speaker A
Since then, the reaction has shrunk back to 0.3 to 0.4%. Inflation reports have still been the biggest scheduled mover every year since 2021, but the panic is gone, which matters if you back test news strategies. An edge measured in 2022 was
03:50
Speaker A
measured in a market that no longer exists. Most news trading courses were recorded in 2022. Draw your own conclusions. Next, which markets actually react? To compare them fairly, I measured each market's first 5 minutes in multiples of its own normal
04:09
Speaker A
volatility at that time of day. Inflation reports hit stocks hardest. The NASDAQ and the small cap Russell move about 2 and 1 half times their normal range. Fed decisions look completely different. The euro, the yen, and 10-year treasuries move four and a
04:25
Speaker A
half times their normal range. Stocks only about twice. The Fed is a rates and currency event. Stock traders just assume it's about them. Like a guy who thinks the waitress is flirting. She isn't. She's working. Gold reacts strongly to the Fed, too, at more than
04:42
Speaker A
three times normal. And oil barely registers. After all three releases, crude moves about as much as on an ordinary morning. The most powerful central bank on Earth speaks. Oil doesn't notice. Now to trading the intuitive approach. Read the reaction
04:59
Speaker A
and go with it. The problem is speed. In the median event, 86% of the first 5 minutes of movement is already over inside the first minute. I tested it anyway. See the first minute enter in that direction at the next minute's
05:15
Speaker A
open. Hold 15 minutes. One point of slippage per side. - $24,765 over 208 events. Four points minus $49,725.
05:30
Speaker A
8 points minus 83,05. No slippage assumption makes it profitable. 1 minute data can't show who wins the first seconds. It does show who doesn't. Anyone who reads the headline first. If your news strategy starts with reading the headline, you are not the
05:48
Speaker A
trader. You are the liquidity. Second intuitive idea. The first move shows the direction of the day. It doesn't. The first 5 minutes match the direction of the following hour 47% of the time after inflation reports, 48% after Fed
06:06
Speaker A
decisions, and 40% after jobs reports. A coin would have done better, and a coin doesn't need a Bloomberg terminal. For jobs reports and the Fed, the correlation is negative, around minus0.3.
06:20
Speaker A
Big first moves tend to be partly taken back. The Fed shows it best. The statement comes out at 2 p.m. The press conference starts at 2:30. Across all 52 decisions, the press conference produced the larger move 73% of the time,
06:36
Speaker A
typically twice as large, and it went the opposite way in nearly half of the cases. Entire YouTube careers are built on live streaming the 2 PM statement. 30 minutes later, the chairman walks out and the stream gets very quiet. That
06:51
Speaker A
reversal tendency has a real cost for anyone running an intraday strategy. I took the thousand strategy settings from my day trading study, nine strategy types on 10 markets, and compared their results on news days with every other day. The pattern is clean. Mean
07:08
Speaker A
reversion strategies, the ones that bet against stretch moves like VWAP fades and RSI and Ballinger reversals, did better on news days in 70 to 78% of their settings. The opening range breakout and the gap fill did the opposite. Worse on news days in 62 and
07:27
Speaker A
70% of their settings. The opening range breakout, the one strategy that survived that entire study, lost money on all three kinds of news day. Across 200 news days, minus $52,270.
07:42
Speaker A
On all other days, plus $149,95. Just skipping those 200 days would have raised its total from 96,825 to 149,95.
07:56
Speaker A
The best trade it could make on a Fed day was none. It lost on news days in six of seven years, and fewer than two in a 100 random sets of 200 ordinary days were as bad. Every trading guru has
08:10
Speaker A
an economic calendar on screen. Almost none of them use it to switch anything off. If first moves overshoot, the obvious test is to bet against them. The rule for jobs reports, the number is released at 8:30. Wait five minutes,
08:26
Speaker A
then take a position against the direction of those five minutes and hold until 9:30 when the regular stock market opens. One NASDAQ contract, two points of slippage per side plus commission, over 75 jobs reports, plus $22,795.
08:45
Speaker A
Profitable in six of seven years and in both halves of the sample. the control.
08:50
Speaker A
The identical trade at 8:30 on days with no release drawn 10,000 times. The median outcome was minus $7,375.
09:01
Speaker A
Two random baskets in 10,000 did as well. That's rarer than a signal group that posts its losing trades. I checked this result three times because I didn't believe it either. The timing is the interesting part. exiting after 15 or 30
09:16
Speaker A
minutes loses money in every variant I tested. The reversal doesn't come right after the release. It comes around 9:30 when the cash equity market opens. My interpretation which the data is consistent with but can't prove. At 8:30 only futures trade, liquidity is thin
09:36
Speaker A
and the first reaction is set by relatively few participants. algorithms, a few hedge fund desks still running on last night's cocaine, and people paying $99 a month to be told what the algorithms already did. At 9:30, pension funds and mutual funds arrive, read the
09:54
Speaker A
same report, and frequently disagree. It survives double the slippage at plus 16,795 and eight points per side at plus 4,795.
10:06
Speaker A
Waiting 1 3 10 or 15 minutes instead of five also works as long as the position is held for about an hour. And the same fade does not work for inflation reports or Fed decisions. It is specific to the
10:20
Speaker A
jobs report. One more result for people who don't trade intraday at all. On the 74 jobs report days in the sample, the NASDAQ returned minus0.25% on average. Other Fridays without a release averaged plus0.06%.
10:38
Speaker A
Only 1.7% of random sets of ordinary Fridays came out as weak. But it's not a consistent pattern. It comes from 2020, 2022, 2025, and 2026. While 2021, 2023, and 2024 were positive. And fewer than half of all jobs days were actually
11:00
Speaker A
down. A handful of big sell-offs drive the average. This is the kind of result that becomes a proven seasonal edge in a newsletter. It isn't one. For comparison, Fed decision days averaged plus 0.36%, but 1 in8 random samples matched that.
11:18
Speaker A
And the well-known prefed drift, stocks rising in the 24 hours before a decision was visible. But 1 in 10 random samples matched it. and it lost money in 2025 and 2026. Academic edges have a habit of retiring shortly after the paper is
11:35
Speaker A
published, usually to a hedge fund in Connecticut. The limitations. Every result here rests on 39 to 208 events. Enough to rule ideas out, not enough to be certain about any edge. I asked a lot of questions of this data.
11:52
Speaker A
The random basket tests measure luck within each result, but they don't fully correct for how many things I looked at.
12:00
Speaker A
One minute data says nothing about the first seconds, and fills during releases can be worse than the slippage I assumed. And one extraordinary year, 2022, dominates several results. The right next step for the jobs report fade is a forward test with small size.
12:19
Speaker A
That's four limitations, which is four more than the average trading course discloses. To summarize, scheduled news packs a disproportionate share of all market movement into a few minutes per month, and what the market reacts to changes with what it's afraid of. The
12:36
Speaker A
Fed is mainly a rates and currency event. Inflation reports hit stocks hardest. Oil doesn't care. Reacting to the headline lost money under every cost assumption and the first move was not a signal. Mean reversion strategies did better on news days. The opening range
12:54
Speaker A
breakout did worse. For the strategy that survived my thousand back tests, skipping news days raised its result by more than half. And fading the first 5 minutes of the jobs report until the 9:30 open passed every test I ran on 75
13:09
Speaker A
events. A lead, not proof. Everything else sold under the label of news trading, the alerts, the squawk boxes, the breaking news thumbnails, is mostly speed sold to people who will never have it. The full research is in the Algo
13:25
Speaker A
Vibes research lab, the verified event calendar, the minute level event engine for all 10 markets, the cross market reaction analysis, the news day filter for intraday strategies, the jobs report fade, and the random basket luck test.
13:41
Speaker A
The link is in the description. And if there's a release you want tested with the same method, put it in the comments.
Topics:news tradingeconomic calendarNASDAQ futuresFed decisionsinflation reportjobs reportintraday tradingbacktestingmarket microstructuremean reversion

Get More with the SozAI App

Transcribe recordings, audio files, and YouTube videos — with AI summaries and speaker detection. 30 minutes free.

App Store Google Play
Or transcribe another YouTube video here →

Free tools: TXT to SRT · SRT Validator · Merge SRT · Subtitle to Text · All tools

More from Algovibes

View All
I Gave AI 2.4 Million Minutes of Nasdaq Data — Transcript19:51
I Gave AI 2.4 Million Minutes of Nasdaq Data — Transcript
2,585 1 English
Everyone Says Market Regimes Fix Trading Strategies. I Tested It. — Transcript8:26
Everyone Says Market Regimes Fix Trading Strategies. I Tested It. — Transcript
1,273 1 English
Archangel Raphael’s Urgent Message for Humanity | Gabrielle… — Transcript44:05
Archangel Raphael’s Urgent Message for Humanity | Gabrielle… — Transcript
7,167 1 English
Yellowstone National Park Vacation Travel Guide | Expedia — Transcript8:06
Yellowstone National Park Vacation Travel Guide | Expedia — Transcript
855 1 English
Into Thin Air: Full Summary | Jon Krakauer — Transcript15:41
Into Thin Air: Full Summary | Jon Krakauer — Transcript
2,285 1 English
I Found the Loneliest Mob in Minecraft… — Transcript0:41
I Found the Loneliest Mob in Minecraft… — Transcript
143 1 English
Inside a 0DTE Options Strategy Targeting 50–100% in 24 Hours — Transcript41:12
Inside a 0DTE Options Strategy Targeting 50–100% in 24 Hours — Transcript
7,277 1 English
How Did Ancient Humans Get Any Privacy? — Transcript7:14
How Did Ancient Humans Get Any Privacy? — Transcript
1,258 1 English
Volatility Forecasting: From GARCH to Gradient Boosting and Language Models — Transcript6:53
Volatility Forecasting: From GARCH to Gradient Boosting and Language Models — Transcript
960 1 English
Varicocele Surgery (1 year update) — Transcript25:23
Varicocele Surgery (1 year update) — Transcript
4,442 1 English
The Truth About What COVID Did To Your Body — Transcript18:52
The Truth About What COVID Did To Your Body — Transcript
3,751 1 English
Why Ray William Johnson Can Get Away With Using AI — Transcript7:05
Why Ray William Johnson Can Get Away With Using AI — Transcript
1,402 1 English