News Blackout Windows for Trading Bots: Handling CPI and FOMC Without Babysitting Every Trade

CPI drops at 7:30 AM Central. For about ninety seconds, the ES order book turns into a rumor. Spreads widen, liquidity steps back, and price covers more ground in a minute and a half than it did in the previous two hours. If your bot has an entry condition that happens to line up with that move, it takes the trade — because that is exactly what you built it to do.

The bot is not the problem. The calendar is. Automated futures traders who run into trouble on release days almost never have a broken strategy; they have a strategy running at a time of day it was never tested for. The fix is a news blackout window, and once it is set, you never have to sit at the screen on FOMC day again.

What actually happens to futures in the 60 seconds around a release

Three things change at once, and all three work against an automated entry.

The spread widens. ES and NQ normally trade one tick wide with real size at the top of book. Into a scheduled release, market makers pull size. That one-tick spread can go to several ticks for a few seconds. Your bot does not know that. It sees its signal and sends the order.

Stops get filled far from where you placed them. A stop is not a guarantee of price, only a guarantee of execution. In a thin book, a 10-point NQ stop can fill 25 points away. On a single NQ contract that is the difference between a $200 loss and a $500 loss — and if you are running micros or minis at eval size, the math matters. We broke that contract math down in NQ vs ES vs MES.

Your backtest probably lied to you about this. Historical fills are modeled, not real. Most backtest engines assume you got filled at the signal price, or one tick worse. Neither is true in the first minute after CPI. This is one of the specific ways 4.5 years of backtest data can look clean and still hand you an outlier loss in live trading.

The three releases that actually matter

You do not need to track fifty economic events. For an index futures trader, three move the needle, and all times below are Central:

• CPI — 7:30 AM CT, monthly. Lands a full hour before the equity index open, which means the opening range you are about to trade is being built on top of a news move.
• Nonfarm payrolls — 7:30 AM CT, first Friday of the month. Same profile as CPI, often a wider reaction.
• FOMC — statement at 1:00 PM CT, press conference at 1:30 PM CT, eight times a year. This one is different: it hits mid-afternoon, when your bot may already be in a position, and the 1:30 presser frequently reverses the 1:00 move.

The 7:30 releases and the 1:00 FOMC create two very different problems. A pre-open release distorts the setup your bot is about to build. An afternoon release distorts a trade your bot is already holding.

Blackout windows that actually work

A blackout window is two numbers: how long before the release you stop taking new entries, and how long after you start again. Start here and adjust from your own data:

For the 7:30 AM CT releases: no new entries between 7:25 and 7:45. If you run an opening range breakout, consider skipping the day entirely or waiting for the second range. The whole premise of the automated ORB is that the opening range represents real, contested value. On CPI morning, the range is often just the tail end of a news spike — same pattern, different meaning.

For FOMC: no new entries after 12:50 PM CT, and flatten anything open before 1:00. Do not plan to "watch it and decide." That is a decision you will make badly, in real time, with money on the line.

Widen these windows if you trade NQ instead of ES, and widen them again if you are running a mean-reversion or snapback strategy. Fading a move is the wrong trade to be in when the move has a headline behind it.

How to set it up so you never touch it

In NinjaTrader 8, the cleanest approach is a trading hours template with your blackout minutes carved out, applied to the strategy rather than the chart. Add a flatten time so nothing carries into the 1:00 CT statement. Then, on FOMC days, the only manual step is choosing whether the strategy runs at all — and that decision gets made the night before, not at 12:58.

Log what you did either way. A "sat out" day belongs in your journal with the same weight as a traded day, which is why session context is one of the five journal fields worth tracking. Three months from now, the only way you will know whether your blackout window is too wide, too narrow, or exactly right is if you wrote down what happened on the days you skipped.

What not to do: hand-managing the release

The tempting move is to leave the bot on and "just take over if it gets crazy." That is not risk management, that is the override loop wearing a helmet. Manually pulling a bot's trade on CPI morning is the same behavior that ends evaluations on ordinary Tuesdays — you are substituting a live judgment call for a tested rule, at the exact moment your judgment is worst.

The reason automation works for busy traders is that the rules were decided in advance, on purpose. That is the entire argument for letting bots trade the open while you work. A news blackout window is not an exception to that logic — it is the same logic, applied to the calendar. And if the release day does go against you, the cap you already set does its job: your daily loss limit ends the session, not your willpower.

Check your firm's rulebook before you assume

Prop firms differ on news trading, and the rules change. Some restrict holding positions through high-impact releases, some restrict it only during the evaluation phase, and some care mainly about what you did around the release when a payout request lands on their desk. Apex, Lucid, BluSky and TakeProfit each publish their own version — read yours before you build the window, not after. A blackout that is stricter than your firm requires costs you a few setups a month. A blackout that is looser than your firm requires can cost you the account.

Put it in the plan, not in your head

Two dates a month. Eight afternoons a year. That is the entire scope of the problem, and it is completely solvable with a calendar and a session filter. Decide your windows once, build them into the strategy, and let the release come and go while you are doing something else — which was the point of automating in the first place.

Want the bots, the portfolio tools and the setup walkthrough in one place? Start at pushbuttontrading.co.