“What is the best time to trade?” is probably the question most often answered with guesswork in all of trading. Every guru has his magic table, almost always copied from another guru. We prefer to ask the data: we built a grid of 1,296 combinations of hours, days and expiry, and judged each one by the harshest possible criterion — the FLOOR of 3 validation windows (that is, each configuration is judged by its worst result, not its best, to eliminate lucky selection). What survived that funnel is what you will read here: hours that confirmed an edge in both windows, a “mirage hour” that would have fooled any naive backtest, the weekday that turned into a trap, a popular myth debunked — and the final honest configuration, with a number in place of a promise.
This study used real M1 data downloaded from Deriv’s official API — real forex 24/5, with a free demo account.
Open a Deriv account →How the study was done (full methodology)
Data: 45 days of REAL M1 candles from Deriv (official API), 8 forex pairs.
3-layer validation: 70% training / 30% test + FRESH WINDOW — a block of data (Jun 12 → Jul 2) that no study saw during development.
Metric: accuracy on the direction of the next candle.
Break-even: with an 87% payout, you need 53.5% accuracy just to break even. Win rate ≠ profit.
Golden rule: a number that only shows up in one window is luck or overfit. Only what survives in all of them counts.
A methodological detail that makes all the difference: with 1,296 combinations, some will always shine by pure chance — that is basic statistics. That is why every combination was ranked by the floor of the 3 windows: the worst performance among them. An hour only makes the final list if it worked in all windows, including the fresh one. It is the antidote to the lucky selection that contaminates 99% of the “time tables” on the internet.
The strong hours: a short morning and a long evening (BRT)
Two time bands confirmed an edge in both validation windows: 9-10h and 18-23h, Brasília time (BRT, UTC-3). In UTC, that is 9-10h BRT = 12-13h UTC and 18-23h BRT = 21h-2h UTC. The highlights, with the result in both windows (test → fresh window):
| Hour (BRT) | Test → Fresh window | Verdict |
|---|---|---|
| 18h | 67% → 77% | Strong in both windows |
| 23h | 68% → 68% | Strong and stable |
| 21h | confirmed in both windows | Strong |
| 9-10h | 62% → 58-65% | Strong |
| 16h | 65% → 40% | UNSTABLE — mirage hour |
| 7h, 0h, 4-5h | worst of the grid | Avoid |
The whole afternoon, from 11h to 17h BRT, dilutes the edge — not a disaster, but it drags the average down. And the case of 16h deserves a frame: 65% in the test, 40% in the fresh window. A single-window backtest would have put 16h on the “golden hours” list — and that mirage hour would have devoured the profit of the good hours. It is exactly the kind of trap that floor-based validation exists to catch. The worst hours of the grid were 7h, 0h and the 4-5h early morning — and it is worth remembering that the middle of the night is precisely when many signal rooms push OTC, a problem we dissect in OTC is a random walk.
Days of the week: Tuesday and Thursday lead — and Friday became a trap
Looking at the weekday breakdown, Tuesday and Thursday were the best days, in the range of ~57% accuracy. The warning came from Friday: performance deteriorated in the fresh window, dropping to 48.8% — below even a coin flip, and well below the 53.5% break-even. The data’s recommendation is direct: avoid Friday. Monday through Thursday make up the trading window; Friday stays out.
Myth debunked: the “safe hours” filter
There is an almost universal belief in binaries: “block the news hours, trade only in the safe hours”. We tested the filter. Result: for mean-reversion strategies, the safe-hours filter was neutral to negative — and it also cut 22% of the trades. In other words: you give up a fifth of your volume (and volume is what turns edge into results) to gain nothing in return, or even to get worse.
Why the myth fails: for mean reversion, volatility is not the enemy — it is fuel. Stretched moves are exactly what the strategy exploits when the price gives back. The approach the data supports is the opposite: instead of a negative list of “dangerous” hours, use a POSITIVE list of proven good hours — and trade only in them.
Expiry: 2 candles beat 1
Another counterintuitive finding: the same signal, measured 2 minutes ahead, hits more often than measured 1 minute ahead. In the numbers of the two windows: 2-candle expiry scored 60.8% / 58.7%, against 57.5% / 57.4% for 1-candle expiry. The explanation fits the strategy’s mechanism: the post-exhaustion reversal is not instantaneous — it takes 2-3 minutes to materialize. With a 1-candle expiry, you are often right on direction and wrong on the stopwatch.
The final honest configuration
Hours: 9:00-10:59 + 18:00-23:59 (BRT)
Days: Monday through Thursday (Friday out)
Expiry: 2 candles
Result: a floor of 59.2% on the validation grid — a realistic expectation of 57-60% in live operation.
Notice the tone: we do not promise 70%, we promise the floor — 59.2% was the worst result of this configuration across the windows, and the realistic operating expectation sits between 57% and 60%. Above the 53.5% break-even, that is real edge; we explain why this range is the honest ceiling of M1 in the study 70% win rate is a myth. And none of this survives without risk management: a good window is not a license to size up, and martingale remains the fastest way to turn 59% accuracy into a zeroed account.
If you want to automate this trading window instead of watching the clock, you can configure time filters in a bot — see the no-code Deriv Bot guide or our open-source bot for Quotex and IQ Option.
FAQ — Hours and days to trade
What is the best time to trade forex M1? In our data: 9:00-10:59 and 18:00-23:59 Brasília time (BRT), with highlights at 18h (67% → 77%) and 23h (68% → 68%) in both validation windows.
What is the worst day of the week? Friday, which dropped to 48.8% in the fresh window — below break-even. Tuesday and Thursday were the best days (~57%).
Should I avoid news hours? For mean reversion, the “safe hours” filter was neutral to negative and cut 22% of the trades. Better to use a positive list of good hours.
1-candle or 2-candle expiry? 2 candles: 60.8/58.7 against 57.5/57.4 in the two windows. The post-exhaustion reversal takes 2-3 minutes to happen.
Do these hours apply to OTC? No. OTC has no sessions and no real liquidity — our tests showed ~50% on everything, at any hour.
Test the 9-10h + 18-23h (BRT) configuration on a demo account before any real cent.
Create a Deriv demo account →Read next
→ I Tested 20 Strategies on 45 Days of Real Data: 70% Win Rate Is a Myth
→ OTC Is a Random Walk: Proof With Real Data (and Where Real Edge Exists)
→ Double Reversal: Band + Exhaustion Strategy Rules
→ Bot Results on Demo Accounts at 3 Brokers (July 2026)
→ IQOptionAPI: check_win Freezing and OP_ASSETS Fix
→ Deriv Bot: How to Create Trading Bots Without Coding
→ Risk Management for Trading Bots
→ Does Martingale Work in Trading? The Math Without the Hype
About the author — Dan Machado
Founder of IA Trader Pro. Builds open-source trading bots and publishes studies with real data — no account screenshots, no promises of riches. Everything in this article came from code and data that anyone can reproduce.
Disclaimer: binary options and derivatives are extremely high-risk products and most retail traders lose money. This content is strictly educational and does not constitute investment advice, an offer or financial counseling. The results presented come from a statistical study of historical data and do not guarantee future results. This article contains affiliate links. Always test on a demo account before risking real money, and never trade with amounts you cannot afford to lose.
