The same system, the same rules, the same strategy. But completely different results at different times. Have you ever thought: Could the reason for the system's failure be your code, not the time?

Although financial markets seem to be continuously open, they do not offer the same liquidity, player intensity, and volatility at every hour. Therefore, if you have not analyzed the profit-loss distribution of the system on an hourly basis, what you think is success may be just a coincidence.

Example:

Let's say a system has a 60% win rate based on the backtest result from a total of 1000 trades. Great. But when you break these 1000 trades down by hours, perhaps only the time period between the London open and just before New York produces positive expectancy. During other hours, the system either struggles or loses money.

📊 Therefore, time-based filtering reveals the true efficiency of the system. Not only the transaction result but also "when it was taken" is part of the system.

Technical Recommendation: Log each transaction data along with a timestamp.

Group all transactions by hour intervals (Turkey time):

🔹 Asia: 03:00–10:00
🔹 London: 10:00–16:30
🔹 New York Opening: 16:30–23:00
🔹 New York Close: 23:00–03:00

Extract the following metrics for each interval:

Win rate
Avg R:R
Expectancy
Drawdown profile
Compare these.

In an analysis conducted in 2023, the average R:R ratio for trades in the spot BTC/USDT pair during Asian hours is 1:1.2, while during the London–NY overlap, this ratio increases to 1:1.9.

So first and foremost, it is not "what" the system does, but "when" it does it that is important.