Seasonality & Time-of-Day. When the market actually shows up.
Crypto trades continuously, but it does not trade evenly. Liquidity and volatility follow the working hours of Asia, Europe and the United States, drop across weekends, and cluster around scheduled events.
These patterns are far more reliable in liquidity than in direction, which is the distinction that determines how they should be used.
Time patterns are strong for liquidity and weak for direction
That the US session is the most liquid part of the crypto day is a structural fact driven by when institutions are staffed, and it persists across years. That the market tends to rise on a particular weekday is almost always noise found by slicing data enough ways. Use time to decide when to trade and at what size; be extremely sceptical of using it to decide which way.
Seasonality & Time-of-Day in a strategy
Spreads tighten and depth improves during the European and US overlap, so execution costs less.
Lower weekend liquidity means larger slippage and outsized reaction to modest orders.
Price often becomes erratic around the eight-hourly funding timestamps.
Macro releases and option expiries land at fixed times and reliably raise volatility.
Restrict a breakout strategy to liquid hours
Same strategy, fewer trades, better fills. Time is used here to avoid bad execution rather than to predict direction.
- IFThe current time falls within the European and US session overlap
- ANDIt is not a Saturday or Sunday
- ANDThe current hour is not within 15 minutes of a funding settlement
- ANDVolume in the last hour is at or above its median for that hour of the day
Every clause targets execution quality rather than direction. Breakouts need follow-through, and follow-through needs participants, which is why the same signal has a much better hit rate during liquid hours. Comparing volume to the median for that specific hour rather than a flat average matters, because a quiet US hour is still busier than a busy Asian one. Logging skipped signals keeps the filter honest: if it is rejecting most of the strategy's trades, the strategy may simply be built on a timeframe the filter cannot accommodate.
Group candles by hour and weekday and the liquidity pattern is immediately visible. No external data needed.
Venue-published volume by period confirms when their books are actually deep.
Macro release schedules and options expiry dates are published well in advance.
Seasonality is the easiest thing in trading to overfit
Slice history by hour, weekday, month and moon phase and you will always find a pattern that looks significant. Liquidity patterns have a structural cause and survive out-of-sample; directional calendar effects almost never do. If you cannot explain why a time pattern should exist, assume it is noise.
Trade the hours that have actually paid.
No screens to babysit: define the conditions and the bot does the rest.
- Session liquidity
- Weekend thinness
- Funding settlement