Seasonality Forecast 4H

A seasonality indicator shows recurring patterns in data that occur at the same time each year, such as retail sales peaking during the holidays or demand for ice cream rising in the summer. These indicators are used in fields like business, economics, and finance to identify predictable, time-based fluctuations, allowing for better forecasting and strategic planning, like adjusting inventory or staffing levels. In trading, a seasonality indicator can show historical patterns, like an asset's tendency to rise or fall in a specific month, to provide additional context for decision-making.

Seasonality reasoning basically seasonality works most stably on the daily frame with the input parameter being trading day 254 or calendar day 365, ..
Use seasonal effects such as sell in May, buy Christmas season, or exploit factors such as sell on Friday, ... to track the price movement.

The lower the time frame, the more parameters need to be calculated and the more complicated. I have tried to code the version with 1 hour, 15 minutes and 4 hours time frames
On the statistical language R and Python, Pine script
Tradingview uses the exclusive and unique Pine language. There is a parameter limit, just need to change the number of forecast days or calculate shorter or only calculate the basic end time value, we seasonality still works
but the overall results are easily noisy and related to controlling the number of orders per week/month and risk management.

The 4-hour frame version works well because we exploit the seasonal factor according to the 4-hour trading session as a trading session
Every 4 hours we have an input value that corresponds to the Asian, European, and American trading sessions
4 hours - half a morning Asian session.4 hours - half an afternoon Asian session, 4 hours - half a morning European session, 4 hours - half an afternoon European session, similar to the US and repeat the cycle.

Input Parameter Declaration
Tradingview does not exist declaration form day_of_year = dayofyear(time) Pine Script v5:
Instead of using dayofyear, we manually calculate the number of days in a year from the time components.
// Extract year, month, day, hour
year_now = year(time)
month_now = month(time)
day_now = dayofmonth(time)
hour_now = hour(time)

// Precomputed cumulative days per month (non-leap year)
days_before_month = array.from(0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334)

// Calculate day-of-year
day_of_year = array.get(days_before_month, month_now - 1) + day_now


Input parameter customization window

Lookback period years default is 10, max - the number of historical bars we have, should only be 5 years, 10 years, 15 years, 20 years, 30 years.
Future project bar default is 180 bars - 1 month. We can adjust arbitrarily 6*24*254 - day/month/year
smoothingLength Smooth the data (1 = no smoothing)
offsetBars Move the forecast line left/right to check the past

How to use
Combine seasonality with Supply Demand, Footprint volume profile to find long-term trends or potential reversal points
day_of_year := day_of_year + ((is_leap and month_now > 2) ? 1 : 0)

// Compute bin index
binIndex = (day_of_year * sessionsPerDay) + math.floor(hour_now / 4)
binIndex := binIndex % binsPerYear // Keep within array bounds

The above is the manual code to replace day of year