WITH RECURSIVE years(byyear) AS (
  SELECT 1991 FROM rdb$database UNION ALL
  SELECT byyear+1 FROM years WHERE byyear < 2020
),
s AS (
  SELECT EXTRACT(YEAR FROM order_date) AS byyear,
         SUM(total_value) AS total_value
  FROM sales GROUP BY 1
),
regr AS (
  SELECT REGR_INTERCEPT(total_value, byyear) as intercept,
         REGR_SLOPE(total_value, byyear) as slope
FROM s)
SELECT years.byyear AS byyear,
       intercept + (slope * years.byyear) AS total_value
FROM years CROSS JOIN regr

BYYEAR     TOTAL_VALUE
----------------------------
 1991        118377,35
 1992        414557,62
 1993        710737,89
 1994       1006918,16
 1995       1303098,43
 1996       1599278,69
 1997       1895458,96
 1998       2191639,23
 1999       2487819,50
 2000       2783999,77
...