According to Large (1954) cereals develop as follow Feekes Growth Stages

Stage            
T I L L E R I N G
   1            One shoot (number of leaves can be added) = "brairding"
   2            Beginning of tillering
   3            Tillers formed, leaves often twisted spirally. In some varieties of winter wheats,
                   plants may be "creeping" or prostrate
   4            Beginning of the erection of the pseudo-stem, leaf sheaths beginning to lengthen
   5            Pseudo-stem (formed by sheaths of leaves) strongly erected

S T E M    E X T E N S I O N
   6            First node of stem visible at base of shoot
   7            Second node of stem formed, next-to-last leaf just visible
   8            Last leaf visible, but still rolled up, ear beginning to swell
   9            Ligule of last leaf just visible
  10          Sheath of last leaf completely grown out, ear swollen but not yet visible

H E A D I N G
10.1          First ears just visible (awns just showing in barley, ear escaping through split of sheath
                  in wheat or oats)
10.2          Quarter of heading process completed
10.3          Half of heading process completed
10.4          Three-quarters of heading process completed
10.5          All ears out of sheath

F L O W E R I N G  (WHEAT)
10.5.1       Beginning of flowering (wheat)
10.5.2       Flowering complete to top of ear
10.5.3       Flowering over at base of ear
10.5.4       Flowering over, kernel watery ripe

R I P E N I N G
11.1          Milky ripe
11.2          Mealy ripe, contents of kernel soft but dry
11.3          Kernel hard (difficult to divide by thumb-nail)
11.4          Ripe for cutting. Straw dead

Reference

Large, E.C. 1954. Growth stages in cereals. Plant Pathol. 3:128-129.

•Exp. 222, r2= 1.305+0.0204(GDD>0), when GDD<106; plateau for r2 at 0.87 when GDD>=106

• Exp. 502, r2= 0.046+0.0088(GDD>0), when GDD< 87; plateau for r2 at 0.81 when GDD>= 87

Coefficients of determination (r2) were recorded for linear regression equations from NDVI data collected over the entire growing season (Figure x).  This included a range in GDD>0 of 50 to 126.  These coefficients were then paired with the number of days from planting to sensing where GDD>0.  Once the r2 values were plotted against GDD>0, the objective was to determine the point at which r2 values were maximized and to simultaneously determine a point where changes further into the season were capped. 

A linear plateau model was then used to determine the point at which r2 values no longer increased or no longer showed a benefit in terms of predicting yield.  This question can go the other direction in that it helped to determine how early yield could be accurately predicted and not sacrifice precision. 

Using the linear plateau model, the earliest number of days from planting to sensing where GDD>0, and that would be required to accurately predict wheat grain yields was 87.  

It should be noted that the data also suggests that yield should be predicted prior to reaching 112 days (GDD>0) where after that time yield prediction was diminished.