The PN software is written in Fortran. A near dead language 25 years ago... (remember the y2K death watch) it has a limit to the number of variables it can handle.

All data is pre screened for being an outlier... IE the computed rating is 10 % faster or slower then the current rating... MIND YOU... the data is computed in the wind speed reported for the race.

The number of ratings used to calculate the rating is 100 data points in EACH wind speed. (If you did not roll in new data and kick out old data... the average would never change.)

Beaufort 2 data is tabulated separately from Beaufort 3 data and if memory serves for each wind speed but so far... the data don't reflect the performance difference and so 2 and 3 are lumped together. Again... it's not just A class and F18's to consider....the Hobie 16, 17 and 18 performance are still not going to vary much over the B 2-3 range.

This analysis was done a couple of years ago... So Darline may have more data with light boats, spin boats and secondary yardstick boats to evaluate. It may have changed since then.... but since the Area L results only contribute 6 or 7 data points....AND the race did not have a well sailed secondary yardstick boat.... eg a hobie 16 Its tough!

The DPN rating is weighted to the B4 rating and is some formula of the 4 wind ratings.... (new boats B4 = DPN)

So... if you don't include an accurate wind speed for the race then your data becomes difficult to use.