Some very good points here from Geoffers and Jim. I would be very interested to hear more Jim, especially with reference to modeling of different types and specific data. I have long tried to reinforce the need for a better understanding, instead of just flying numbers, and whilst it makes perfect sense and is logical that the amount of pitch down would decrease with wind speed, one has to be careful instructing and advising without good test data and or training to support. For most of us the RFM is the only thing we have. I personally feel very uncomfortable using 20 degrees nose down in an S92 in anything other than calm conditions, especially at night or in poor weather. I believe that some of the RFM information serves to confuse rather than clarify in some areas, and where the RFM procedure states simply "rotate to a maximum of 20 degrees nose down" this leaves some discussion as to whether this is referenced from a hover datum - which in an S92 is often around 8-10 degrees up - or referenced from the AI.
If referenced from the AI we are talking about a rotation of nearly 30 degrees, which I believe to be a potentially dangerous route to over controlling and could lead to a dangerous further decay in Nr in many cases, with a greatly increased initial rate of descent. I believe Jim hit the nail not he head with the wording rotation through 20 degrees, not to 20 degrees.