Race Analysis Insights

There has been some question of what is considered a race for analysis purposes and why a minimum number of boats is “required” to use the data.  I offer these insights:

  1. The primary purpose of the race analysis is to determine the relative difference of speed potential between boats.  In our case it is an average speed potential taken over various wind and sea conditions on a closed course.  If we determine that Boat A is faster than Boat B it doesn’t mean we believe that the difference is correct under all conditions but rather represents the velocity difference under the average conditions that the underlying races were held.   The more boats of a single type and the more races, the more reliable the speed potential prediction becomes.
  2. The computer program (algorithm) running the analysis has no idea of who is sailing the boat, how the boats were equipped, how well the boats were individually sailed, or any other attribute particular to an individual race such as race course or environmental conditions. It can however mathematically compare the corrected finish times of boats using their handicaps and elapsed time around the course and decide relative differences (a velocity difference).  

A common problem however, is that it is quite likely that one of the boats in a race may have gotten a wind shift, hit a mark, started late, or a whole variety of other events common to racing sailboats which contributed to the perceived velocity difference between the boats.  The perceived “hull speed” difference with only two boats in a race is therefore extremely questionable and probably unusable unless there is other knowledge, unknown to the number crunching computer program, that might help validate the difference.

Looking at thousands of individual races it was found that the absolute minimum number of boats in an individual race that may produce a usable prediction is three and that’s only when other criteria are met that shows that the boats under analysis show a level of performance that is consistent with their handicap.  For example: if a boat sails so far from their predicted performance that it is obvious that other factors besides the handicap caused the perceived velocity difference, their data (race results for that race), are not usable for handicapping purposes.  Since the computer algorithm knows nothing but numbers, that race result may likely automatically be excluded.  My opinion is that “while there is a limit to how fast a boat can be sailed there’s almost no limit to how slow a boat can be sailed” (and that comes from 45 years of personal observation).   From a handicap adjustment perspective we need results that are near the maximum performance limit not the minimum.

 When analyzing a year’s data from all the races around the lake combined (thousands of individual results), the program must rely on certain limits to automatically cull the bad data out. Hopefully the data is still available for review within the individual club which can be useful to that club.

  1. The mixing of Spinnaker and Non-spinnaker boats is not recommended by PHRF-LO for handicapping purposes because PHRF-LO, while believing the non-spinnaker results are valid amongst themselves, do not necessarily correlate well to the spinnaker results.  I think this is certainly a debatable subject and PHRF-LO has made some great progress in adjusting velocity differences due to sail configurations over the years.   I am sure there are smaller clubs who race spinnaker and non-spinnaker boats together with success.  Perhaps in the future PHRF-LO’s position may change but presently they are keeping spinnaker and non-spinnaker results separate.

  2. Someone questioned the use of results of boats that were single handed within the analysis.  While I’m sure there may be some boat types that can be effectively sailed single handedly, in the majority of cases, more than one person is required to sail the types of boats for which PHRF-LO provides handicaps. Boats that are sailed single handed are generally at a severe disadvantage to boats sailed with an optimum crew number in almost all cases.  The results of any boat sailed with this “disadvantage” are unreliable and unusable with our present handicapping analysis techniques.