Handicap racing is now the bedrock of the sailing scene, but traditional handicapping techniques have failed to keep up with the wide range of boat designs we now see on the race course, covering classes as diverse as Toppers and Foiling Moths, and the corresponding wide range of speeds. With the vast amount of data generated by GPS tracking is is now possible to have a more dynamic form of handicapping that adapts itself to the conditions experienced in each race to become more context sensitive.
Traditional techniques will allocate each class an average handicap that has to cover all conditions, but the problem is that there is no such thing as an average race. Evidence from the big handicap races suggest certain weather conditions favour certain classes, for example, in light winds the faster boats struggle to sail to their handicap as they cannot make use of all their performance, a bit like a F1 car driving around a busy city at rush hour.
Background
The Portsmouth Yardstick Scheme (PYS) for handicapping dinghy racing has been around for over 70 years and has served sailing well, allowing boats of varying performance to race against each other on equal terms.
The key principles of PYS have remained consistent over the years, using observed data in the form of elapsed times for each boat, in a particular race, to be compared and a suggested handicap calculated. Clubs would submit an annual return to the RYA with their suggested handicaps across the year for the classes they race. The official RYA Portsmouth Yardstick list would be calculated using the club numbers, to effectively give a blended handicap for each class.
2008 saw one of the most significant changes with the introduction of an online system developed by SailRacer that allowed clubs to directly submit their timed results and suggested handicaps were calculated automatically. For the first time there was a centralised database of timed results across thousands of races with their associated suggested handicaps.
With the raw data available for the first time, one of the observations was that the typical handicap race was eight boats, sometimes with eight different classes, which makes comparison less accurate, other than for generating a personal handicap as the sample size is too small. Fortunately groups such as the Great Lakes took a more focussed view on handicapping using the larger handicap events to give a bigger and more representative sample.
Another issue that the data highlighted was the vast range of suggested handicaps for an individual class, referenced to sail number. In broad terms a newer boat is faster than an older one, obvious, but when you drill down and consider why, many one design classes have seen some radical changes in construction and design. Take the Solo, dating back to the 50s and conceived in an era when many boats were home built and measurement tolerances wide, that has now evolved to a highly refined dinghy, with stiff FRP epoxy construction, optimum hull shapes and highly efficient modern rigs. From a handicap perspective, a home built Solo from the 50s is treated the same as a brand new highly optimised professionally built Solo. The data would suggest there is at a least a 6% differential between these two extremes.
An analysis of the data highlighted a massive range in suggested handicaps, particularly for the newer high performance classes such as the 49er and Foiling Moths. In recent years there has been major progress in design, manufacture and materials for dinghies, with some classes now foiling and many benefitting from apparent wind, where speeds can easily exceed wind speed. When boats are sailing in displacement mode, it is easy to calculate Hull Speed (HS) using the well proven formula :
HS = 1.34 x √LWL
For a typical dinghy this equate to circa 5 kts. When all boats are sailing in displacement mode, comparisons becomes reliable and easy. But this ideal scenario rarely happens, some classes will start planing on the faster off wind legs even in relatively light conditions. As the wind builds more classes will be planing on some part of the course, with some designs starting to benefit from apparent wind sailing. Generally speaking as the wind increases so do speeds, until you get to a situation when conditions are too severe and survival becomes the order of the day.
Broadly speaking there are seven speed modes : drifter, displacement, marginal planing, planing, apparent wind, foiling and survival. For some classes speeds can verry massively, from under 5 kts to 30 + kts, BUT the handicap remains the same, therefore as so often seen in large handicap races if the conditions don’t suit a particular class they will struggle to sail to their handicap. This was clearly illustrated at the 2017 Draycote Dash, which is light conditions saw the faster boats struggle to get into the top ten, despite being sailed by some top sailors, for example the Fireball of Ian Dobson and Richard Wagstaff both class Champions and regular winners over the SailJuice Winter Series, who for the average lap races on corrected time finished in positions 12, 21, 21 and 34 as they struggled to sail to the handicap of 948. The problem is this handicap had been calculated on a average basis across a variety of wind conditions, with the Fireball coming to the fore in strong winds. So, unless you have average conditions the handicaps will struggle.
With the advent of GPS tracking technology, providing a wealth of data across every leg, it is possible to produce a far more dynamic handicapping system that is context sensitive to the conditions experienced in any measured race. Using the methodology outlined below, Ian Dobson would see a score line of 2,3,5 and 8 (total 18 compared to 88).
Rescored Races Methodology
With dynamic handicapping the aim is to calculate the optimum handicap for each class to suit the conditions of each race. So the first task is to calculate the Course Speed Profile (CSP) for each class to the conditions experienced during the race being scored. The CSP gives a measure of how well each class is performing to its optimum performance, you could consider this to be like gears, with six being top and the class achieving its best performance.
The CSP is calculated using the GPS leg data which gives a wealth of analysis : leg length, leg angle, leg times, speeds and VMGs.
For many races, different classes are likely to be in different gears, for example in marginal planing conditions, the lighter boats will be planing some of the time, but many classes will be struggling to get out of displacement mode (1st gear).
Once the CSP has been calculated for each class, we can cross reference the historic data from previous races to find similar conditions and calculate the associated handicap, ie one that is context sensitive to the conditions experienced.
Vast amount of data
SailRacer has a huge amount of data available to allow cross referencing and improve the accuracy of derived handicaps, including 1.25 million race results and GPS data covering over 1.5 million legs, across 110 different classes.
Self learning
Traditional handicapping has inherent latency as part of its methodology, waiting to collect the data then analyse to produce the average handicaps. This process typically works on an annual cycle. Using live GPS data, dynamic handicapping can utilise the latest data in conjunction with all the historic analysis, even to the extent that the current race being scored contributes its data to the overall dataset.
Checks and balances
Using statistical techniques to calculate factors such as standard deviation and confidence interval, outlying data is discarded and calculated handicaps fit an appropriate range to ensure a high level of accuracy.
New Designs
With traditional handicapping methodology, it can take a significant amount of time for new designs to amass sufficient data to reliably calculate a handicap. Dynamic GPS handicapping allows a speedier collection of data and new designs seeing a realistic handicap applied more quickly, with the potential to compare the CSPs of similar established classes.