Wednesday, August 27, 2014

Triathlon Handicap System™

Triathlon is a challenging sport that would benefit from a more effective way to measure results.

I really like triathlon. I like swimming, biking, and running, I like training, I like improving, learning, stumbling and getting up again, races, venues...almost everything about it. I love bringing people into the sport, and I equally love helping those already in it to improve and hopefully love it more.

What I don't like about triathlon is how, by its results system is currently viewed, most of the participants walk away from a race feeling they aren't very good at it. I'm talking of course about the age group awards system, meaning that the top 3 finishers in each 5 year age group are recognized post race, awarded with something, and gain bragging rights if so they choose (and most do). To 'podium' is generally considered doing well, and to 'not podium' means to not do well. Age groupers will sometimes bemoan their involvement in the sport by saying 'I'll never podium', as if to suggest that somehow their effort is recognizably less than others and they accept that.

I have a problem with that. Triathlon - at any level - is hard. It takes a lot of time, money, dedication and athleticism to finish one at any distance. I want to help grow the sport and grow the enjoyment of it for those involved. I look at other sports for a model; a hard sport to play that has millions of happy amateurs at all levels...golf. Golf's wonder is the handicapping system. A 30 year old who just started playing the game can play a round with a 70 year old who's been playing for years, and the field is absolutely leveled by the handicaps. Both walk off with the simple pronunciation of hitting over or under par, and a story of how that was arrived at for the day. Sure, there are 'scratch' golfers, but they *alone* compete for top prizes in competition. Those players don't enter the conversation unless an amateur player is seeking a coach, or offered in amazement as a pinnacle of achievement, a sight to behold in awe. So why can't this work for triathlon?

Golf counts your strokes over or under "par", which is a number of strokes assigned to a course by the sum of the hole lengths. Additionally, a course has a "slope", rating how difficult the course is beyond simple par. Reason being, the length of the hole may dictate the par, but reaching par may be more challenging given terrain, hazards, or other conditions. Comparing this to triathlons is fairly direct - certain factors, such as elevation, road conditions, water type (fresh vs salt) apply to 'course slope'. However, 'par' is a tough concept. You can take a golf hole and say it should take 4 strokes (a drive, an approach, and two putts) to reach the hole, but how do you estimate a triathlon?

If you compare to a professional, you're chasing a moving target, literally. What year did the pro attempt the race? What were the conditions that day? What was the competition? What level of training were they at? Do you average all pro performance over time? What if a pro has never raced the course you're on, or not enough pros have raced the course to provide a statistically significant average? I propose an alternative.

In running, there are generally accepted tables that compare the world record pace at a given distance adjusted for age and gender. Really, there are, take a look. This lovely table says take your best time at a distance and compare it to the record, then apply a discount for your age. For instance, a 30 year old female that runs a 10K in 45 minutes would sort out to  (1820 seconds * .9997) / (2700 seconds or 45 minutes) = 67.39%. So that puts you in the 67th percentile. If you were to run faster, you'd move to a higher percentile. Given that discount, one can also use the generally accepted formula to extrapolate expected run times given another run time, where T2 = T1 * (d2/d1)^1.06, where T1 and d1 represent T pace 1 and distance 1, and T2 represent pace 2 and distance 2, respectively.

Similarly in biking, the generally accepted model is to compare FTP, or functional threshold power - the amount of watts you can output for an hour. Given an FTP and several other inputs, such as gradient,  weight, drag, and rolling resistance, you can arrive at a velocity, which when applied to distance provides an expected time. Likewise in swimming, you can compare swim paces at a given distance and reasonably extrapolate a time from a set of inputs.

So how do you level the playing field? I've developed a program that takes inputs such as your FTP, 5K best, average 800yard swim pace, and other static information about the course and athlete, and produces an 'expected time'. There's par. If you finish within 1% of par, you raced at par. For every percent above or below, you pick up or lose a stroke (floored, meaning if you were 1.1% faster than par, that's 2 below par). I'm going to go about incorporating this into a web-based widget and API that anyone can use.

The same program can also of course take inputs from a theoretical athlete - for instance, the best overall athlete from USAT AG nationals - and be used on an annual basis to provide handicaps. For example, if the theoretical athlete finished the USAT National sprint course in X, and your model suggests a time of 1.1X, your handicap is 10 for the year. I can imagine a world where athletes talk excitedly after a race by saying, "I'm a 14 handicap and I raced 2 under par today - my best result yet!".

I also see this being something you could use in a fantasy setting - create teams of the same number of players (now regardless of gender or age!) and add the combined over\under par across a season of racing.

All of these ideas are intended to grow the sport and people's involvement in it.