Preparing for a fall marathon, a runner posts an 18-minute 5K PR in a tuneup race. Excited, he punches the time into three different race-equivalency calculators to find his projected time for the marathon. The first calculator, created by coach Craig Godwin of FasterRunning.com, spits out a time of 2:56:05. Greg McMillan’s namesake calculator predicts 2:55:23. He tries Jack Daniels’ Running Calculator and it reads 2:52:45.
Which one does the runner believe? The fastest, of course. But the pace for a 2:52 (6:35/mile) is 8 seconds faster than that for a 2:56 (6:43/mile), which could result in disaster if the runner goes out too fast.
The truth is that race-equivalency calculators estimate a best-guess, ballpark time. Why do calculators’ predictions vary? “Math isn’t the way our body works,” McMillan says. “Our body works on threshold, so you have to understand the physiology of the body.” A race calculator must balance abstract, predictable mathematical formulas with messy, real-world data. As to what percentage of each is required to generate the most accurate algorithm, no one has yet perfected the recipe, hence the three different marathon forecasts.
CRAIG GODWIN’S CALCULATIONS
An engineer based in Eugene, Oregon, and a distance runner since high school in the 1970s, Craig Godwin has devised two different race-equivalency calculators. The first is for coaching his own runners, and the second allows members of the Oregon Track Club masters group to compare PRs across age and gender lines. To develop his algorithm, Godwin has compiled the top 20 world-class performances at various distances. He bases the formula off of elites because the pros adhere to steady training schedules, pace themselves better than most and don’t experience much time fluctuation between race distances. But this method still isn't an exact science.
Godwin’s calculator assumes that a runner will perform at the same percentage off of a world-class performance, no matter what event she is racing. In reality, Godwin says, “It’s not a straight line.”
The elites can unintentionally alter the formula. “In recent years, we’ve seen some very fast marathon times, whereas the mile record hasn’t changed in a number of years," Godwin says. “Should the table be updated for the marathon performances, or would that hurt the normal runner?” Major marathons offer large purses, whereas 5Ks on the track generally don’t, so what happens if the fastest athletes bypass the track in favor of a payday on the roads? Outliers also impact the data, due to performance-enhancing drugs or a breakthrough that other elites can't replicate.
Elites also tend to train thoroughly for every distance raced, while many citizen runners have to skimp in areas of preparation, particularly for longer races. "While the calculators are derived primarily from performances of elite runners, you do not need to be an elite runner or train at an elite level for the calculators to be accurate," Godwin says. "However, the calculators assume you are equally talented and equally trained for all distances."
GREG MCMILLAN’S METHOD
McMillan incorporated research from his exercise physiology master’s thesis, his own running and his experience coaching recreational runners—from beginners to those trying to qualify for Boston—to create his race-equivalency calculator in the 1990s. “When you work with such a range of runners, you need a tool to prescribe training to each of those athletes,” McMillan says. Using his athletes as guinea pigs, he was able to plot data points to predict race performances. To date, more than 20 million people have used his calculator. To help refine the algorithm, McMillan monitors a sample of the 1 million regular users.
McMillan stresses that the key is the varied population. “If it’s just [based on] elite runners, that’s not always the best for the rest of us,” he says. Still, users might have to handicap the estimates, depending on their fast- and slow-twitch fibers, McMillan says. “If you are a speedster, it will probably be harder for you to hit the predicted time of the longer races,” he says. “The reverse can be true as well.”
JACK DANIELS’S VERSION
Jack Daniels has drawn on his more than 50 years of coaching experience to invent his VDOT system. It is based on VO2 max as well as the gray area measurements of an athlete, such as proper biomechanics and mental toughness. “The data we used involved testing hundreds of serious runners,” Daniels says of his proprietary regression formula.
Daniels has added an advanced feature that takes wind, temperature and altitude into account. To establish the equivalency standard for wind, he tested eight competitive runners who covered flat ground alongside an airport runway while he measured the wind velocity. He had them go north at 6-minute pace, then turn around and go south at the same tempo. They repeated the run at 5-minute-mile pace. “The subjects repeated these same tests on three days with different wind velocities,” Daniels says. “We measured oxygen consumption (to get running economy), heart rates and blood-lactate values in every run.”
For the temperature adjustments, Daniels had 32 subjects race a 25K road race on three consecutive weekends, with different air temperatures. “Each subject’s body-weight loss and fluid intake were measured exactly during each run,” he says. Daniels took data from his nearly 50 years of testing athletes at altitude to create adjustments based on feet above sea level.
Daniels’s calculator reveals the noticeable effect these factors can have on a runner’s finishing time. For example, an 18-minute 5K run into a 10 mph head wind equates to 17:05, while an 18-minute 5K lifted by a 10 mph tail wind is worth only a 19:38.
Making Your Prediction More Accurate
To compensate for the variability in runners, McMillan follows what he calls the “hybrid calculator,” a method of tweaking results for your strengths. A runner with natural speed doing shorter races or workouts would use a recent 5K race time as his training and racing baseline, but for longer workouts and races, this runner should use a slightly slower 5K time, to give him a slower set of bases for a stamina run. A farther-is-better runner would do the opposite—use a slower 5K (or add 30 seconds to a recent time) when using it as a baseline for interval work, but use his PR for any workout longer than 3.1 miles.
While these three race-prediction calculators using an 18-minute 5K varied by 3 minutes over the 26.2-mile distance, they were much closer determining a 10K result: between 37:23 and 37:35. “Predictions of race performances that are closer together are stronger than those farther apart,” McMillan says. For the most accurate forecast, use a distance that is closest to the upcoming race—rather than a new PR at a farther-removed distance--as long as it is a solid effort unaffected by weather or hills.
“A race-prediction calculator gives a starting point, but you should use what you know about yourself as a runner and your current training to make manual adjustments to what the calculator is telling you,” Godwin says. “I recommend that most runners allow a plus or minus 5 seconds per mile pace range around what the calculator predicts.”
