Unfair Advantages: Circular Current in the 2016 Rio Olympic Pool
How circulation in the Olympic pool impacted the 50m Freestyle
Over the last ten years, there have been consistent and reoccurring rumors of a water current in competition pools at major international level swim competitions. In 2015, researchers at Indiana University found evidence of a lane bias at the 2013 FINA World Swimming Championships (Cornett, Brammer, & Stager, 2015). They found that athletes competing in lanes 1–4 of the pool experienced a uniform, advantageous change in time when moving in a given direction as compared to a detrimental time change for athletes competing in lanes 5–8. These findings support the rumors of a circular current, one which could have had a dramatic impact on not only athlete performance in 2013, but on future performances in similar pools.
The main concern is the impact that the current would have on the 50 freestyle event. For every other event, the swimmer will race more than one lap (length down the pool), so theoretically, the impact of the current should be negated. If a swimmer competing in lane 8 were to swim 1% faster as the result of an advantageous current, it could impact their qualification into a semifinal or final heat of Olympic races. The top 8 times for the women’s 50 meter freestyle from the 2016 Rio Summer Olympics are shown in the table below, along with a 1% increase and decrease in time for each athlete. In 2016, there was a difference of 0.02 seconds between first and second place, 0.04 between first and third place, and 0.59 seconds between first and last place in the women’s 50 meter freestyle. If a current were present in the pool, it would have undoubtedly impacted the results of this high profile, international competition.
This simulation was built using the following assumptions:
1. It uses a right skewed distribution based on the top 8 performances of each athlete. The races chosen were not swam in the Rio facility.
2. The majority of swimmers competing at the Olympics do not typically set world records, but generally their performance is close to their career fastest times. Therefore, using fastest times outside of the Olympics is a relatively accurate predictor of their performance at the Olympics.
3. The simulation does not factor in whether the swimmer is competing in the preliminary heats, semi-final heats, or final heats. This could significantly influence the performance of the swimmer as athlete effort may vary by heat and a pool current would be present during all session of the meet.
The initial simulation predicted a time performance for each swimmer in the final heat of the 50 meter freestyle and subsequently predicted their place in the heat based on the time. I ran 10,000 Monte Carlo simulations to compare the modeled times to those achieved at the Rio Olympics. Below are example distributions for Aliaksandra Herasimenia and Etiene Medeiros, athletes who competed in lanes 8 and 1, respectively, during the 50-meter freestyle final at the Rio Olympic Games. The histograms of their simulated swims are shown below:
When comparing the time swum by Herasimenia in Rio with the simulated times, it is evident that she swam significantly faster than would have been predicted based on her previous times. This finding is consistent with the hypothesis that there was a circular current in the pool. Medeiros on the other hand, swam a time that fell within the prediction interval based off her past performances (24.5, 24.9).
To judge whether a current had an impact on the swims from more than just one heat at the Olympics, I then ran the simulation on 3 different heats from both genders: the final and both semifinals (48 individual swims in total). In five of the six heats, the times of the athletes competing in lane 8 were below the lower confidence limit of the 95% confidence intervals from the simulated races. I then calculated the mean squared error for each of the races by lane and combined them to judge the overall error by lane (shown below).
The plots of mean squared error and absolute error are both consistent with the hypothesized results of the circular current theory. Lanes 1, 2, 7, and 8 would have been the most impacted by a circular current, and all four of these lanes have the highest errors.
The most controversial part of this circular current is its potential impact of the finishing places of the athletes. To look at the impact of the current on finishing place, I took my original simulation and ran iterations of each race 10,000 times. I then combined the simulations for the three women’s heats, the three men’s heats, and finally all of the heat simulations together to create a result matrix. The table below is the result matrix for the women’s heats.
The table above clearly demonstrates that there are differences between the simulated place finishes and the real place finishes for some lanes, especially lane 1. In the simulation, swimmers in lane 1 would finish first 3 percent of the time, and finish in eighth 43 percent of the time. In the simulation, compared to the real results, where 0 swimmers from lane 1 finished first in their heat, but, 43 percent of them finished last. In other lanes, the simulation produces larger probabilities than the actual lane occurrence. For example, according to the simulation, a swimmer in lane 3 would finish first about 6 percent of the time, while in actuality, a swimmer in lane 3 finished first 17 percent of the time.
While this simulation supports the hypothesis of a circular current in the pool, there are other elements that could be factored into the simulation to create a more accurate prediction of the race results. Some of these factors include: the inclusion of lane bias into the simulation to demonstrate the effect of starting from a faster-seeded lane on a swimmer, or how much faster/slower a swimmer typically performs when competing at the Olympic level. However, this result is another nail in the coffin for the company that built the pool, and the Olympic committee’s dedication to fair competition. The difference between 3rd place and 4th place could be a sponsorship deal for these athletes, and the difference between 1st and 2nd could be tens of thousands of dollars lost.
Delaney Ambrosen is a junior economics major and statistics/computer science minor at Kenyon College.
