Applied Sports Science newsletter – July 11, 2017

… Frustration and negativity became constant companions, causing Smoak to tinker and brood and doubt, desperate for ways to unlock the abilities he still believed he had.

Atkins knew how much churn existed behind Smoak’s stoic façade, so he asked the slugger if he’d consider talking to a sports psychologist.

“I was like, ‘I’m open to it. I’ve never done it before,’” recalls Smoak. “I’d talked to sports psychologists before, but it was more, ‘hey, whatever.’ But I was open to learning and open to trying to figure things out.”

… “It’s been a long time coming,” Vaughn said. “I’ve been putting in work for a number of years but never could put in a performance that quite matched my fitness.”

Vaughn and her husband, Brent, who ran professionally in Portland, Oregon, prior to retiring from the sport, had their first daughter, Kiki, now 10, when Vaughn was a junior at the University of Colorado. The couple welcomed Calia (now 7) three years later and then Cassidy, who turns 2 in August.

In a recent phone interview with Runner’s World, Vaughn talked about what it’s like to be a professional athlete, mother, wife, and full-time realtor.

In May, Kerri Walsh Jennings left the AVP tour and split with her 2016 Olympic partner, April Ross, because she believes the current power structure is preventing athletes from making a living wage in beach volleyball. Now, she’s trying to change that.

The head-spinning, mouth-drying, bone-crushing fatigue of jet lag is enough to make anyone think twice about changing time zones. But sometimes, traveling is worth the suffering — or, it’s just unavoidable. So we quizzed Stanford sleep expert Jamie Zeitzer about how to minimize the misery of jet lag.

The body operates on a biological schedule known as the circadian clock. Seeing daylight at specific times of day helps set this clock — but it’s slow to adjust when we rapidly jet across time zones. That lag accounts for some of the nighttime sleeplessness and daytime sluggishness we experience while traveling, Zeitzer says. Packing into an airplane and eating unusual food at unusual times probably doesn’t help much, either. “There are so many things that are going wrong at once,” he says.

Introduction
The ‘Barça structured training’ method involves simulating multiple team-sport specific situations to expose the player to a varied training stimulus. The increased training variability is believed to enhance physical performance during football competition. The purpose of the present study was to investigate how the degree of variability in training influenced player physical performance during matches using a clustering algorithm and statistical test approach.

Methods
Physical performance data was gathered from 25 male professional football players of Barça B team during one season, including 153 training sessions and 34 matches, and using STATSports Viper pods. 9 variables were identified: distance (DIS), number of sprints (SPR), max speed (MAX), high metabolic load distance (HML), high metabolic efforts (HEF), relative metabolic power (PER), dynamic stress load (DSL), total loading (TLO) and accelerations (ACC). Variability was calculated by dividing the mean session value by the standard deviation of the microcycle, for each given variable and player. A dataset was created by grouping the variability of 3 week microcycles, including training and matches. From this dataset, K-Means clustering was applied to group the microcycle variability in an optimal number of clusters. Physical performance of each player during matches following each the microcycle was labelled respectively. For both clustered training groups and matches groups, independent t-test were performed and the standardized difference of means was assessed.

Results
Of the 9 variables in the training groups, very large effect sizes (ES > 2.0) were found for DIS (ES=2.1, p<0.001), PER (ES=2.2, p<0.001) and TLO (ES=2.4, p<0.001); large ES (1.2< ES <=2.0) for DSL (ES=1.9, p<0.001), HEF (ES=1.9, p<0.001) and SPR (ES=1.3, p<0.001); and moderate ES (0.6< ES <=1.2, p<0.1) for ACC (ES=0.64, p=0.01), HML (ES=1.16, p<0.001) and MAX (ES=0.81, p<0.001). Cluster analysis showed high variation and low variation groups. The high variation group presented higher physical values in matches, showing moderate ES for PER (ES=1.14, p<0.001), TLO (ES=0.72, p<0.001), DIS (ES=0.77, p<0.001), HEF (ES=0.7, p<0.001) and HML (ES=0.68, p<0.001); and small ES (0.2< ES <=0.6) for ACC (ES=0.40, p=0.066) and DSL (ES=0.44, p<0.001). Discussion
In conclusion, training involving high variation, was associated with significantly superior physical performance during matches, in comparison to low variation training. Training variation in microcycles can be classified as ‘high’ or ‘low’ based on those variables identified in the present study. Understanding of the relationship between training variability and match physical performance allows an objective assessment of training practice in football.

I want to monitor my athlete but where do I start?

Given the relationships among athlete workloads, injury1 and performance,2 athlete monitoring has become critical in the high-performance sporting environment. Sports medicine and science staff have a suite of monitoring tools available to track how much ‘work’ an athlete has performed, the response to that ‘work’ and whether the athlete is in a relative state of fitness or fatigue. The volume of literature, coupled with clever marketing around the ‘best approaches’ to optimising athlete performance, has resulted in practitioners having more choices than ever before. Furthermore, the range of different practices used in sport and the lack of agreement between parties emphasise the importance of having a clear rationale for athlete monitoring. The aim of this paper is to provide a practical guide to strategic planning, analysing, interpreting and applying athlete monitoring data in the sporting environment irrespective of data management software. [full text]

This study aimed to investigate the effects of wearing a compression garment (CG) during night sleep on muscle fatigue recovery after high-intensity eccentric and concentric knee extensor exercises. Seventeen college male students participated in two experimental sessions under CG and non-CG (NCG) wearing conditions. Before night sleep under CG or NCG wearing conditions, the subjects performed a fatiguing protocol consisting of 10 sets of 10 repetitions of maximal isokinetic eccentric and concentric knee extensor contractions, with 30-second rest intervals between the sets. Immediately before and after and 24 hours after the fatiguing protocol, maximal voluntary isometric contraction (MVIC) force for knee extensor muscles was measured; surface electromyographic data from the vastus medialis and rectus femoris were also measured. A two-way repeated-measure analysis of variance followed by Bonferroni pairwise comparisons was used to analyze the differences in each variable. Paired sample t-tests were used to analyze the mean differences between the conditions at the same time points for each variable. The MVIC 24 hours after the fatiguing protocol was approximately 10% greater in the CG than in the NCG condition (p = 0.033). Changes in the electromyographic variables over time did not significantly differ between the conditions. Thus, it was concluded that wearing a CG during night sleep may promote localized muscle fatigue recovery but does not influence neurological factors after the fatiguing exercise.

… Vi is an AI whose sole job is to make you a more proficient runner. It won’t open the Notes app or help you with productivity at work. It is a specialized personalized trainer — that’s it. On top of the AI component, Vi is a voice-activated Bluetooth headset, one complete with Harman Kardon speakers.

The AI features a built-in barometer to gauge elevation, along with a heart rate sensor, gyroscope, and an accelerometer, which will track your steps and a variety of other fitness metrics. Vi can also respond to basic questions you pose during your run. You can ask her about your heart rate in real time, whether you’re currently exercising within your optimal range, and a host of other questions. Simply say “step to the beat” and Vi will play a consistent beat, allowing you to develop a steady cadence for an even pace.

Prior to your workout, Vi will access weather data to determine whether you should run inside or outside on a particular day. The virtual assistant also learns from your previous workouts and health data, allowing her to adjust your regimen and coach you along the way.

We propose Human Pose Models that represent RGB and depth images of human poses independent of clothing textures, backgrounds, lighting conditions, body shapes and camera viewpoints. Learning such universal models requires training images where all factors are varied for every human pose. Capturing such data is prohibitively expensive. Therefore, we develop a framework for synthesizing the training data. First, we learn representative human poses from a large corpus of real motion captured human skeleton data. Next, we fit synthetic 3D humans with different body shapes to each pose and render each from 180 camera viewpoints while randomly varying the clothing textures, background and lighting. Generative Adversarial Networks are employed to minimize the gap between synthetic and real image distributions. CNN models are then learned that transfer human poses to a shared high-level invariant space. The learned CNN models are then used as invariant feature extractors from real RGB and depth frames of human action videos and the temporal variations are modelled by Fourier Temporal Pyramid. Finally, linear SVM is used for classification. Experiments on three benchmark cross-view human action datasets show that our algorithm outperforms existing methods by significant margins for RGB only and RGB-D action recognition.

… For those with sickle cell trait, which has killed 11 college football players since 2000, according to the Lincoln Journal Star, the exhaustion comes sooner. “If you have a flare-up, you might feel it on the eighth or ninth sprint,” Powers explained.

The athlete’s body demands a break. But his mind—and his coach—may have other plans: No football player wants to appear weak or out of shape. So he keeps going.

If he doesn’t stop when the flare-up occurs, his own blood cells rapidly start trying to kill him. They form into crescent shapes and clog the blood supply to his muscles. The muscles die due to lack of oxygen. They dump their contaminants into the bloodstream, which, according to Dr. Kimberly Harmon of the University of Washington, interrupts the electrical system in the heart and causes cardiac arrest.

Being described as a “tweener” in the NBA used to be considered a dirty word, an adjective used for a player too big or small to fit a traditional position. The league, however, has evolved.

Now teams seek out players that have a Swiss Army knife-like skill set that can fit into a more free-flowing, positionless game. There’s value in being able to segue into multiple roles.

“I don’t have the five positions anymore,” Celtics coach Brad Stevens said. “It may be as simple as three positions now, where you’re either a ball-handler, a wing or a big.

… The Golden State Warriors’ first foray into attempting to use analytics and other such data science came from a test of their G-League affiliate, the Santa Cruz Warriors. Joe Lacob’s son Kirk Lacob, fresh out of Stanford, was made assistant GM and was put in charge of the G-League team, where he employed a statistics-heavy approach to front office management. The plan here was for Santa Cruz to be a test bed for the team’s more forward-thinking plans, where they dealt with much lower-stakes scenarios. The successes from Santa Cruz would then be evaluated and then employed by the main organization in Oakland.

The Santa Cruz Warriors became a model developmental team under the younger Lacob. In his first two seasons in Santa Cruz, the team made the development league finals each year. In addition to the consistent winning, Santa Cruz consistently pumped out solid NBA contributors such as Kent Bazemore and Dewayne Dedmon. It became clear that the experiment in Santa Cruz was a success, and the younger Lacob was brought up to join the head front office of the Golden State Warriors

As a football industry we’ve been trying harder and harder in recent years to find ‘undervalued talent’, with mixed success.

Broadly, there are two approaches we can take. The first is bottom-up; finding traits or actions that our competitors dismiss, but actually tells us more about a player. For example, valuing the quality and quantity of chances created, rather than actual assists. Or prioritising what a player has done in his league matches rather than international tournaments.

The second approach is top-down. This is a more literal approach on finding undervalued talent – it involves looking for talented leagues and teams that pay less. Take the Czech First League, below. The latest UEFA Benchmarking Report revealed that the league’s average club earned just €3.2m in 2014-15, which makes them the 29th wealthiest top division in Europe.

Real-world data is almost always incomplete or inaccurate in some way. This means that the uncertain conclusions we draw from it are only meaningful if we can answer the question: how uncertain?

One way to do this is using Bayesian inference. But, while Bayesian inference is conceptually simple, it can be analytically and computationally difficult in practice. Probabilistic programming is a paradigm that abstracts away some of this complexity.

There are many probabilistic programming systems. Perhaps the most advanced is Stan, and the most accessible to non-statistician programmers is PyMC3. At Fast Forward Labs we recently shared with our clients a detailed report on the technology and uses of probabilistic programming in startups and enterprises.

… There are two areas to explore. First, to understand how analytics can disrupt existing business models, set aside the time to focus on the long term. What can be learned from other industries that are farther along? What customer needs can be better met through new business models?

Second, to capture new opportunities, start with the data, analyzing what they are worth, how distinct they are, who would find them valuable, and how they can be combined with other sources to increase their value. Then, think through the business model. A simple way to get started is to conduct a market scan of the data and analytics players, as well as a competitor scan to understand what others may be doing. Identify where and how to play within this ecosystem.

Surprisingly few companies know where and how analytics can create value

Analytics create value when big data and advanced algorithms are applied to business problems to yield a solution that is measurably better than before.