Category Archives: Sports Science

Comparing Barry Sanders and Emmitt Smith

Barry Sanders and Emmitt Smith are arguably two of the greatest running backs of all time. They dominated the 90s with their running styles and were both eventually enshrined in the NFL Hall of Fame. Fans and media folk have often debated “Who was better as a running back”. It is a bit of a tough question to answer. First, stating the obvious, Emmitt played an extra 5 years than Barry. Emmitt played 15 seasons while Barry retired after just ten seasons, most people feeling he cut his career short. Secondly, Emmitt was on some exceptional Cowboys teams and had a supporting cast of great offensive linemen, a solid blocking fullback, and a hall of fame quarterback with great receivers, which set him up for opportunities that weren’t always present for Barry (the Lions only finished first in the, what was then referred to as the NFC Central, twice during Barry’s ten year career).

But what if Barry had played the extra 5 years that Emmitt did? Barry ended his career with 15,269 yards to Emmitt’s 18,355 career yards. How many more could Barry have gotten?

Side Note: I removed receiving yards from this analysis as I only wanted to look at rushing yards. Barry had 2,921 career receiver yards while Emmitt had 3,224 career receiving yards, most of those coming in his first nine seasons with the Cowboys.

First, let’s look at some visualizations of the career both of these superstars had. We will look at their yards per season, their average yards per attempt each season, and their average yards per game each season.

Screen Shot 2015-01-04 at 7.30.30 PM

Screen Shot 2015-01-04 at 5.46.02 PMScreen Shot 2015-01-04 at 5.46.54 PM

Looking at the graphs, it seems like Barry was a more consistent runner than Emmitt and, after the age of 27, Emmitt’s production began to dip a little bit, while Barry stayed relatively consistent. During Barry’s 10 year career, he averaged 4.98 +/- 0.6 yards per attempt,  99.9 +/- 14.2 yards per game, and 1,526.9 +/- 270.8 yards per season. Looking at the first 10 years of Emmitt’s career, he averaged 4.3 +/- .46 yards per attempt, 90.3 +/- 17.6 yards per game, and 1,396.3 +/- 267 yards per season.

In the last 5 years of his career, Emmit averaged 3.62 +/- 0.5 yards per attempt, 59.42 +/- 19.9 yards per game, and 878.4 +/- 362.6 yards per season. It is important to remember that Emmitt missed six games during the 2003 season due to a fractured shoulder blade, causing him to only get 90 attempts that year (a far cry from his average of 269 attempts per year that he got during that time period, after removing the 2003 season from the data set).

Equivalence Coefficient

So, how much better would Barry have been had he played the same amount of games that Emmitt played – what amounts to 73 extra career games (since Emmitt missed 6 games in 2003, 2 in 2001, and 1 in 2004)?

In baseball, Sabermetricians will sometimes use the Equivalence Coefficient (EC) as a means of projecting out performance in specific metrics given an equivalent scenario.

First, we determine that, had Barry been given 73 extra games in his career, he would have had approximately 1461 total attempts (based off his previous career numbers). This allows us to calculate Barry’s EC:

1+(1461/3061)*1.00 = 1.477295

By multiplying this coefficient by Barry’s career total yards, 15,269 yards, we project that, given 73 extra games (or 1461 extra attempts) Barry would have rushed for a career total of 22,557 yards (4202 yards more than Emmitt).

Of course this assumes that during these extra 73 games Barry is 100% of the player he was in the past. Oftentimes, people will talk about the the magic age of 27 representing a “cliff” for running backs where their performance begins to drastically decline (it seems like Emmitt may have started to decline around this age). Looking at Barry’s charts above, it certainly doesn’t look like he was showing signs of decreased performance at the age of 27 and there is no reason to believe that he wasn’t going to be the same player in those next 73 games that he was in the previous 153 games. However, let’s assume that perhaps Barry is not the same player in those next 73 games. Looking at Emmitt’s production in his final 4 seasons (I took out the injury plagued 2003 season where he only had 90 attempts in 10 games), Emmitt saw a 16% decrease in average yards per season, a 6% decrease in average yards per attempt, and a 15% decrease in average yards per game. So, let’s say, hypothetically, that during the next 73 games, Barry would be 10% less of the player that he was in the first 10 years of his career (a 10% decline in his overall performance). If this is the case, Barry’s new EC is calculated to be 1.429566 and his projected total for career yards would have been 21,828. That is 3,473 more yards than Emmitt Smith had in his career.

What does it all mean?

It is fun to play with projections like this. This doesn’t necessarily mean that Barry was a better running back (remember, I didn’t factor in receiving yards) but it certainly does show the incredible skills that Barry had and what he was able to accomplish in a career that he ended too soon (in the opinion of many). From a sports science/strength & conditioning perspective, projecting out performance allows us to understand the potential that our athletes have and what their projected performance may look like given a decrease in their overall output. This is important in a team sport as the chaotic nature of performance is difficult to tie back to the S&C coach (it certainly is not as easy to make the connection as it is with an individual athlete sport, such as track and field, swimming, or cycling). With this information we can begin to grasp the value that our programs may have on prolonging an athlete’s career by keeping them healthy and on the field/court where they can perform at the highest level, helping them to maintain their performance without seeing as large a decline with age as other athletes may see.

NBA Super Teams – “They Just Need to Learn To Play Together”

Being a life long Cavs fan, I was excited to see LeBron return to Cleveland to try and make a run at a championship. I’ll admit, in 2010, when he made the decision to leave for Miami, I was pretty upset; but, I don’t fault players for going to different teams if it means more money and better opportunities to win championships. Additionally, how can you fault a guy for wanting to play on a super team with athletes like Dwyane Wade and Chris Bosh? What makes LeBron’s second stint in Cleveland so exciting is the possibility of another super team, this time with Kevin Love and Kyrie Irving. Of course, when these teams get together and don’t automatically win 20 straight right out of the gate, fans tend to get a bit unruly and start to jump ship.

Similar to LeBron’s first year in Miami, it took some time for things to click between the players on the team. The media pundits always like to bring it back to the fact that the players need to, “Learn to play together”,  because, as some point out, they have such similar games and are all three such dominant players on the court that they have to figure out who is going to play which role (which might actually change from game to game).

With that in mind, I decided to look at some of the data of these players on the super teams to see how similar they are and, perhaps, try and understand how similar LeBron’s current super team, the Cavs, is to his former super team, the Heat.

The Data

Since I don’t have access to a ton of NBA data, I took whatever I could get a hold of from I compiled the player data for all of the years LeBron was at the Miami Heat, all of the years Kevin Love was at the Minnesota Timberwolves (his entire career), and all of the years Kyrie Irving has been at the Cleveland Cavs (his entire career).

Only players who participated in more than 30 games per season where included (NOTE: Doing this removes Kevin Loves 2012-2013, as he only played in 18 games that season do to several hand injuries).

I then created a cluster analysis to evaluate how similar or dissimilar players in the data set were (you can click on the picture to make it larger).

Screen Shot 2014-12-07 at 5.14.50 PM As we see, LeBron lies on a node all to himself, and rightfully so! LeBron is a truly unique player, who can play every position on the court and is both a significant defensive and offensive threat. To the right of Lebron, we see a second node, which then breaks down into two more nodes. It is in this cluster on the tree that we see the main players we care about – those that make up LeBron’s former and current super team. Kevin Love and Chris Bosh are clustered close together while Dwyane Wade and Kyrie Irving are clustered close together, owing to the similarities in their game.

LeBron’s Supporting Cast

We see that the supporting cast for LeBron have some similarities when we evaluate their performance metrics within the data set of these three teams (Cavs, Heat, and Timberwolves). Looking deeper at LeBron’s supporting cast in both Miami and Cleveland allows us to see how each of these players compare to each other.

Screen Shot 2014-12-07 at 6.07.04 PMLooking at Field Goal Percentage, we see that LeBron’s former teammates are a clear favorite when it comes to making shots. However, it is important to remember that this data is showing us the years that Wade and Bosh also played with LeBron. Perhaps having LeBron on the team created new opportunities for Wade and Bosh to score, opportunities that Irving and Love did not have given the weaker teams they were on?

Screen Shot 2014-12-07 at 6.09.57 PM

When it comes to rebounds we see that Kevin Love has had more success in this category than the other four players, while we see that Bosh is roughly equal with James.

Below is some comparison of the four players against each other for a few of the other metrics in the data set (all graphs are showing mean +/- SD).

Screen Shot 2014-12-07 at 6.16.28 PM Screen Shot 2014-12-07 at 6.16.46 PM Screen Shot 2014-12-07 at 6.17.03 PM Screen Shot 2014-12-07 at 6.17.18 PM


It is interesting to compare these players against each other. I am a bit surprised to see Bosh with such a low number of assists compared to the rest of the group. Additionally, it is interesting to see how many blocks Wade and Bosh had in comparison to Irving and Love. The Heat were definitely a defensive juggernaut, while also being a huge offensive threat, making them one of the most dangerous teams in the NBA. While it appears, from the cluster analysis, that Love and Irving play similar games to Bosh and Wade, it seems like Bosh and Wade are slightly better in some of the statistical categories.

Wade was already a great player (one of the best ever) before LeBron, and Bosh was no slouch himself. I wonder if “figuring out how to play together” allowed them to improve different aspects of their game, as having a unique player like LeBron on the court tends to draw a lot of interest from the opponents defense, opening up new opportunities for Wade and Bosh. One way to evaluate that might be to look at their individual statistics during their career before LeBron and then after LeBron. Additionally, it might be interesting to compare how much LeBron’s game changed from his first years with the Cavs and then his years with the Heat, where he “figured out how to play” with the other superstars on the court.

It will be interesting watch James, Love, and Irving, figure it out over the next few seasons. They have potential to turn this into another big three and hopefully dominate the NBA. With the similarities between Bosh and Love, and Wade and Irving, perhaps LeBron can put his new teammates into a good position to elevate their game and take things to the next level.

Minimum Effective Training Dose

You hear the phrase all the time, “I’m a real minimum effective dose guy”, or, “We train only as much as we need and then no more”.

Everyone says these things, but what do they really mean? What is a “minimum effective dose”? Is the minimum effective dose different for different people? Do some people need more training and some less? While the phrases sound good on paper or when uttered at a training conference, how do we take the theory of the minimum effective dose and turn it into practice?

To be fair, these are great ideas and statements that really do resonate with me in my approach to program design. Why expend physical resources (energy) on training that are unnecessary and potentially limiting your recovery from the previous session, thus diminishing your ability to train harder the next time around? As I like to say, “There is always a cost of doing business. All training comes at a cost and in order to reap the benefits you need to make sure you pay that cost and then replenish the checking account before paying again.”

Recently, I had a great discussion with two colleagues I respect – Sam Leahey and Nate Brookreson. We were discussing concepts around an individualized training approach, and the main discussion points began with us first reading and talking over two papers by Kiviniemi, et al., Endurance Training Guided Individually by Daily Heart Rate Variability Measurements (Eur J Appl Physiol, 2007) and Daily Exercise Prescription on the Basis of HR Variability Among Men and Women (Med Sci Sports Exer, 2010).

Both studies utilized a similar type of training approach for the two training groups. One group performed a standard, predetermined training program – just like a coach would write for an athlete, dictating what should be done each day of the week (exercises, load/intensity, sets, reps, etc). The other group performed their training based on their HRV readings taken first thing in the morning, upon waking. The mode of exercise in the studies was endurance training, and days were broken into high intensity (40min at > 85% of maxHR) or low intensity (40min at 65-70% maxHR) or complete rest.

The way it worked for the HRV-dictated training group was that they would take their HRV, and based on the outcome, compared to a rolling average, they would alter their training for the day performing either a high intensity session, a low intensity session, or taking a rest day. Thus, training was guided by what the body was prepared to do.

Interestingly, the HRV-dictated training groups improved their fitness while training high intensity sessions less frequently during the study period than the predetermined training group (More is not better. Better is better). Basically, on days when their body was ready for a high intensity training session they went for it, and when their body was not ready they backed off and allowed the body time to replenish the checking out, so to speak, before repaying the cost. They gave the body what it needed.

Some of my thoughts

Heart Rate Variability is not the be-all-end-all of athlete monitoring, as some make it out to be. It is one small piece (a small piece with rather noisy data, mind you) in a much larger puzzle. That being said, I do believe it can have a role in athlete monitoring if you understand its limitations, standardize the collection process, and couple it with other methods of monitoring the athlete and evaluating their capability and capacity on a given day.

These studies seem to move us closer to understanding the concept of a minimal effective dose and perhaps offer a newer approach to program design and periodization – similar to the concept of auto-regulation. Earlier this year I put together a decision tree for training, similar to the one shown in one of the studies mentioned above, where a few factors were taken into consideration and put into the tree, and the results of those factors allowed the athlete to alter their training program based on the input they plugged in. This allowed us to adjust the program up or down on a given day based on how the athlete was responding. Instead of writing training programs that told the athlete to do “x” on Monday, “y” on Wednesday”, and “z” on Friday, the athlete was given different workouts with different training targets (2 workouts reflecting the main physiological targets of the training block, 1-2 workouts reflecting the secondary, or maintenance, physiological targets of the training block, and 1 recovery based work).

Depending on how the athlete was reporting that day, we would choose which workout to do. This would end up sometimes pushing our training week out longer than 7 days (sometimes it would take 10 days to get through the training cycle). This was apparent, particularly, in older athletes whose bodies took longer to recover from the training session or athletes who were out of shape and lacked fitness and needed the extra time to make appropriate adaptations to the training stimulus imposed upon them. If we were working on a timeline and had a set duration of time to perform a block of training (for example the athlete would only be able to train 10 weeks in an offseason), we would adjust the workout on a given day by lowering either training volume or training intensity (which of those we lowered was dependent on the physiological targets of that phase of training and what the main objective was).

What was also interesting in the studies above was that if the subject had recovered the following day from a high intensity training day they would then perform another high intensity session (although after two successive high intensity sessions they were asked to take a rest day). The high-low training concept of organizing high intensity stressors on one day and low intensity stressors on another day is a great one and one that I have used for many years; however, there are times when the athlete needs to be able to put together back-to-back days of high intensity work due to competition (i.e., basketball or hockey) or hard practices (i.e., NFL training camp) being stacked together. By using a training approach driven by monitoring the athlete’s response and adjusting the workout to suit their needs and abilities on a given day, we can slowly build up the athlete’s resilience to tolerate high intensity work to a level that allows them to train hard, recover quickly, and then train hard again. This is a key piece that ties together the stress resistance/stress tolerance and fitness components of my Physiological Buffer Zone methodology, which I discussed in THIS interview.

What it basically boils down to is that each athlete is an individual. Each athlete has a different way of responding and adapting to the training stress you apply to them (and even to the treatment stress if you are using soft tissue work!). The time it takes to recover and make favorable adaptations to a training session may differ from one athlete to the next, and an individualized approach, based on monitoring various qualities, is essential to understanding what the athlete needs. Too often coaches try and force fit an athlete into their training program without respecting these laws of individualization. Hopefully the future will allow for better methods to test athletes, monitor/evaluate athletes, and adjust training for athletes to ensure that their body receives the type of training it needs – the correct amount at the correct time.

A Scale of Perception for Bar Velocity

Questionnaires have been around for a long time and been found to be valid and reliable once the athlete is properly anchored to the scale. While it may sound simple, there is actually a lot of complexity within the simplicity of just asking a person a few questions regarding how they feel today or how hard they felt a particular activity was (RPE). However, once the individual understands what they are being asked, and gains some experience rating themselves, usually about 4 weeks,  questionnaire data can be very helpful in planning training. (I have been a fan of using questionnaire data as a method of understanding how an athlete is tolerating training for several years and wrote about the daily questionnaire I use in a previous blog article.)

Recently, Bautista and colleagues (2014), have attempted to create a new scale, which allows the athlete to rate their perception of bar velocity in the bench press (CLICK HERE for full paper).

Measuring bar velocity is incredibly helpful and is done by attaching some sort of linear position transducer to the bar to objectively measure the speed at which the bar is moving through various lifts (E.g., bench press, squat, deadlift). The 1RM of the subjects in the study was established prior to using the rating scale, during an incremental load protocol. A linear position transducer was used to understand bar velocity at various percentages of the individual’s 1RM during the incremental load test:

  • Light = < 40%
  • Medium = 40% – 70%
  • Heavy = > 70%

Over a 5 day testing period, the subjects performed each set in a random order, using the intensity parameters above, and were blinded to the amount of load on the bar via partial occlusion pads, which prevented them from seeing the weight. The subjects performed 2-4 repetitions with a given load and then provided their perception of bar velocity using a scale developed by the authors, based on bar velocity during the incremental load 1RM test:

Screen Shot 2014-09-07 at 5.09.39 PM

The verbiage below the numbers, used to anchor the subjects during the experimental portion of the test, was established using the corresponding bar speed form the incremental load 1RM test and the verbal qualification provided by the subjects following each of their lifts during the initial test.

A high correlation was found between the actual bar velocity and the perception of bar velocity provided by the subjects, particularly as their use of the scale increased. Thus, greater exposure and time using the scale improved their ability to properly classify their lift.

Practical Use

As stated earlier, I am a big fan of questionnaires. While they are especially helpful when combined with other objective data (GPS, HR, Fitness Testing, Bar Velocity, etc) as a stand alone they can provide rich information once the athlete is properly anchored to the scale.

I see the Rating of Bar Velocity scale used in this study being practical in a few ways:

  1. Not all strength and conditioning programs have funds to provide a linear position transducer unit at each lifting platform. However, if athletes gain an understanding and awareness of how to rank their bar velocity, this method can be useful as an inexpensive means of determining individual percentages for power training. (NOTE: I do think it would be of value to at least have one or two linear position transducers available to allow the athletes to initially understand how fast they are moving the bar, as well as to have available on testing days.)
  2. Not all athletes will move the same relative intensity at the same speed. This will allow the coach to adjust the training intensity up or down for the athlete, in order to stay in their ideal zone of bar speed, depending on the training goal for the day.
  3. Similar to using a Rating of Perceived Exertion on a fitness test, the Rating of Perceived Velocity can be used on a strength test or Rep Max test and charted over time to show improvement with the same load or the same relative intensity.
  4. Finally, having athletes rank their efforts like this, I find, increases their awareness of the training session and engages them more in what they are doing. Rather than going through the motions, the athlete has to now be conscious of what (s)he is trying to do.


Tommy John & Pitch Counts

This year, eighteen pitchers have undergone Tommy John surgery in the MLB (last I checked) and we are not even half way though the season. There is tons of speculation about why we are seeing such a rise in ulnar collateral ligament (UCL) injuries and I don’t think it is ever easy to boil any injury down to one single factor. Even when you think you have everything figured out and have crossed your “T’s” and dotted your “I’s” injuries can creep up because of factors that you might not be aware of or factors that are outside of your control.

That being said, one of my good friends works in major league baseball as a manual therapist and strength coach and during a conversation one day he mentioned that it would be interesting to know how many pitches Jose Fernandez (the second year phenom pitcher for the Florida Marlins who pitched in 8 games this year before undergoing Tommy John Surgery and will miss the remainder of the season) threw in those eight games. Again, hard to pin-point injury to one single factor and just having something like pitch count doesn’t provide you with enough information to really understand the whole situation (although famed orthopedic surgeon, Dr. James Andrews, does feel that throwing too much and overuse may be one potential factor in UCL injury). However, I thought it would be interesting to look at so here are some of the things I quickly found after looking at a few websites and getting pitching statistics.

During Jose’s 2013 season (his rookie year) he:

  • Threw 2604 total pitches over 28 starts
  • Averaged 93 pitches per game

During the first 8 games of the 2013 season he:

  • Threw 665 total pitches
  • Averaged 83 pitches per game

During Jose’s 8 games in 2014 he:

  • Threw 770 total pitches (105 more during the first 8 games of this season than the 2013 season)
  • Averaged 96.25 pitches per game (averaged 13 more pitches per game over the first 8 games than the 2013 season)

Graphically, this is what Jose’s 2013 season looked like:

Screen Shot 2014-06-02 at 7.08.44 AM

You can see that over the first 11 games the teamed seemed to take it a bit easier on the rookie and ease him into the swing of things (there may be reasons for this – again, tough to know with just pitch count information) – before really pushing it the rest of the way.

In comparison, here is what the first eight games of the 2013 and 2014 seasons looked like together:

Screen Shot 2014-06-02 at 7.12.59 AM

I don’t know that there is a big take away here other than it is interesting to look at the way Jose was managed from his rookie year to his second year in terms of pitch count. There was a considerable jump in the first 8 games of the 2014 season compared to the 2013 season and the second half of the 2013 season. It would be great to have more information – daily readiness information, fitness information, how much throwing he does/did between starts, how much he did in spring training, etc, on-and-on.

Perhaps one factor (of many) that plays a role in UCL injury is the management of pitchers in terms of how much they throw or how it is determined that they are fit enough to go out and throw a certain amount or tolerate greater volumes of throwing? I’m sure the picture is much larger than this though.