Analyzing Video and High-Speed Photography of Andre Agassi Using the EASI Tennis System

Contents

In our article "What We Must Do to Revitalize Tennis," we described fundamental challenges to teaching tennis. Those challenges suggest we must reformulate our teaching methods to be responsive to how students learn rather than organizing them around how we like to teach. In this article, we address the problem of providing accurate and factual information to the student in place of metaphors, templates, and rules, using an analysis of Andre Agassi's forehand as a basis for our discussion. Copyright restrictions prevent us from showing the actual footage of Agassi. The original footage we used can be obtained from TennisOne for a small donation to their research project, and we strongly encourage everyone to do this.

The most accurate physical data available on how pros hit their shots comes from high-speed video. Additionally, high-quality data can be found in the ProStrokes Gallery of TennisOne, and on ESPN and other televised matches. Our success in using this data to improve student performance depends on our ability to analyze this data and transform it into information.

The challenge of using both conventional and high-speed video lies in the volume of data it provides (this is generally called the raw data). Our specific challenge is to transform the raw data into useful information. To do this, we must organize the data into a format from which information can be extracted, tested, and transferred in a systematic and orderly fashion.

Technical Explanation of Our Video Analysis

The reader who is not interested (at this point) in the technical explanation of our video analysis may skip this section.

Video data provides motion and direction. The organization of such data can be approached by dividing the motion into a series of stages and the transitions between these stages. The two related key questions are: how many stages are there, and what are they?

Optimally, the stages should be numerous enough to capture all of the important transitions, but not so numerous as to break important units into fragments. Classical tennis instruction divides the forehand stroke into three stages: the takeback, the swing, and the follow-through. The three classical stages are clearly too few to capture all of the important information about a stroke. As a result, the classical approach leaves a large burden on the student to explore a vast array of possibilities. This is not an indictment of the classical theorists but a reflection of the limited technology available to them (for example, the classical approach was formulated in an era in which video analysis was not practical for teaching tennis). The key limitation of the classical approach lies in the lack of details about the swing stage of the tennis stroke. But it is a fact that the swing stage is where all the important details for hitting a clean professional forehand lie, and therefore the swing must be analyzed further if we are to accelerate the process of learning.

Our starting point to analyze the swing was to note that, to transform motion data into information, the selection of stages of the motion must be based on their purpose or intent (which might not be conscious). Any other selection does not provide as good an infrastructure for transferring the information to the student (refer to our first article). Hence, when selecting stages to be used to organize the video information for teaching, transferability must be the highest priority.

Second, we must select the fewest possible stages needed to get most of the information from the data. It is not possible to get all of the information. Also, the stages must be distinctive and not merely the result of an arbitrary division of the swing into equal time periods. (As a technical point, a division into equal time periods would likely be less useful than no division at all.) One stage (the classical approach) is clearly too few, and ten is clearly too many. An analysis of the relevant transitions as determined by evaluating higher-order derivatives suggested to us that the swing can be divided into four meaningful and distinct stages, each with a primary purpose that is easy to explain, easy to teach, and easy to learn. This is because, when there are significant changes in the second derivative of motion, the brain is shifting control from one part of the body to another for a purpose. We sought to capture these transition points and their purpose.

As a means of identifying and organizing the stages, we observed that any stage of any tennis stroke can have one or all of the following four purposes:

  • Provide stability to ensure that the stroke is controllable and can be used to carry out your intent
  • Provide racquet speed
  • Ensure that the racquet path is a straight line as the racquet approaches the point where it will strike the ball
  • Bring the racquet into alignment with the path of the ball

Every other aspect of a stroke can be explained in terms of these four dynamical features.

In short, the results of our dynamical systems analysis led us to organize tennis strokes into five stages:

  1. The classical takeback stage, whose purpose is to start the swing from an efficient and stable position
  2. The contraction stage, whose first purpose is to stabilize the stroke and whose second purpose is to provide initial acceleration
  3. The rotation stage, which supports further controlled acceleration and brings the racquet into alignment with the path of the ball in the most efficient manner
  4. The acceleration stage, during which the racquet is accelerated to maximum controllable speed
  5. The strike stage, during which the racquet path is directed into a straight line just before it strikes the ball

Any stage may contain one or more of these purposes. Further, these stages are not set forth as rules to be followed or templates to be duplicated. They are starting points for exploration and experimentation and provide an organizing structure around which to develop strokes. We have found that this division of the stroke provides a very effective infrastructure for rapid learning. In fact, this structure will enable an individual who may be considered to be, in the current vernacular, "untalented" to learn professional technique in as little as three years.

The Technical Setup for the Illustrations 

Our discussion of the EASI tennis stages of Andre Agassi's forehand uses a series of still photo illustrations of a professional player replicating this forehand as well as can be done by someone who is not Agassi. If you are not interested in reading an explanation of our technical setup at this point, you may skip to the following section.

The still photos were derived from high-speed video of Andre Agassi as follows. First, we studied the video using a VCR with full stop-action and advance tools. Our objective was to identify plausible transition points to test in the next stage of analysis. After a preliminary analysis, we transferred the video from the VCR to a computer, using the Dazzle Digital Video Creator II, where the five stages were determined and described.

The complete analysis took more than 15 months. Over this time period, we set forth numerous hypotheses and tested them both by experimentation and in the teaching environment of a girls' high school varsity team. The team was chosen because they did not recruit and because the school had no emphasis on sports achievement other than recreation. Further, the head coach was not a former player: her primary sport had been soccer. We imposed these requirements to rule out the factors of "talent," preconditioning, and prior professional knowledge.

Once we had identified the key transitions by experimentation and analysis, we imported the digital MPEG file of Agassi into MovieStar5 in order to break down each frame and capture the key transitions in static form. After we captured the five most important frames on the computer, we printed them for a reference point and went into our photography studio to produce the final illustrations for this article, using a professional player as a model. We kept the illustrations as faithful to the originals as possible, though we had to make adjustments to compensate for the fact that the video camera cannot always be placed optimally and the fact that still photos cannot record the dynamic tension that occurs in an actual high-speed stroke.

Our procedure was to first shoot all photos on high-quality 4 by 5 professional film using professional lighting to ensure clarity in the important features of the stroke. We then imported these photos into our computer through a high-resolution scanner. From there, we edited them in Photoshop to simplify the background and focus attention on key details.

The Stages of the Agassi Forehand

We will use a series of five static and four rollover illustrations as a basis for our discussions. Following is the takeback stage and a Shockwave video of the entire stroke sequence.

Takeback

In the following illustration, move the cursor on and off the image on the right to see the transition between the takeback and the next stage. This illustration shows the most important feature of the first stage, moving the racquet forward as if pulling a rope. Many amateurs begin the forward motion by some form of rotation.

Contraction

The contraction consists in dropping the racquet head and pulling the racquet forward (as noted above, by the butt, as if it were a rope) so that it is in front of the body plane (Agassi rarely allows the elbow ever to go behind the body plane). This is done by contracting the arm from the shoulder (to bring the arm into closer proximity with the body) and by bending the elbow to a degree that the racquet face becomes perpendicular to the ground (for a flat ball; for topspin, the face of the racquet may be parallel to the ground). The degree of elbow bend needed depends on the grip. For a western grip, the forearm and upper arm meet at about 90 degrees. For the eastern grip, they meet at about 120 degrees.

This is the stage that Bollettieri illustrates using a towel in his excellent video The Killer Forehand. The contraction stage should place the elbow about 9 to 12 inches from the body and in front of the body plane by about 9 inches or more to achieve a stable configuration to support the remainder of the stroke. Some exploration is needed to get the most efficient placement for a given individual.

We should mention, for the purpose of relating our work to that of other researchers, that Bollettieri hit upon this notion heuristically with the idea of placing a ball under the student's arm to force it close to the body. Once we know we are trying to achieve stability, this motion makes more sense. If the contraction stage does not impose stability, the remainder of the stroke is in essence "out of control." A further point about this movement is that stability is most important when operating at high speeds. The contraction toward the body may be relaxed at lower execution speeds.

Rotation

Once stability has been established, the second state transition may occur from contraction to rotation. The rotation stage consists in rotating the hips, shoulders, or both to better position the player to bring the racquet face into alignment with the path of the oncoming ball. At high speeds, the effect of the rotation force on the racquet arm must be countered by other forces in the muscles of the waist if it is not to destabilize the arm position.

A further subtlety is that a successful rotation (at high speeds) depends on the fact that, at the end of the contraction stage, the elbow must be in front of the body plane. If this does not occur, the rotation can destabilize the motion by causing the racquet face to shift downward, causing the ball to fall short or go into the net. The only remedy to offset this instability once it starts is to use physical body force to maintain stability at the cost of efficiency (you will begin to get tired).

Acceleration

The transition from rotation to acceleration is shown in the illustration.

The acceleration in the Agassi video from which this sequence was derived arises in part from the rotation of the upper arm. While some acceleration does occur in the previous two stages, this stage has as its main purpose to create the final burst of speed that is to be delivered to the ball. It is at this point that the player is in a position to alter or adapt the level of acceleration to fit the circumstances of the exchange. If the player is unbalanced for some reason, this final acceleration burst must be attenuated accordingly.

The acceleration is a separate stage from the rotation because it is a dynamical state change in the movement of the upper arm. In particular, the upper arm begins the acceleration through initiating a rotation that also brings the racquet into final alignment with the path of the oncoming ball.

Strike

The final transition is shown in the previous illustration. This is a state change in that this is where the motion of the racquet is rectified (straightened out) and directed into the the path of the oncoming ball to maximize the chance of a clean strike.

The Stages Have a Purpose

Each stage has a dynamical purpose (making it easier to learn) that contributes to the quality of a stroke. As mentioned above, there are four dynamical elements that can be considered: stability, alignment, acceleration, and rectification (SAAR). The methods pros use to produce SAAR are not always efficient and effective. In fact, players can be partitioned according to how they produce SAAR and how efficient and effective they are at doing this. Our analysis shows that Agassi has the most efficient forehand among all the pros.

The takeback varies widely and can be personal. Some players have a takeback that puts their elbow behind their body plane. This is just fine. However, if you choose to develop your stroke around this model (we call this the Dokic Model), you must resolve the problem of getting the elbow in front of the body plane quickly, or your stroke will become destabilized and inconsistent. A key point is that a smooth transition to the contraction stage is most efficient. Though the contraction has a primary purpose of stabilizing the remainder of the stroke, it does provide some acceleration and rectification.

The rotation provides some acceleration by definition and some stability, but no rectification. Its primary purpose is to make a partial alignment of the racquet with the path of the oncoming ball.

The acceleration stage is predominately to provide acceleration, but it does include some rectification. The strike stage is almost purely rectification.

As the stroke unfolds, the degree to which a stage contains multiple purposes diminishes until the strike, which is necessarily the simplest stage and the most important for clean contact.

It is clear that the purpose of acceleration is to generate racquet head speed, the purpose of alignment is obvious, and the purpose of stability is to provide control. So what does rectification provide? Rectification provides accuracy in placing the racquet on the ball. There are two stages that provide large components of rectification: the contraction stage and the strike stage. Both of these stages are usually absent in amateur play, and our research shows that, while they are easy to understand, most players do not have the ability to execute these stages initially, so they must be learned.

Caution: These Are Not Rules!

The foregoing analysis is not a set of rules. Each stage of the stroke contains some of the elements of stability, alignment, acceleration, and rectification. These are inseparable; hence, there is no rule that governs them. Also, our stages are approximations, not absolutes. They are organized around meaningful objectives that make sense to students and thus can be remembered and practiced when they are on their own.

Because we have formulated fairly accurate approximations of the stages of the swing, they greatly reduce learning and experimentation time. It is not uncommon for a player to develop good strokes in one-tenth the time of traditional methods.

Another point about our system is that the course of study usually starts by experimenting with the strike stage first. It can be shown mathematically that, if one starts with the last stage first and works backwards, the learning time will be further dramatically reduced. Also, by starting with the strike stage, one directs the student's attention to what is purposeful from the outset, and at each subsequent stage of development. Because the students are focused on purpose rather than rules, they learn faster. Also, because there are no rules, students are free to use their own initiative to aid in the development of their strokes. In the EASI Tennis System, the instructor transitions from being a director to being a facilitator.