Question: How OptimalFlight can improve the accuracy of launch monitor summary data?

Answer: Choose the drives most similar to each other (ex: most similar 4 of 5 shots, or most similar 3 of 4 shots).

Use the OptimaFlight’s TrueFlight Average to determine which ball flights are most similar.

Reason: Picking the best 4, or 5 or 10 shots on the basis of carry distance alone can unintentionally introduce additional variation into the launch condition average by permitting a wider range of data to come into play (ex: occasional low burners & high fliers).

Drewspin sent me launch monitor study data on seven drivers, 5 shots per driver, on Sunday November 19, 2006 in coded form (A,B,C,D,E,F,G).

Top tier driver studies would normally take 10 shots, discarding the best and worst 3 and use the remaining 4 based on carry for the final analysis. For this driver study, the most similar 4 drives of 5 was used for the final analysis. This strategy works out well because the method of averaging launch monitor ball speed, launch angle and spin data works out best with similar shots. Otherwise, the recreated ball flight using averaged launch monitor inputs will not match up very well with the 'average' carry.

All shots from Drewspin’s driver study were entered into OptimalFlight (drives 1-4) and their ball flights recreated. The initial round of ball-flight analysis replaced the most different "looking" drive of the group by drive 5. I had an interesting conversation with John Muir (Clubmaker-online.com) on a few reports. Most of the time, it is very obvious which drive's ball-flight is the odd one of the bunch. There will be occasions where the most similar 4 of 5 decision making is not a very straightforward one to use (all 5 shots similar, or not everyone will be making the same selection.) As a result of our conversation, OptimalFlight was updated to report additional ball flight details to help this decision making to be as unbiased as possible to the initial launch conditions or results.

What's different in the current release of OptimalFlight is the creation of an "in the air" average ball flight path called the TrueFlight Average. The program can now measure the difference between a ball flight and the TrueFlight average line. This difference is reported as the "Average Difference" and can be calculated for each of the 4 flights. The ball flight with the highest average difference among drives 1-4 was substituted with Drive 5. This approach eliminates any visual or individual bias in to deciding which drive is the most different "looking". The 4 drives with the lowest AvgDiff value became the one used for the final analysis.

To illustrate how this works, a closer look at a simpler example was done. 4 drives are available, and need to pick the most similar 3 of 4 drives.

12/10/2006 Update: Added the following summary table of individual launch monitor conditions for drives A, B, C, and D. We can easily observe how Flight C has a big jump in SPIN without seeing a graph and looking only at launch monitor SPEED, LAUNCH and SPIN.

A ball-flight graph of all 4 shots provides a clear picture of what’s happening. Drive C does jump out a bit because of a high launch with high spin result.

Picking the best 3 of 4 drives based on carry result alone shows that the best 3 drives for carry distance are A, C, and D. Are these 3 drives really the most similar?

OptimalFlight was used to recreate all 4 ball flights for further review. The TrueFlight Average line (smooth line, no markers) represents the drives with a * by the drive letter (which is A,B,C). Ball Flight B is the closest to the line and Ball Flight C is the furthest away from the line. (Updated the following graphs to focus only on Drives A, B, and C)

The average difference for Flight C is an average of the in-the-air flight differences from the TrueFlight Average ( purple region ). The average difference for Flight C is 4.40 yards:

The overall average difference of all 3 drives is 2.98 yards. Let’s now take a look at only ball flights A, B, and D (omitting flight C because it had the largest difference from the TrueFlight Average).

The overall average difference of all 3 drives is 1.53 yards, which is nearly a 50% reduction over the ball flight variation observed with drives A, B, and C. The in-the-air shot variation is much more tighter with Drives A, B, and D.

Question: Which set of data has the most value for club fitting decisions? We now have two options: using drives A, C, D based on best carry distance or drives A, B, D based on proximity to the TrueFlight Average line. Let's now review each set by averaging of launch condition inputs:

Note for Flight ID D now represents an average of the launch conditions (ex: Average Launch Angle of (13, 15, 14) = 14). It is no surprise to see how the AVERAGE ball flight tracks closely to Drive D. OptimalFlight computes the carry result to be 3.1 yards longer than the average carry of Drives A, C, and D.

Let's take a look at how things shape up for the other set of data based on the most similar 3 of 4 drives to the TrueFlight Average line.

OptimalFlight re-recreated AVERAGE ball flight is only 0.4 yards longer than the carry average of Drives A, B and D.

It raises a very important point: The decision to use similar drives for the final analysis leads to a more accurate summary. Picking the best 4, or 5 or 10 shots on the basis of carry distance alone can unintentionally introduce additional variation in launch condition averages. A wider range of launch conditions to come into play (ex: occasional low burners & high fliers) which can artificially elevate or dampen the results. This issue is exactly why the top tier studies focuses on the middle 4 of 10 drives (discarding the top and bottom 3 for carry) to permit a more narrower range of launch conditions to be involved in the final summary.

OptimalFlight provides basic tools to objectively see what's happening with launch monitor data and how to maximize the value of it. Averaging the best 3 of 4 carry flights produces a longer flight by 3.1 yards. It is not as accurate as selecting the most similar 3 of 4 drives (0.4 yards difference)

The following graphs shows both average flights have desirable launch conditions. Averaging the launch conditions of the most similar 3 of 4 shots produces near optimal for total distance (carry + roll) and has landing angle at a peak to maximize roll.

The differences are probably too small to be a major concern because the optimal results are comparable to each other.

A set of inconsistent drives can contain extreme differences and produce more dramatic differences than shown here. If accuracy of average results is important to you, the most similar subset of n drives is a sound strategy to follow. On occasion, it can make a difference in your dialogue with your club-fitter: whether to leaving things alone as is or make an unnecessary equipment change.

The concept of the most similar 3 of 4 drives approach was applied to drewspin's launch monitor study. Each driver had 5 shots and the most similar 4 to the TrueFlight Average were chosen for the final analysis.