Matteo Campagna
ITEN
Menu
Menu

Research line

From the total on the scorecard to the arrow's coordinates

The score is the one measure almost every study of archers reports, often without setting out to. This page follows, in chronological order, how research learnt to read it: from the end-of-day total of 1966 to points per arrow, to the width and centre of the group, and on to the coordinates a camera tries to read by itself.

Studies in the registry
102
On archers
87
Years
1966-2026

A measure nobody chose

Almost every study of archers reports a score, and almost none sets out to study it: the score describes the sample, splits the groups, counts the effect of a programme. The question that holds this line of research together is a plain one: what does the number on the scorecard measure about an archer? This page follows, year by year, how research has changed its reading of that number. The practical questions sit in the articles on how many points it takes to say one archer is better and on what the qualification ranking is worth.

1966-1992: the round total

It begins in 1966. Fifty-eight beginners shot twenty-four arrows from each distance, and the total halved between twenty and forty yards, from about 142 to 70 points. Even then the number depended on distance, not in proportion: from ten to twenty yards they lost fifty-eight points, then about twenty for each step.

The first comparison between levels comes in 1986: 188 US archers grouped by annual FITA average, 1,272 points at world level, 1,221 at upper national-collegiate level, 1,157 at collegiate level and 1,054 in the weakest group. Alongside the mean sat a second number, the standard deviation, which grew from 21 to 38 points going down the levels, except the national-collegiate group, the most compact at 14. The top, though, has no fixed number: the fifteen highly skilled archers measured in 1990 had an annual average of 1,211, below the 1986 world-level group, while in 1992 the same lead author, with another sample, gave 1,300 to 1,350 as the scores that normally win an international event.

The mean and the spread

Later samples showed that the score carries two pieces of information. In the Greek national team the standard deviations of competition scores were 27 to 40 points, against 69 to 77 in non-selected archers. In 1999 twenty young Swedish archers gave the first direct measure of the link: score and the standard deviation of ends correlated at minus 0.65. The link is partly geometric, since an archer shooting close to the ten has little room above, and the entry on consistency and level remains reported.

Meanwhile the step between levels got a size. In Turkey ten national team archers and ten club archers scored 1,296 and 1,322 against 1,157 and 1,147 points, a gap of about 150. The entry on the gap between club and national level has been corroborated since 2003, but across different formats and eras those points are an order of magnitude.

From the round to the arrow

In 2011 the unit of measurement moved to the single arrow. At an indoor competition for adolescents, across a thousand arrows, shot time and heart rate variability before each arrow together explained 11.8% of the single arrow's score, while in a separate analysis the difference between beginners and experienced archers weighed about twice as much. A year later a Korean study showed that at a fixed fifty metres the mean per arrow rose with the diameter of the face, from 7.7 on 60 cm to 8.93 on 100 cm: points per arrow depend on how large the target looks from the line. In 2016 eighteen expert archers shot from eighteen or twenty distances, and the decline was linear, with r squared around 0.90 and, at competition distances, a single departure, seventy metres in men; in official Chinese competitions with 3,196 archers the score drops by 0.0286 points per arrow for every extra metre. This is the entry on scores falling with distance; the article on why groups open up explains the geometry.

Read per arrow, the score becomes a common currency, provided distance and face are stated. In 2018 thirty Japanese archers grouped by ranking scored 633, 593 and 557.5 out of 720, and the entry on points per arrow and level has filled up with bands measured one group at a time. From 2023 the question shifted to archers who compete well below the top. Sixty Indian academy archers started at about 7.4 points per arrow, and in 2026 five independent samples in four countries landed between 5.9 and 7.65, so that the entry on the mean score sitting around seven is corroborated. Even eight French recurve archers in the national top twenty, at seventy metres, shot 163 arrows of eight or less out of 549.

The group: width and centre

An older path measured the group instead of the points. In 1990 three Olympians and six Belgian national-level archers shot at eighteen metres: the group outline spanned 7.8 to 10 cm in the former and 11.9 to 26.7 cm in the latter. In 2014 twenty-nine recurve archers at three levels had the coordinates of 72 impacts each recorded: the beginners' spread was five to seven times that of the high level, but the most off-centre group was the intermediate one, so centring does not improve in an orderly way.

In 2020 the group was measured in competition. Photographs of every target face at two major 2013 events, reanalysed with a model, show that in men's compound the top twenty lost 0.30 points over 36 arrows to off-centre groups, against 1.35 for the division average. The entry on tighter, better centred groups has seven samples in agreement, all cross-sectional. What its position says about the shot is covered in the article on impact patterns.

Coordinates read by a machine

Someone, though, has to plot the coordinates. In 2012 ten operators with no archery experience took them from photographs of 72 arrows shot at 70 m, correlating at 0.977 or more with the measured positions, with four one-point errors in 720 arrows. A year later an algorithm assigned the correct zone in every sequence from four archers at 10 m indoors. Then the limits appeared: on torn faces a model trained on smartphone photos dropped to around 40% in recall and precision, and in the first field test checked against the judges, on a hundred arrows from ten university archers, the system detected 91 and scored 67 correctly. The most accurate validation, from 2026, locates 5,084 holes with an error of 1.41 mm, but mostly misses arrows within one shaft diameter of each other, exactly where a judge is most needed. The entry on automatic scoring from images is contested, and the method is described on the page on machine learning and computer vision.

Predicting the score

The last question asks how much of the next score is already in the data. In 2025 a model trained on one session from thirty-two Indian archers and tested on the next had an error of 6.26 points out of 300; among the predictors, years of practice came first and past competition scores third, while heart rate around the release came sixth and tenth. Two small Indonesian groups show an internal ranking that changes little: 0.898 between two very different training periods, 0.985 over six weeks. The entry on predicting the score is reported.

Where the line is thin

Almost everything comes from cross-sectional comparisons between groups, and nobody has followed the same archers as their level changed. Many scores come from regional or questionable journals, in formats that do not line up. The machines have never been tested outdoors at 70 m or over a full round, and no prediction model has been compared with the simplest prediction of all, the average of the last few competitions.

How we came to know it

  1. 1966OriginSinger 1966 Transfer Effects and Ultimate Success in Archery due to Degree of Difficulty of the Initial LearningFifty-eight beginners, twenty-four arrows per distance: the total halves between twenty and forty yards, from about 142 to 70 points. The measure is the total, and distance shows at once that it is not absolute.
  2. 1986ExtensionLanders et al. 1986 A Psychobiological Study of Archery PerformanceOne hundred and eighty-eight US archers grouped by annual FITA average, from 1,272 to 1,054 points. Alongside the mean comes the standard deviation, from 21 points at world level to 38 in the weakest group.
  3. 1990ExtensionClarys et al. 1990 Muscular Activity of Different Shooting Distances, Different Release Techniques, and Different Performance Levels, with and without Stabilizers, in Target ArcheryFor the first time the group is measured instead of the score: at eighteen metres its outline spans 7.8 to 10 cm in the three Olympians and 11.9 to 26.7 cm in the six national-level archers.
  4. 1991ReplicationZervas et al. 1991 Visuomotor Behavior Rehearsal in Archery Shooting PerformanceIn the Greek national team the standard deviation of competition scores is 27 to 40 points, in non-selected archers 69 to 77. A second sample in which spread follows level.
  5. 1999ExtensionNorlander et al. 1999 Primary process in competitive archery performance: Effects of flotation RESTTwenty young Swedish archers: score and the standard deviation of ends correlate at minus 0.65. It is the first direct measure of the link between mean and consistency.
  6. 2003ReplicationErtan et al. 2003 Activation patterns in forearm muscles during archery shootingIn Turkey ten national team archers and ten club archers sit about 150 FITA points apart, with twice the years of training. The 1986 step reappears in another country.
  7. 2011ExtensionCarrillo et al. 2011 Autonomic nervous system modulation during an archery competition in novice and experienced adolescent archersThe unit becomes the single arrow: across a thousand arrows at an indoor competition, shot time and heart rate variability explain 11.8% of the score, less than level does.
  8. 2012MethodCallaway et al. 2012 Inter-Rater Reliability and Criterion Validity of Scatter Diagrams as an Input Method for Marksmanship Analysis: Computerised Notational Analysis for ArcheryTen operators with no archery experience plot 72 arrows shot at 70 m from photographs: their coordinates correlate at 0.977 or more with the measured ones, with four one-point errors in 720 arrows.
  9. 2012ExtensionLee et al. 2012 An Archer's Perceived Form Scales the "Hitableness" of Archery TargetsAt a fixed fifty metres the mean per arrow rises with face diameter, from 7.7 on 60 cm to 8.93 on 100 cm. Points per arrow depend on how large the target looks from the line.
  10. 2013MethodZin et al. 2013 Image Processing Approach to Automatic Scoring System for Archery TargetsThe first algorithm to recognise arrows in images and assign the scoring zone: four archers at 10 m indoors, the correct zone in every sequence tested.
  11. 2014ExtensionErtan 2014 Hit distribution patterns in recurve archeryTwenty-nine recurve archers at three levels, coordinates of 72 impacts each: the beginners' spread is five to seven times that of the high-level group, and the most off-centre group is the intermediate one.
  12. 2016ReplicationCzyż et al. 2016 Specificity vs. Generalizability: Emergence of Especial Skills in Classical ArcheryEighteen expert archers shoot from eighteen or twenty distances: the score falls linearly, with r squared around 0.90, and at competition distances there is a single departure, seventy metres in men.
  13. 2018ExtensionSerrien et al. 2018 The Uncontrolled Manifold Concept Reveals That the Structure of Postural Control in Recurve Archery Shooting Is Related to AccuracySix recurve archers from a national team, ninety arrows each: the best shoots no six and no seven, the weakest fourteen sevens and a six. Level shows in the tail of the poorest arrows.
  14. 2020ExtensionPark 2020 The impact of off-centred groups on an archer’s scorePhotographs of every target face at two 2013 events, reanalysed with a model: in men's compound the top twenty lose 0.30 points over 36 arrows to off-centre groups, against 1.35 for their division.
  15. 2021ExtensionQiu et al. 2021 Adaptation and the distributional effects of heat: Evidence from professional archery competitionsOfficial Chinese competitions with 3,196 archers: the score drops by 0.0286 points per arrow for every extra metre. It is the first count of the distance effect made on real competitions.
  16. 2023ContradictionMa et al. 2023 Automatic Archery Scoring System Using Deep Learning and Image ProcessingA model trained on smartphone photos finds every arrow on clean faces and drops to around 40% on torn ones. The limit of automatic reading now has a figure.
  17. 2023ExtensionChatla et al. 2023 Isolated and Combined Effect of Strength Training and Specific Strength Training on Performance Levels Among ArchersSixty Indian academy archers with national competition experience start at 531 to 536 points out of 720, about 7.4 per arrow: the first full round in the seven-point band.
  18. 2025ContradictionAdam et al. 2025 Real-Time Arrow Detection and Scoring on Archery Targets Using YOLOv8 with Euclidean Distance-Based Zone EstimationThe first field test checked against the judges' scores: ten archers, a hundred arrows, 91 detected and 67 scored correctly.
  19. 2025ExtensionGuru et al. 2025 Can we predict archery performance using shooting heart rate and other biophysical parameters? A machine learning approachThirty-two Indian archers, a model trained on one session and tested on the next: an error of 6.26 points out of 300, with years of practice and past competition scores ranked above heart rate.
  20. 2026MethodShepherd 2026 Frozen Vision Transformers for Dense Prediction on Small Datasets: A Case Study in Arrow LocalizationAcross 5,084 holes on 40 cm faces the localisation error is 1.41 mm and the error on the mean score 1.8% at the median; the arrows closest together are still the ones missed.
  21. 2026ReplicationPrastyawan et al. 2026 From play to performance: effects of fun-based circuit training on muscular strength, endurance, and competitive performance in Indonesian archery athletesThirty-eight competitive club archers at about 7 points per arrow, with single arrows scattering 2.4 to 2.5 points around the mean. One of the year's five samples below eight.
  22. 2026ExtensionJacquot 2026 Analyse multidimensionnelle de la performance en tir à l'arcEight French recurve archers in the national top twenty at seventy metres: 163 of 549 arrows score eight or less. Even at the top of a national ranking, almost one arrow in three falls below nine.

Studies in this line

Related claims

Claims in the registry with at least one of these studies among their evidence.

252 claims · 90% with evidence on archers

Articles

Methods used

Research lines

Cite
APA

Campagna, M. (2026). From the total on the scorecard to the arrow's coordinates. In Research. matteocampagna.com. Retrieved 16 October 2026, from https://matteocampagna.com/en/research/score-as-a-measure-of-performance

BibTeX
@misc{campagna2026_enresearchscoreasameasureofperformance,
  author = {Campagna, Matteo},
  title = {From the total on the scorecard to the arrow's coordinates},
  year = {2026},
  howpublished = {Research, matteocampagna.com},
  url = {https://matteocampagna.com/en/research/score-as-a-measure-of-performance},
  note = {accessed 2026-10-16}
}