Fair Decathlon Model. Part 2: How Much Is an Advantage Worth? (1)

From Mayer, Eaton and Warner to Osaka 2007 and Roma 1960

July 2026

Abstract

The first article on the Fair Decathlon Model (FDM) described a broad pattern: throwing events generally gained points, jumping events generally lost points, and running events changed less. A reader then raised an apparent contradiction. If Kevin Mayer is the strongest thrower among the leading 9000-point decathletes, why do Ashton Eaton and Damian Warner gain more under FDM? The answer is that the overall scoring level of an event and the point differences between athletes are two separate issues. FDM recalibrates both. This follow-up examines the official point spreads at three representative performance levels, then applies them to five head-to-head comparisons: Mayer versus Eaton, Mayer versus Warner, Eaton versus Warner, Maurice Smith versus Roman Sebrle at Osaka 2007, and Rafer Johnson versus Yang Chuan-kwang at Roma 1960. The examples show that the number of events won is not the essence of the decathlon. What matters is the combined value of ten performances and the size of the advantages created in each event.

1.  The question raised by the first article

Thomas noted that the first article appeared to contain a contradiction. FDM generally gives more points to throwing performances, yet Mayer falls from first to third when the personal best decathlons of Mayer, Eaton and Warner are recalculated. Eaton and Warner, commonly viewed as sprinter-jumpers, each gain more than 100 points, while Mayer loses 28.

The contradiction disappears once two different properties of a scoring table are distinguished:

• Nominal scoring level: how many points a representative performance is worth.
• Point spread: how many points separate representative performance levels within the same event.

An event may be scored too low overall while its internal differences are too large. Conversely, an event may be scored generously at ordinary levels while its point curve is too flat to reward exceptional performances adequately. This distinction is central to understanding FDM.

2. The official scoring environment

The published version of FDM uses the average 10th-best performance in each event as an 885-point anchor and the average 75th-best performance as a 779-point anchor. The average 140th-best performance is an independent control level. Before looking at individual athletes, it is useful to examine how the official tables value these same performance levels.

Table 1. Official points for the average 10th-, 75th- and 140th-best performances across the 21 calibration seasons. Red cells show point spreads below the all-event average; green cells show spreads above it.

Colours indicate the magnitude of the official point spread, not the scoring level of the event itself.

2.1 Two different imbalances

The table reveals an important distinction. At the 10th-best level, the three throwing events are worth only 820 to 829 official points, while 110 m hurdles is worth 955 and long jump 962. This is the familiar reason why combined-events followers often say that the throws are undervalued: they are referring to nominal values.

However, the point spreads tell a different story. Between the 10th- and 75th-best levels, the all-event average is 106 points. The official spreads are only 86 in the 100 m, 84 in the 400 m and 80 in the 110 m hurdles. These three sprint events are consistent with one another, but the entire group is compressed relative to the decathlon as a whole.

The jumping events do not form a uniform group. Long jump and high jump are close to the all-event average, at 103 and 111 points. Pole vault is the clear outlier at 134. The throwing events are even less uniform: shot put is relatively compressed at 95, discus is expanded at 120, and javelin is strongly expanded at 138.

The same pattern remains visible over the wider 10th-to-140th range. The sprint spreads remain low, long jump and high jump remain close to average, pole vault and javelin remain very high, and discus remains moderately high. The 1500 m is close to average between the 10th and 75th levels, but becomes relatively steep in the lower part of the range.

2.2  A striking comparison

The nominal imbalance can be seen without any formula. The average 140th-best 110 m hurdles performance, 15.49, is worth 791 official points. That is 12 points more than the average 75th-best performance across all ten events, and almost exactly the same as the average 10th-best 1500 m performance, which is worth 793 points.

The average 10th-best performances in shot put, discus and javelin are worth only 820, 829 and 828 points. They are therefore only 29 to 38 points above the average 140th-best hurdles performance. FDM addresses both parts of this problem: how much each representative level is worth and how much the difference between levels is worth.

6.  What is the essence of the decathlon?

A decathlon is not a match-play contest consisting of ten separate wins and losses. No athlete receives one point for winning an event and zero for losing it. The objective is to produce the strongest possible combination of ten performances.

The number of events won can make for a powerful narrative. A 3-7 or 6-3-1 score immediately attracts attention. But it does not measure the overall quality of the ten-event performance. A narrow win and a dominant win cannot have the same value. The scoring system must estimate how large each advantage really is and do so consistently across all ten events.

This is the common denominator linking Johnson and Yang, Mayer, Eaton and Warner, and Smith and Sebrle:

• Johnson loses more events but creates much larger advantages in the events he wins.
• Mayer, Eaton and Warner split their strengths differently, and two of their pairings are exactly 5-5.
  • Smith wins more events than Sebrle, but the official tables do not value the size of their respective advantages consistently.

The purpose of FDM is not to select a preferred athlete type. It is to ensure that equivalent performance levels and equivalent differences between athletes are rewarded consistently across all ten events. A sprinter-jumper, a sprinter-thrower, a balanced decathlete or a more polarised athlete should all be able to win when their complete set of ten performances is genuinely the strongest.

7.  Conclusion

Thomas's question exposed an important gap in the first presentation of FDM. Saying that throwing events generally gain points does not explain how the point differences between athletes change. The official tables contain two separate forms of imbalance: differences in nominal scoring levels and differences in the steepness of the event scales.

The official sprint scales are internally consistent but too compressed relative to the decathlon as a whole. Long jump and high jump are close to the average point spread, while pole vault is much more expanded. Shot put, discus and javelin have low nominal scoring levels, but their internal spreads range from relatively compressed to strongly expanded. These facts explain why Eaton and Warner can gain more than Mayer, why Smith can overtake Sebrle, and why Johnson can still remain ahead of Yang.

The head-to-head examples also show that FDM does not rewrite every result. Eaton's advantage over Warner remains almost unchanged. Johnson's original Olympic margin is reproduced almost exactly. The model changes results only when the valuation of the underlying advantages changes materially.

FDM remains a developing proposal rather than a final scoring table. Further criticism and historical examples are therefore welcome. The questions raised by Thomas and S. Steele have already led to a more precise understanding of what the model is attempting to measure: not which athlete wins the most events, but which athlete produces the strongest total decathlon.

Data and calculation notes

• The published FDM coefficients from the first article are used throughout this article.
  • For modern examples, official event points use the 1985 scoring tables. Individual event points are truncated to whole numbers before summation.
  • Roma 1960 official totals and the retrospective 1985 totals follow the Decathlon2000 results listing supplied for this analysis. Only the 1960 totals determined the Olympic medals.
  • The 10th-, 75th- and 140th-level performances are 21-season averages from the calibration dataset described in the first article.
• ASCII spellings are used for names in tables for consistency with the project dataset.
Read part 1

Explore the foundation of the Fair Decathlon Model (FDM) in Part 1: 
Are the Current Decathlon Scoring Tables Properly Balanced?

Rafał Snoch for Decathlon 2000