Predicting the results of the 2004 olympic games in selected track and field sports

Introduction

The method of extrapolation is used in biomechanical filtering of kinematic data when IIR type filters (such as Butterworth) are used in order to overcome the endpoint error. We applied this methodology to predict whether we should expect better results in specific track and field sports during the 2004 Olympic Games in Athens via a mathematical model.

Methods

The method of linear prediction was used [1]. All the data of consecutive Olympic games were collected via the official site of the International Olympic Committee. Prior to the prediction of the 2004 values the model parameters was adjusted in order to predict most of the outcomes of the 2000 Olympic Games (far example Graph 1). Pilot results indicated that small deviations of the values of the model parameters do not affect the outcomes.

Results

The model predicted more than 70% of the outcomes for the 2000 Olympic Games. An example of the men’s 100m output is presented in Graph 2. The model indicated that we could expect better results compared to the previous Olympic Games in the sports of (where * indicates a possibility for an Olympic record) 200m, 400m*, 800m, 110m*, 4x400m, Discus*, High Jump, Pole Vault and Triple Jump for men. Similarly the 100m, Discus and High Jump for women.

Discussion / Conclusions

The prediction of results in future games is always a risky attempt and this study was based on one factor only which was the previous winning results. We should wait until September 2004 to confirm/reject the above findings.

References

1. Giakas et al (1998). Journal of Biomechanics 31(1), 87-91.