Prognostic Times

Prognostic Times

Prognostic times (PTs, or Gold Standards) are widely used in rowing communities for various purposes, such as: 1) Defining training intensity; 2) Evaluation of performance in various boat types; 3) Selection of rowers and crews; 4) Benchmarking and setting targets for short and long-term rower’s development; 5) Motivation for rowers and coaches to achieve the highest standards.

Usually, PTs are developed at National Rowing Federations (NRFs). Our Google search reveals only two open sources (Table 1) from Australia (1) and Canada (2), which means that most of NFRs keep their PTs in secret. All published NRF’s prognostic times are anonymous and methods of their development were not disclosed, so it is likely that they are subjective, which creates the following problems: it is not possible to discuss PTs to make them more accurate and eliminate discrepancies between them, which may lead to errors in crews selection. Usually, NRFs PTs are set close to or higher than World Best Times (WBTs), which are unrealistic to achieve without very fast weather conditions.

In an attempt to find an objective method of PT calculation, two different methods could be used:

1. Statistical method using trends of speed of the winners of World regattas. Advantages: Relatively simple and obvious, does not require biomechanical measurements. Disadvantages: Low reliability due to the effect of random weather conditions.

2. Biomechanical modelling using rowing power and drag factors in various boat types. Advantages: Ability to model rowing speed for various rower’s power/weight ratios and various weather conditions. Disadvantages: Requires additional complex measurements, uncertainty of the real power production in a race (usually, only erg scores are available) and the effect of weather.

Here is a description of both methods to make them available for others to produce their own objective PTs (say, for juniors, U23 and other rower’s categories), as well as our BioRow PTs-2024 for elite rowers. Analysis of speed trends with method 1 consists of two steps: 1) Data filtering, 2) Selection of the most adequate trend type.

Step 1. The data must be filtered to exclude outliers, mainly on the slow side, and the filtering range must be chosen adequately: e.g., in W2x (Fig.1), filtering data within ±1.5SD range with exclusion of only one (the slowest) data point in 2016 (1) improves the PT by 9s, from 6:49 down to 6:40 and increases the significance from 8% up to 31%.

Step 2: An adequate trend type should be selected, which is usually a linear, 2nd and 3rd order polynomial trend. E.g., in W2x, using a linear or third order polynomial trend would give PT2024 6:40 (Fig.1), but using the second order trend would give 6:49. In M2x, using a linear trend would give PT2024 6:03, the second order – 6:06, the third order – 5:57. The following rule can be used to make the selection of trends more objective: if two trend types produce a similar PT, they should be used for this boat type (as was in W2x above); if all three trends produce different PTs (in M2x), the median PT should be selected.

Biomechanical modelling (method 2) is based on an equation relating the prognostic speed V with rowing power P and drag factor DF in a specific boat and rower’s mass:

V = (P / DF)1/3                                                  (1)

Rowing power P could be derived from the known erg score of the rower(s). However, it is not a fact that this exact amount of power is really produced in the race on-water, especially in big sweep boats, which is a limitation of the method. Drag factors DF in various boat types were derived from the BioRow database of biomechanical measurements (n>30k), usually in a tail wind. Two types of DF were identified (RBN 2015/04): DFnet , which excludes blade and boat velocity efficiencies, and DFgross, which is the ratio of rowing power to the cube of the boat speed. DFgross was used here for simplicity and derived as a linear function of the average rower’s mass.

 Table 1 shows that the most of NRFs PTs are set at 99.8-99.9% of WBT speed level, while BioRow PTs are derived at average speed 98.6% (trends) and 99.0% (model) of WBT, which is closer to the real results of the winners of World regattas. In some boats (M2-, W1x, W2x, W8+), very good correspondence was found between the trends and the model. Other events have abnormally low (M1x, M4-) or high (W2-) PTs derived from the trends.

Hopefully, the development of prognostic times will become more open and objective in a future, assisted with attempts from FISA to bring more information to the rowing community about weather conditions during World regattas, and possibly about other indicators of rowing performance.


1.                 Australian Gold Standard times.

©2020 Dr. Valery Kleshnev