Modelling the Transfers of Training Effects on Performance in Elite Triathletes
Four elite triathletes were followed for 40 weeks of swim, bike and run training, with training load calculated from exercise heart rate. A first-order transfer-function model linked each sport's training to swimming, running and triathlon-race performance.
Running performance followed running training and swimming performance followed swimming training (specificity); cycling training also carried over to running performance, but nothing carried over to or from swimming. Running training had a major effect on triathlon race performance.
Participants
n=4 elite triathletes (abstract: "four elite triathletes"). Sex, age and competitive record are not given in the abstract.
Protocol
40 weeks of normal swim, bike and run training; training stimulus calculated from exercise heart rate.
Performance: running = a submaximal 30 min run; swimming = a 5 x 400 m all-out test; triathlon competitions rated subjectively.
Model: one to three first-order transfer functions linking actual to modelled performance, fitted by minimising the residual sum of squares.
Key Results
| Training → performance | Time constants (abstract) | Fit |
|---|---|---|
| Running → running | tau(1) = 20; tau(2) = 10 | r = 0.74; p < 0.001 |
| Swimming → swimming | tau(1) = 31 | r = 0.37; p = 0.03 |
| Cycling → running (cross-transfer) | tau(1) = 42 | r = 0.56; p < 0.001 |
| Running → triathlon competition | tau(1) = 52; tau(2) = 4 | r = 0.52; p < 0.001 |
| Swimming ↔ cycling or running | — | no cross-transfer identified |
The abstract does not state the unit of the time constants.
Practical Application
In four elite triathletes modelled over 40 weeks, cycling training carried over to running performance (r = 0.56), running training drove triathlon race results (r = 0.52), and swimming neither gained from nor gave anything to the other two sports. Bike hours help your run a little, but swim fitness only comes from swimming, and run training is not something the bike can replace.
When hours are short, do not cut the swim on the theory that bike or run fitness will cover it: in this model nothing crossed over into swimming.
Inference, not tested here: when running volume is limited (injury, impact tolerance), extra cycling is the discipline that showed a carry-over to running; the paper modelled transfer, it did not test a substitution.
The swimming model fitted poorly (r = 0.37) compared with running (r = 0.74), so the swim result is the least certain part.
Not from this paper; the base's own practice: how to split hours for sprint and Olympic distance is in SYNTH-52.
Limitations
n=4: a case series, not a sample from which a general rule can be drawn.
Correlations between modelled and actual performance, not a controlled comparison.
Training load by heart rate; triathlon performance judged subjectively; running performance from a submaximal test.
Time constants given without units in the abstract.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level
elite case series, n=4, modelled correlations