More than one ground simulation per run in the track: which markers your dry session has to produce, how to measure them — and where the transfer ends.
Most of your skeleton start training happens without ice — and that is not a workaround, it is the documented normal case. On one athlete's way to the Olympics, roughly 300 start simulations on the ground stood against roughly 220 runs in the ice track [1]. That is more than one simulation per run. If you are waiting for ice time to work on your start, you are waiting for the wrong resource.
The reason lies in the structure of the start itself. Biomechanically it is a hybrid of sprint acceleration and the dive — the deliberate load onto the moving sled [2]. The sprint half can be reproduced on the ground almost completely. The dive half only partly. Drawing that line honestly is the difference between a useful simulation and busywork.
This guide shows you which task a start simulation actually has to copy, how to measure it — and where the transfer ends.
Roughly 300 simulations against roughly 220 runs [1] means in practice: your start is built on the ground and only retrieved on the ice. Treat the dry session as a warm-up for the "real" work and you invert the ratio — training the most important thing least often.
Copy the requirement, not the look. The start analysis gives you concrete reference values: start velocity is around 6.97 ± 0.42 metres per second for men and around 6.25 ± 0.58 for women — roughly 22 to 25 km/h, reached with one hand on the sled [2]. The step-length profile is characteristic too: the push-off step measures around 1.60 ± 0.14 metres, the following step around 2.09 ± 0.18 metres [2] — so the start opens up rather than running evenly. And peak horizontal force sits around 827 ± 217 newtons, roughly the weight of 85 kilograms that you push forwards instead of downwards [2]. Those are the markers your simulation has to produce.
The established diagnostic protocol is simple: a 40-metre sprint with splits at 5, 10 and 20 metres, started from a three-point start, because that mirrors the one-handed skeleton start [4]. Plus video: markerless motion analysis has shown that push technique can be captured objectively in the field [5]. Without splits you only know that it felt fast.
What trains well on the ground: approach, acceleration steps, step frequency, running with one hand on the sled, direction of force. What only partly transfers: the dive — loading onto an already moving sled, with ice friction, runner behaviour and the position you end up in [2][3]. The competition start takes you past 40 km/h within seconds over the first metres [3]; that dynamic cannot be fully reproduced dry. Plan it as an ice task, not a dry-land goal.
If ground training carries the sprint half, your ice time has a clear job: dive timing, the transition into the driving position and everything after it. Do not burn a run on something you could practise in a car park. That principle is what makes the documented ratio of simulations to runs meaningful in the first place [1].
Start simulations look harmless and are not: you sprint bent over, guided with one hand, delivering high force forwards — that loads the ankle, the Achilles tendon and the lower back. Build intensity over weeks, not days, and train simulations fresh at the start of a session, never as a tired add-on.
Anything involving a real sled on the ground belongs in your club's hands: setup, securing the run-out area and the sequence are site-specific and not transferable from a text. And the dive itself — loading onto a moving sled — is learned exclusively under supervision [3], not alone on a field.
Measure your start simulation twice within the same session: once fresh after the warm-up, once at the end. The difference between your 20-metre splits [4] is your personal fatigue index — and it tells you more about competition readiness than any personal best. Because in a race what counts is not your best push but the one you can still deliver in the third run of the day.
Through start simulations on the ground: approach, acceleration steps and one-handed guidance, measured over a 40-metre sprint with splits at 5, 10 and 20 metres from a three-point start [4]. The goal is to produce the markers of the competition start — start velocity, step-length profile and horizontal force [2].
In the documented case more than one to one: roughly 300 start simulations against roughly 220 runs in the track [1]. That is not a prescription for every athlete, but it shows the magnitude — start training is mostly ground work.
Above all the dive, that is, loading onto the already moving sled including ice friction, runner behaviour and the position afterwards [2][3]. The competition start takes you past 40 km/h within seconds over the first metres [3] — that dynamic cannot be fully reproduced on dry land.
There is no published dosing recommendation. Only the long-term magnitude is documented: roughly 300 simulations on the way to the Olympics [1]. In practice, few very clean repetitions in a rested state have proven useful — that is experience, not study evidence.
By numbers, not by feel. Splits at 5, 10 and 20 metres show whether you accelerate early or late [4]; video shows step rate, angles and trunk position, which can be captured objectively in the field [5]. The target picture comes from start velocity and the step-length profile in the start analysis [2].
Start Faster: Sprint Power, Horizontal Force and 30-Metre Benchmarks