AthletePro Super Record User Guide

01Introduction: what the platform does and why

AthletePro Super Record turns three ingredients —the athlete’s physiology, the real course and race-day weather— into a segment-by-segment power plan the athlete can actually hold. Unlike an intuition-based strategy, every watt-by-watt target is derived by solving the physics equation of cycling and optimizing how the anaerobic reserve is spent.

The mechanical power needed to move at a given speed decomposes into three terms: aerodynamic drag, rolling resistance and the gravitational component (climbing or descending). Formally:

P = ½·ρ·CdA·v³ + Crr·m·g·cosθ·v + m·g·sinθ·v

Where ρ is air density (a function of altitude, temperature and humidity), CdA the effective aerodynamic frontal area, Crr the rolling-resistance coefficient, m the total mass (athlete + bike + gear), g gravity, θ the gradient and v the speed relative to the air. The aerodynamic term grows with the cube of speed, which is why it dominates above ~30 km/h and why CdA gains are so valuable on the flat and in time trials.

This guide follows the natural workflow: first define the engine (the athlete), then the vehicle (the bike), and finally build and tune the plan over a specific course. We recommend reading it once from start to finish before creating your first plan.

02The athlete and their physiological parameters

The athlete is the engine everything is computed on. Under “Athletes → New” you enter the anthropometric data and the performance parameters. Their accuracy determines the validity of the whole plan: an error in critical power propagates to every segment.

Anthropometrics and position

Body mass combines with bike and gear mass to give the total system mass, which enters the rolling and gravity terms. Height and, above all, riding position set the body’s CdA: a time-trial position lowers frontal area versus riding on the drops, and that difference can be worth minutes in a TT.

Critical Power (CP) and anaerobic reserve (W′): what they are

Critical Power (CP) is the highest power the athlete can sustain in a metabolic steady state, i.e. without lactate and other metabolites accumulating progressively toward exhaustion. Physiologically it marks the boundary between the heavy and severe intensity domains, and it is a better descriptor of functional threshold than estimates based on a single 20-minute test.

W′ (read “W-prime”) is the finite amount of work —in joules— the athlete can do above CP before exhaustion. It is a reserve: it depletes when riding above CP and recharges when riding below. Think of it as a fixed-capacity “anaerobic battery.” A sprint, a short climb or an attack spends W′; a descent or an easy stretch recovers it.

The relationship between power and time-to-exhaustion follows a two-parameter hyperbolic model:

P(t) = W′ / t + CP (equivalently: t = W′ / (P − CP))

The key intuition follows: the closer to CP you ride, the longer the effort lasts; above CP, time-to-exhaustion depends on how much W′ remains. The platform’s plan manages exactly that W′ budget across the course.

Why entering CP and W′ correctly is critical

The whole power plan is calibrated around CP and optimized by spending W′ down to the margin you set. If CP is overestimated, the plan will ask for watts the athlete cannot hold and they will blow up before the finish; if underestimated, the plan is conservative and leaves time on the table. A 5% error in CP is not a 5% error in the result: because the model is non-linear, it can be the difference between finishing with reserve and cracking on the final climb.

How to obtain CP and W′: multi-point fit vs. manual entry

The rigorous method is the multi-point fit. You record several maximal efforts of different durations —typically in the 2–3 to 12–20 minute range (e.g. 3, 5 and 12 min)— done rested and under comparable conditions. From those (mean power, duration) pairs the platform fits the hyperbolic model and estimates CP and W′ simultaneously. The more points, and the better spread, the more reliable the separation between the two parameters.

• Use genuinely maximal efforts: a sub-maximal test shifts the curve and biases CP.
• Spread the durations: two very similar efforts (e.g. 5 and 6 min) do not separate CP from W′ well.
• Same conditions: rested, same position, with a calibrated power meter.

Manual entry is the alternative when no recent tests are available: you input CP and W′ directly from the coach’s knowledge, historical data or an estimated threshold power. It is fine to get started, but refine it with a multi-point test as soon as possible, because manual entry carries the bias of whoever enters it.

The platform also uses a three-parameter model that incorporates instantaneous maximum power (Pmax), improving the fit for very short efforts and preventing the model from predicting unrealistic sprint powers.

Rule of thumb: create the athlete with a reasonable manual estimate to start working, and schedule a multi-point test in the first weeks. Re-enter CP and W′ after each significant block; the plan is only as good as these two numbers.

03Build the bike in the garage

The bike contributes three quantities to the model: its share of CdA, the rolling-resistance coefficient (Crr) and weight. Under “Garage → New bike” you assemble it from real components and the platform aggregates their effects.

Components and what each contributes

• Frame and cockpit: set the geometry and, together with the athlete’s position, most of the CdA.
• Wheels: rim depth and a rear disc affect CdA, especially in crosswinds (yaw). Their weight matters on climbs.
• Tyres and pressure: set the Crr. The gap between a fast and a slow tyre can equal tens of watts at race speed.
• Drivetrain: its mechanical efficiency subtracts a small but real fraction of the delivered power.

CdA, Crr and weight: why they matter together

CdA governs the aerodynamic term (∝ v³) and therefore rules on the flat and in time trials. Crr governs the rolling term (linear in v) and weighs relatively more at low speed and on poor surfaces. Total weight enters rolling and, above all, the gravity component: it is decisive uphill, where speed drops and the aerodynamic term loses prominence. A good plan recognizes that the “fast” bike depends on the profile: aero for the flat, light for the mountains.

If you have data from real rides, the Lab can estimate CdA using the Chung method (energy regression over a loop), replacing a guess with a measurement.

04Measuring CdA: the aerodynamic laboratory

CdA is by far the most influential parameter in the model and the least well known. At time-trial speeds the aerodynamic term accounts for 80–90 % of the power, so a 10 % error in CdA translates almost entirely into an 8–9 % error in the prescribed power. Estimating it from a catalogue is fine to get started; measuring it on the actual athlete, in their position and with their equipment, is what turns the plan into a prescription.

The virtual elevation method (Chung)

The laboratory implements Robert Chung’s method, also known as virtual elevation. The idea is to invert the power equation: given measured power, speed, mass and Crr, the instantaneous gradient can be solved for and, by accumulation, an elevation profile reconstructed. That reconstructed profile —the virtual elevation— only matches the real one if the CdA used is correct.

Δh = ( P·η/v − ½·ρ·CdA·v³/v − Crr·m·g·v/v − m·a ) / (m·g)

The procedure is therefore to sweep CdA values until the virtual elevation reproduces the real one. Too low a CdA makes the profile drift upward; too high, downward. The correct value is the one that leaves the residuals flat.

How to run the test

• Choose a closed loop, flat or gently rolling, with little traffic and no stops: the method requires returning to the starting point so the elevation closes.
• Repeat several laps at steady speed in the position you want to characterise. Each position (drops, extensions, TT) is a separate measurement.
• Do it in calm air. Wind is the method’s main source of error on the road: the equation assumes air speed equals ground speed.
• Upload the .FIT file to the laboratory, enter the real total mass and the tyre’s Crr, and let it fit.

On Crr: the fit estimates CdA only, with Crr held fixed. That matters less than it seems at high speed —rolling resistance is a small fraction of the total— but it does matter in slow tests. Enter the actual tyre’s Crr, not a generic one.

How to read the quality of the fit

The platform returns not just a number but how much to trust it. The primary indicator is RMSE, the standard deviation of the residuals between virtual and real elevation, in metres. Below 1.5 m the fit is rated excellent; up to 3.5 m, good; up to 7 m, fair; above that the ride is not usable for measurement and should be repeated.

Alongside RMSE the platform reports the closure error —how much elevation is left over or missing on returning to the start— and, when there are several laps, the spread of CdA across them. High spread with good RMSE usually betrays wind: each lap fits well on its own but to a different CdA.

A measured CdA is stored against the bike and the position, and from then on the plan uses it instead of the estimate. Because CdA depends on position, the platform stores road, time-trial, triathlon and velodrome values separately: using the TT figure for a track event would prescribe too few watts.

05Create the plan over a course

With the athlete and bike defined, under “Plans → New” you build the plan for a specific event. The process chains importing the course, loading the weather and optimizing power.

Import the course

Upload the .gpx by dragging it onto the import area, or import it directly from your Strava account (“Connect with Strava” button → pick one of your routes). The platform resamples the track every 50 metres, smooths the elevation to remove GPS noise and segments it automatically by gradient. You can trim the start and end by kilometre if the file includes warm-up or transfers unrelated to the event.

Load the weather at your ETA

Enter the start date and time and press “Load weather.” The platform does not apply the start-time weather to the whole course: it estimates the time of passage (ETA) for each segment and queries the wind, temperature and rain for that moment and place, with multi-point sampling on long routes. Wind is projected onto each segment’s heading to obtain the real head/tailwind component, and rain adjusts Crr for a wet surface.

Optimize power

Optimization decides how many watts to assign to each segment to minimize total time while respecting the W′ budget. Choose the mode first:

• W′-optimized: dynamic programming over the CP/W′ model. It distributes power so the reserve empties down to the margin you set as you cross the line, spending more where it pays off (climbs and headwinds) and less where saving does not (descents).
• By % of CP: sets intensity as a percentage of CP, useful for constant-pace plans or workouts.

The W′ depletion parameter controls how much reserve you allow to be spent: 0% keeps power below CP (sustainable effort, W′ untouched); high values (70–90%) squeeze the reserve for a one-day event. “Set Power Limits” let you cap maximum power by duration so the plan does not propose unrealistic efforts on short segments.

Press “Power Plan Optimizer” to compute the plan, or “Calculate preview” to see the result without saving it. The output is a per-segment power target, coloured by effort, with the W′bal trace showing how much reserve remains at each point.

06Weather: wind, air density and rain

Weather enters the model through two distinct doors, and it pays not to conflate them. First, air density ρ, which multiplies the aerodynamic term directly and depends on pressure, temperature and humidity. Second, wind, which changes the air speed relative to the rider and therefore the speed that gets cubed.

Why wind is computed at the time of passage

A common mistake is to apply the start-line forecast to the whole course. On a three-hour event that is simply false: wind veers and strengthens through the morning. The platform first estimates the time of passage through each segment and then queries the forecast wind for that specific hour and location. The result is a wind field along the course, not a single vector.

This has a consequence that surprises many coaches: the optimal power distribution depends on wind as much as on gradient. Into a headwind it pays to invest watts, because time lost there is greater; with a tailwind it pays to save them, because extra watts yield less. It is the same reasoning applied to climbs and descents, and the platform solves both jointly.

Head, cross and tailwind

The platform resolves the wind against each segment’s bearing. Only the longitudinal component adds to or subtracts from air speed; the transverse component does not reduce drag — it increases the yaw angle and with it the effective CdA. A strong crosswind is not a neutral wind.

Rain

Precipitation probability is queried along with the rest of the forecast. When it is appreciable the model raises the rolling resistance coefficient on the affected segments —wet tarmac rolls worse— and, more importantly, reduces the available lateral grip, which lowers cornering speeds. The effect on total time is usually larger through the corners than through rolling resistance.

Forecasts are only reliable a few days out. Recompute the plan the day before: that is when the forecast is good enough and there is still time to adjust strategy and equipment.

07Racing line: corners, roundabouts and drafting

Two factors that point-mass physics ignores can cost minutes in a real event: corners, which force braking and reacceleration, and drafting, which changes aerodynamic resistance depending on who you ride with.

Cornering speed

In a turn the limit is not power but lateral grip. Maximum cornering speed is set by the friction circle:

v_max = √( μ · g · R )

Where R is the corner radius and μ the available lateral grip coefficient, which the platform takes as 0.60 in the dry and 0.38 on wet tarmac — a reduction of almost 40 % that explains why a technical circuit becomes so much slower in the rain.

The delicate part is estimating R from a GPS file, which comes with noise and irregular sampling. The platform uses Menger curvature: the circle through three trace points separated by a fixed arc distance of about ten metres either side. By fixing the separation in distance travelled rather than in number of points, the estimate is robust to irregular sampling and to GPS jitter.

κ = 4·Area(P₋, P₀, P₊) / ( |P₋P₀| · |P₀P₊| · |P₋P₊| ) , R = 1/κ

Consecutive points whose limit speed falls below the threshold are grouped into a corner, and the minimum-radius point —the apex— sets the speed through it. The time cost is not just the arc: it includes the braking beforehand and the reacceleration afterwards, which is markedly slower than braking because it is limited by available power.

Drafting

Riding sheltered reduces aerodynamic drag and with it the power needed for the same speed. The saving depends on group size, on position within it, and on the fraction of time spent on a wheel rather than on the front.

The platform models the maximum saving with a logarithmic curve in the number of riders, calibrated against the literature: around 20 % in a pair, 45 % in a group of twenty-five and 66 % in a peloton of two hundred, with an asymptote near 68 %. The references are the paceline study by Íñiguez-de-la-Torre and Íñiguez (2009) and the CFD and wind-tunnel work of Blocken and colleagues (2018) on a 121-rider peloton, where the most sheltered positions see less than 10 % of the isolated rider’s drag.

Drafting is applied to the scenario you declare, not the one that happens. If the plan is for an individual time trial, leave it at one; if it is for a bunch stage, state the group size and the sheltered fraction you expect. A road plan computed as though it were a time trial will ask for watts nobody needs to spend.

08Tune and refine the plan (Time Analysis)

The “Time Analysis” tab is the plan’s test bench. It lets you move four parameters and see, live, how the estimated time changes and which segments gain or lose the most.

• Drag (CdA): simulates a more or less aerodynamic position. Its effect concentrates on the flat and the TT.
• Power: adds or removes watts globally to explore form scenarios.
• Weight: adds or removes system mass; its effect spikes uphill.
• Crr: simulates a better or worse tyre or surface.

Below the sliders, three figures summarize the experiment: base time (the current plan’s), gain (positive in green if you save time, negative in red if you lose it) and adjusted time. The elevation profile colours each segment by how much time the change gains (green) or costs (red), so you see where each parameter matters: CdA paints the flat green, weight paints the climbs red.

When an adjustment represents a real decision (a new position, a different tyre), save it with “Save adjustments” so it is incorporated into the plan. The W/CdA gauge and the yaw histogram help assess aerodynamic sensitivity and the effect of crosswind along the event.

Work by hypothesis: change one parameter at a time, watch where the gain concentrates and decide whether that improvement is achievable in reality. Time Analysis turns “I think I’d go faster” into “I’d gain 47 s, almost all in the final 8 km of flat.”

09Track: the individual pursuit

The velodrome removes two of the three unknowns of road cycling —there is no gradient and no wind— and introduces others of its own: banking, a single fixed gear and the standing start. The platform treats the track as a separate discipline with its own computation, not as a flat course.

Setting up the event

You define the lap length (250 m at most competition venues, 200 or 333 m elsewhere), the total distance —4,000 m for the individual pursuit— and the gear, which on the track is fixed: there is no shifting. The platform computes the number of laps and offers the chainring and sprocket combinations whose resulting cadence falls in the usable range.

Banking and rider lean

On the bends the rider leans to balance the centrifugal force. The lean angle from vertical is given by:

θ = arctan( v² / (R·g) )

Leaning reduces the projected frontal area, so effective CdA in the bends is lower than on the straights. Since half of a velodrome lap is spent in the bends, the effect on total time is not negligible.

The double-counting trap, and how it is avoided

There is a subtlety worth understanding here. If CdA was measured in the velodrome itself, that value already includes the lean: the virtual elevation method returns the lap-average CdA, not the straight-line value, and it comes out around 8 % lower. Applying the lean model on top would discount the same effect twice and the plan would come out optimistic.

For that reason, when the bike has a CdA measured on the track the platform switches the lean model off automatically. And for that reason the velodrome CdA has its own field: a CdA measured in a road time trial was measured in a straight line, does not include the lean, and must not switch that model off.

The standing start and effort distribution

The first lap is unlike the rest: it starts from a standstill and much of the work goes into accelerating the mass of the system rather than overcoming resistance. The platform solves it by energy balance and assigns it a start power far above cruising, capped by the athlete’s maximum power.

From there the remaining laps are optimised with the W′bal model exactly as on the road: the anaerobic reserve is spent so that total time is minimal without running out before the line. In a well-paced pursuit the power profile falls after the start and rises slightly at the end, when there is no longer any reserve to preserve.

On measuring CdA on the track: virtual elevation works better there than on the road. In a velodrome the real elevation is known and constant, so the criterion stops being “the loop must close” and becomes “the virtual elevation must come out flat”, which is stricter. And there is no wind, the method’s largest source of error outdoors. On the other hand R² is undefined —real elevation has no variance to explain— and the platform reports it as unavailable rather than as zero; reliability there is set by RMSE.

10Export, follow in the race and best practices

With the plan finalized, export it to the athlete’s device without copying numbers by hand. The available formats cover the usual ecosystems:

• Garmin / Wahoo (.FIT): the plan appears as a per-segment power target on the head unit.
• Zwift (.ZWO) and TrainerRoad (.ERG / .MRC): to run the plan on a trainer.
• TrainingPeaks and printable PDF: for tracking and the road book.

Best practices

• Recalibrate CP and W′ regularly; they are the basis of the whole calculation.
• Pick the right bike for the profile (aero on the flat, light in the mountains) and reflect the real gear in the garage.
• Check the weather the day before: wind changes the optimal power distribution more than it seems.
• Leave a realistic W′ margin on events with a demanding finish; finishing with the reserve at zero is only optimal if the finish is the last ramp.

The public demo (“Demo” menu) reproduces this whole flow over a real time trial: it is the best place to get familiar with Time Analysis before working with your own athletes.