RAMTT models what a ride costs the body, not just what it produced. Every number in the platform rests on graded evidence, and this page shows where the evidence comes from, how it is used, and where it runs out.
We model the four systems that decide what an athlete can do today, and we show the evidence behind every number we put in front of you, graded, so you can see what rests on measurement and what is still a guess.
Most training platforms count the work you did. Kilojoules, hours, a stress score, a form number. Counts are useful, and RAMTT computes all of them, but a count answers the question of what you did, not the question an athlete actually has at seven in the morning, which is what the body can do today and what it will cost to find out. To answer that you need a model of the athlete, not a scoreboard, and a model has to be honest about how sure it is.
That is why RAMTT keeps a running estimate of the things that change from hour to hour, fuel, thresholds, recovery and durability, and carries a range with each of them. The width of the range is part of the answer. A narrow range after a lab test means one thing; a wide range after three weeks of unlogged rides means another, and the platform says which it is rather than hiding the difference behind a single number.
Every claim also carries a grade: established, supported, emerging, early, or our own. The grade describes the state of the research behind the claim, not our confidence in you. The sections below walk through each of the four systems, what the model does with them, and which evidence it stands on. Where the evidence is thin we say so, and where our own validation is the only thing holding a number up, that is marked too.
Fueling is the system with the clearest evidence and the most room for error in practice. The literature on how much carbohydrate an athlete can take in per hour, and how that ceiling rises with training, is consistent across populations, which is why the core of the fueling model carries an established grade. What the literature does not tell you is what your gut can take today, after a poor night and a nervous morning, and that is where the model takes over from the paper.
RAMTT keeps a running estimate of what you have stored, what you have taken in, and what the ride is costing, and it turns those into an hourly target with a range around it. The target is never a single number handed down; it is a band, and the band narrows as the model sees more of your rides. When the band is wide, the plan says so and asks for a fueling log rather than pretending.
The gut training ladder, which raises tolerance week by week, moved from early to emerging this season after a replication in a trained population. It is the newest part of the model and the part we watch most closely. The note lists what held, what did not, and what a coach should do while it is still emerging.
A threshold is estimated from many rides and moves slowly, which makes it the most stable number in the platform and the easiest to over-trust. RAMTT treats the fresh threshold, the one you would measure on a good day, as one thing, and the threshold you actually have in the third hour of a ride as another. The second drifts as fuel runs down, and the model says how far.
The reserve above threshold is where a race is decided, and it is the number most platforms get wrong by treating it as fixed. RAMTT recomputes it every minute from the drifting threshold and what you have spent, and shows the range rather than a point. The range is wide on a ride the model has never seen the like of and narrow on the tenth Tuesday interval session in a row.
The power-duration model behind all of this is established and has been for two decades. What is ours is the coupling between fuel and threshold within a ride, and that carries its own grade.
Recovery is the system with the least evidence and the most opinions. Most of what is sold as recovery science rests on small studies in narrow populations, and we grade it accordingly. RAMTT's daily model reads three regulators from a short morning check-in and adds them into a seven-day balance, and it says openly that the balance is a low-confidence estimate that lacks validation.
That honesty is the design, not an apology for it. A recovery number that looks as confident on its worst day as on its best is worse than no number, because it teaches the athlete to ignore it. The balance shows its range, it widens when the check-ins stop, and it never overrides a hard rule such as the return after illness.
We are validating the three regulators against our own athletes this season. Until that is done the grade stays where it is, and the note says what would move it.
Durability is what the last hour of a long ride costs, and it is the part of endurance the standard models miss entirely because they describe a fresh athlete. RAMTT reads it within a ride from the drift of heart rate against power, the fall in the power you can hold as the ride goes on, and the way threshold settles as fuel runs down.
The index that comes out of this is a single number from zero to a hundred, and it is the one number in the platform we are most careful with, because it is ours. It was checked against forty riders over a season, and the note reports where it held and where it did not. The range around it is wider than around any threshold, and it stays that way until the next validation round.
Durability feeds the race-day plan directly: an athlete whose index is low needs more fuel earlier and a more conservative reserve, and the plan says so, with the grade next to it.
| Title | Description | Topic |
|---|---|---|
| Race-day plan now carries its evidence grade | The fueling plan you take into a race now shows the grade behind each hourly target, and the range the target was chosen from, so a coach can see at a glance what rests on measurement and what rests on a population average. | Fueling |
| Thresholds drift within a ride | Threshold is no longer a fixed line for the whole ride. The model now lets it settle as fuel runs down, and the reserve above it is recomputed every minute, which changes what the last climb looks like on the screen. | Thresholds |
| Morning check-in shortened to three questions | We removed four questions that did not move any estimate in twelve months of data and kept the three that do. The check-in now takes under twenty seconds and the regulators are unchanged. | Recovery |
| Coach overrides are shown to the athlete | When a coach overrides a recommendation, the athlete now sees both the model's answer and the coach's, with the coach's name on the override. Nothing is hidden on either side. | Coaching |
| Out of date is now its own state | A number whose data has gone stale is no longer shown as if it were fresh. It is marked out of date, which is a different state from not yet given, and each has its own action. | Platform |
| Gut training ladder graded emerging | The ladder that raises carbohydrate tolerance week by week has moved from early to emerging after a replication in a trained population. The note lists the study and what still needs to hold. | Science |
| Durability index validated against forty riders | The within-ride durability index has been checked against forty riders over a season. The note reports where it held, where it did not, and what the range now looks like. | Science |
| Garmin and Wahoo import runs locally | Ride files are now parsed on your device and only the aggregates leave it. The raw file stays with you unless you choose to share it with a coach. | Platform |
| Weekly review rewritten in plain sentences | The weekly review no longer opens with a score. It opens with what changed, what the model is unsure about, and what would settle it, in the order a coach would ask. | Coaching |
| Letters: why we grade evidence at the level of the claim | A long piece on why one screen can carry three grades, how the grades are set, and what we do when two graders disagree. It is the reasoning behind the grade you see next to every number. | Letters |