How Trailia Builds Your Training Plan: Coaching Principles & Science

The coaching principles behind every Trailia trail running plan: periodization toward your goal race, gradual load, recovery weeks, vertical and descent trained as their own skills, race-specific preparation and tapering, with the published research they draw on.

Last updated 2026-09-19 | Written by Gianluca Belloni, Founder and trail running coach

Trailia writes your plan with AI, but the AI does not get to decide what good training is. It works inside coaching rules set by a CONI-recognised trail running coach: what a trail race actually demands, which loads may grow and in what order, when to back off, and how to arrive at the start line fresh. Every week is checked against those rules before it reaches your calendar.

This page sets out those principles in plain language, with the published research behind them, so you can judge the coaching before you trust a season to it.

It starts from what you actually run

A plan is only as good as its starting point, so the first thing Trailia does is measure where you are rather than ask you how fit you feel. It reads your recent training from Strava, Garmin Connect, Polar Flow or uploaded files: how much you run in a normal week, how much climbing you do, your longest run, how consistent you have been and which direction the trend is going.

Around that baseline it takes the things only you can say: the goal race and its terrain, how many days a week you can train and for how long on weekdays and at weekends, which days suit hard work, whether you have hills within reach, your experience and any injury history.

  • Real activity data sets the baseline. Self-reported fitness is used only when there is no data, and the plan tells you that is what happened.
  • A layoff is detected rather than ignored: someone coming back after weeks off restarts from a comeback volume, not from the training they were doing before they stopped.
  • A well-trained athlete preparing a shorter race is never regressed to a beginner plan for the distance. Normal training weeks do not drop below the running and climbing you already handle.

Honest about the calendar

Before anything is scheduled, the plan compares where you are with what your race demands and asks whether that gap can be closed safely in the weeks available. When it cannot, the plan is built toward finishing rather than performing, and it says so plainly instead of quietly hoping.

The progression is never compressed to make a race fit. A plan that gets you to the start line slightly under-prepared but healthy beats one that rushes the ramp and gets you injured in week nine. This is the single place where a training app is most tempted to flatter you, and the one where flattery costs the most.

Time and vertical, not just kilometres

On trail, the same distance can be a completely different session depending on gradient, surface and conditions. So off-road workouts are prescribed by duration and by climbing and descending metres, with distance shown as an approximate figure. Pace and distance are primary targets only on road, track and flat ground.

When distance and time have to be converted, the conversion accounts for the cost of climbing rather than treating a mountain kilometre as a flat one. A pace target quoted against raw trail distance is not a training prescription, it is an arithmetic error, and it is why plans written for the road read as nonsense in the mountains.

Built backwards from race day

The plan is designed from the fitness you need on race day back to where you are today, and it moves from general to specific as the race approaches. Early weeks build capacities that develop best under controlled conditions. Later weeks spend those capacities on terrain that looks like the race. How the available weeks are divided between those stages depends on how many of them there are, and on the distance you are preparing for.

  • A base stage builds the aerobic engine and raises volume safely, and starts protective downhill exposure in small doses while quality work stays somewhere pace is controllable.
  • A build stage introduces race-specific work: terrain, gradient and duration. The biggest training weeks land here, not in the final block.
  • A peak stage is about specificity rather than volume: race simulation, race-day kit and fueling on race-like ground. The longest run of the plan happens well before race day, never in the final fortnight.
  • A taper reduces volume progressively while keeping intensity and session frequency, and its length is scaled to how long you will be out there. A race sitting in the final week of a plan is treated as the race, not as one of the taper weeks.

Load grows gradually, and one thing at a time

Running volume, climbing and descending are three separate loads, and they are allowed to grow separately. No more than one of them rises meaningfully in a given week. Overuse injuries come from total load spiking across several channels at once, which is exactly what this prevents, and it is the rule most self-written plans break.

How fast volume may rise depends on where you currently are, because a single percentage is meaningless at the bottom of the range and reckless at the top. Descending is progressed most conservatively of all: it is the load that does the muscle damage, and it is what takes your legs away in the last third of a race.

  • Recovery weeks come round regularly, and more often for athletes who are new to structured training or over 50. Training resumes from where it was before the deload, so backing off never costs you ground.
  • No single session jumps far beyond the longest equivalent you have recently done.
  • The long run is a proportion of your week, not a heroic outlier, and weeks are balanced rather than built around one enormous day.
  • New session types are introduced one at a time, and never in the same week as a jump in volume or climbing.
  • There is a floor below which a run is not training, and a ceiling on how far any build can take you, however long the plan. Compounding a weekly percentage for a year produces numbers no coach would ever write.

Mostly easy, and hard when it counts

Intensity is distributed pyramidally and judged by how much time you spend in each zone across the week, not by counting workouts. The large majority of running is genuinely easy, a smaller share is moderate, and a small share is hard. Easy means easy by breathing and heart rate: an easy run can still be technical, steep and on a mountain.

Hard sessions are rationed deliberately. Most weeks hold at most two, and one good quality session with the rest aerobic is a better week for most runners than three mediocre ones. The first weeks of a new plan carry less, because rhythm comes before intensity. Recovery weeks carry none.

  • Zones are anchored either on a lactate-threshold field test or on maximum heart rate, and never mixed between the two. Zones estimated from an age formula are flagged as provisional, with a field test recommended.
  • Effort governs, not the number on your wrist. Heart rate runs high in heat and low in cold at the same effort, and lags behind on steep or technical ground.
  • Heart rate well above or well below your own norm is treated as a reason to reduce load, not to push through.
  • The long run is its own kind of load rather than one of the week's hard sessions, though a long run with race-pace work in it still counts as hard when the week is spaced out.

What your race actually asks for

Peak training is set by what safe progression from your baseline can reach, then checked against what the race demands in volume and in climbing. Where it falls short, the plan tells you rather than forcing the ramp. Where you are already past what the race asks, it stops adding volume you have no use for and spends the block on specificity instead.

The longest session is prescribed in hours and capped, because covering race distance in training is not the goal. Accumulated weekly volume, repeated exposure to climbing and back-to-back weekends build more fitness at less risk than one enormous day that leaves you wrecked for a fortnight.

  • Descending is trained as its own quality: small doses early, progressed carefully, with the most demanding descent work placed so the protective adaptation is still there on race day.
  • Power hiking is real training, not a failure to run. Above a certain gradient it is more economical for most runners, so mountain plans drill it, and ultra plans rehearse the decision of when to switch.
  • Poles, packs, headlamps, shoes and fueling are rehearsed in training at race weight and race timing. Nothing is new on race day, ever.
  • Fueling is practised and progressed across the block rather than started at race rate, because the gut adapts to repeated exposure like everything else.
  • Races with heat, altitude, night sections or technical ground trigger preparation specific to them, scheduled at the point in the block where the adaptation will still be present when you need it.

It adapts to the week you actually had

Synced activities are matched automatically to the session they were meant to be, and you can log how an effort felt, how you slept and what hurt. The following week is generated from what actually happened, not from what was on the calendar.

The governing rule is to adjust rather than skip. When fatigue, poor sleep, soreness, illness or an unusual heart rate shows up, sessions are shortened before they are dropped, and hard work is converted to easy before it is converted to rest. Missing one session changes nothing; missing several changes the plan.

  • Training load is tracked as the ratio of recent to longer-term load, the measure associated with injury risk when it spikes. Consistently high effort on easy runs reduces load and prompts a check for overtraining.
  • Pain is graded rather than ignored. Pain only after activity is managed by removing what aggravates it; pain during activity means reducing load and reassessing; pain with real functional limitation stops progression and sends you to a clinician. A plan never tries to train around that last case.
  • When you run out of weeks, the plan continues from where you are rather than restarting from a template.

Predictions come as a range, with a confidence level

Every goal race gets three predicted finish times, fastest, most likely and slowest, rather than one number pretending to a precision the data does not support. The prediction reads your recent training and, when you upload the course GPX, the actual elevation profile, translating your flat fitness into climbing and descending speed.

Each prediction carries a confidence level that falls when there is little training history, no long run near the race distance, or no comparable race behind you. Predictions update as your fitness changes, and a prediction that says it is uncertain is more useful than one that does not.

Checked before you see it

A plan is not trusted just because it came out of a model. Automated checks run over every week before it is saved, and every finished plan is reviewed against the same coaching principles described on this page, scored and kept on file so that plan quality is measured over time rather than assumed. Where a plan drifts from the principles, it is caught and corrected rather than shipped.

If a plan of yours still looks wrong, it probably is. Email hello@trailia.run with it and a person will look.

What Trailia is not

Trailia is not medical advice. Nutrition figures are framed as ranges to practise, not prescriptions; injury guidance is about training decisions, not diagnosis; and pain with real functional limitation is sent to a clinician rather than trained around. If you have a medical condition, are pregnant, or are returning from a serious injury, involve a qualified professional.

It is also not a substitute for a coach's judgment in an unusual situation. That is why the rules exist, why every plan is checked, why any workout can be edited or the plan regenerated in plain language, and why the person who built it also coaches athletes directly.

Sources

The principles above draw on the following published research and on established trail-coaching practice. Each entry says what it informs. Where we have had to choose a value inside the range the evidence supports, that is a coaching judgment rather than a number lifted from a single study.

  1. Gabbett TJ. The training-injury prevention paradox: should athletes be training smarter and harder? British Journal of Sports Medicine, 2016. https://bjsm.bmj.com/content/50/5/273 Used for: Tracking the ratio of recent to longer-term training load, and treating a spike in it as an injury-risk signal.
  2. Nielsen RO, Parner ET, Nohr EA, et al. Excessive progression in weekly running distance and risk of running-related injuries. Journal of Orthopaedic & Sports Physical Therapy, 2014. https://www.jospt.org/doi/10.2519/jospt.2014.5164 Used for: Why weekly volume is allowed to rise gradually, and why the limit depends on where an athlete currently is.
  3. Damsted C, Glad S, Nielsen RO, Sørensen H, Malisoux L. Is there evidence for an association between changes in training load and running-related injuries? A systematic review. International Journal of Sports Physical Therapy, 2018. Used for: The evidence, and its limits, behind progression rules and the principle of raising one load at a time.
  4. Seiler S. What is best practice for training intensity and duration distribution in endurance athletes? International Journal of Sports Physiology and Performance, 2010. Used for: Distributing intensity by time in zone, and the case for keeping most running genuinely easy.
  5. Mujika I, Padilla S. Scientific bases for precompetition tapering strategies. Medicine & Science in Sports & Exercise, 2003. Used for: Taper design: reducing volume progressively while intensity and session frequency are maintained.
  6. Bosquet L, Montpetit J, Arvisais D, Mujika I. Effects of tapering on performance: a meta-analysis. Medicine & Science in Sports & Exercise, 2007. Used for: How much volume a taper should remove, and how long a taper should run.
  7. Minetti AE, Moia C, Roi GS, Susta D, Ferretti G. Energy cost of walking and running at extreme uphill and downhill slopes. Journal of Applied Physiology, 2002. https://journals.physiology.org/doi/full/10.1152/japplphysiol.01177.2001 Used for: The energy cost of gradient, why climbing is counted as its own load, and where running gives way to hiking.
  8. Vernillo G, Giandolini M, Edwards WB, et al. Biomechanics and physiology of uphill and downhill running. Sports Medicine, 2017. Used for: Why descending is trained as its own quality and progressed more conservatively than climbing.
  9. Giandolini M, Vernillo G, Samozino P, et al. Fatigue associated with prolonged graded running. European Journal of Applied Physiology, 2016. Used for: Eccentric muscle damage from downhill running and the loss of force late in a mountain race.
  10. Jeukendrup A. A step towards personalized sports nutrition: carbohydrate intake during exercise. Sports Medicine, 2014. Used for: Carbohydrate intake during long efforts, and the case for training the gut rather than guessing on race day.
  11. Périard JD, Racinais S, Sawka MN. Adaptations and mechanisms of human heat acclimation. Scandinavian Journal of Medicine & Science in Sports, 2015. Used for: Preparing for hot races: how many exposures it takes, when to schedule them, and how quickly the adaptation fades.

Frequently Asked Questions

Is my plan written by an AI or by a coach?

Both, and in that order matters. A coach set the rules the plan has to obey, AI writes the sessions inside them, automated checks verify every week before it is saved, and the finished plan is reviewed against the same principles. The coaching judgment is in the rules, not in the model.

Why does my plan build volume so slowly?

Because injury comes from load rising faster than tissue adapts. Volume, climbing and descending each grow gradually and separately, recovery weeks come round regularly, and there is a limit on how far any block can take you. If your race is close and the gap is large, the plan will tell you that rather than compressing the ramp to hide it.

Why are my trail workouts in minutes and vertical metres instead of kilometres?

Because on trail the same distance can be a completely different session depending on gradient and terrain. Off-road sessions are prescribed by duration and climbing, with distance as an estimate. Road and track sessions still use pace and distance.

Why is my longest run shorter than my race?

Covering race distance in training is not the goal and rarely survives contact with the next two weeks. Weekly volume, repeated climbing and back-to-back weekends build more fitness at less risk than one enormous day, so the longest session is capped and placed where it still helps on race day.

Can I change how the plan is built?

You can change what it is built from: training days, time limits, which days suit hard work, strength frequency and your intensity preference, and you can edit or move any workout. The safety limits themselves are not adjustable, which is the point of having them.

What if a plan looks wrong to me?

Email hello@trailia.run with it. Every plan is reviewed automatically and kept on file, so a specific plan can be looked at, and a bad one tells us something worth fixing.