Measure
Training load, intensity, power, heart rate, wellness and performance evidence establish the athlete's observable state.
Evidence and methods
The science matters. The athlete should not need to decode it.
Montis combines established endurance physiology, longitudinal athlete data and explicit decision rules. It does not treat individual metrics as answers. Evidence is progressively resolved into capability, adaptation and finally a governed coaching action.
Montis controls the coaching dialogue. The AI does not invent the athlete's physiological state, adaptation status or coaching decision. It receives governed evidence and decisions from Montis and turns them into useful conversation.
Production coaching pipeline
Performance Intelligence
ESPE
ADE
Deterministic decision layer
The Montis intelligence model
Five coaching layers. One governed decision.
Montis does not interpret training from a single metric. Training load, physiological response, expressed performance, longitudinal adaptation and decision governance are resolved as connected layers of the same coaching model.
The Intelligence Stack below is the framework used throughout Montis. Each layer has a distinct role, and the sections that follow explain the science and decision logic behind them.
The governing idea
From training data to a coaching decision.
Training creates stress. Physiology reveals the response. Performance Intelligence describes how capability behaves under that stress. ESPE asks whether the athlete is actually adapting. ADE then resolves current capacity together with phase, recovery and event governance to determine what should happen next.
Language explains the decision. It does not invent the physiology.
Athlete evidence is validated, interpreted and resolved by the Montis coaching intelligence stack before it reaches the language model. Performance Intelligence, ESPE and ADE establish the governed state and coaching boundaries. The LLM can explain, question and discuss that result, but it cannot independently redefine the athlete state or invent an unsupported coaching action.
Interpret
Load, recovery and Performance Intelligence describe how the athlete is responding to recent and accumulated stress.
Compare
ESPE compares power-duration behaviour across rolling windows to identify the direction of adaptation.
Decide
ADE evaluates current operational capacity, risk, forecast fatigue, adaptation context and event state.
Govern
Phase, recovery, taper and event requirements determine the final coaching instruction.
Scientific foundations
Established concepts. Combined rather than isolated.
Montis does not treat any single physiological model as a complete representation of an athlete. Different models provide different evidence, and that evidence is resolved together before a training decision is made.
Fitness and fatigue
Chronic and acute load concepts provide longitudinal context for how much training stress is being accumulated and carried.
Evidence includes CTL, ATL, TSB and recent load direction.How training stress is distributed
Volume alone does not describe training. Montis considers the distribution and density of intensity as part of the overall stress picture.
Influenced by established intensity-distribution and polarisation concepts.CP and W′
Critical Power and W′ provide useful context for sustained performance and finite supra-threshold work capacity.
Used alongside depletion and work-above-threshold evidence rather than as a standalone answer.Cardiovascular drift and decoupling
The ability to preserve physiological efficiency during prolonged work reveals something that FTP alone cannot: how well performance survives duration.
Montis examines repeated decoupling and long-session exposure rather than relying on one isolated ride.Power-duration behaviour
Short-, medium- and long-duration power describe different performance qualities and provide a longitudinal view of what the athlete can currently express.
Used by ESPE to compare adaptation across energy-system-relevant durations.Periodisation and tapering
The fact that an athlete can tolerate another hard session does not mean that session is strategically correct.
Phase, recovery, event timing and projected form can override short-term capacity.Scientific & methodological references
Established research informs the system. It does not replace athlete-specific evidence.
Montis draws on established endurance-training, performance-modelling, recovery and sports-nutrition research. These frameworks provide methodological anchors for individual parts of the system rather than acting as independent coaching algorithms.
| Reference / framework | Scientific contribution | Application in Montis |
|---|---|---|
| Banister Fitness–fatigue / impulse-response | Models the delayed positive and negative effects of training stimulus on performance capacity over time. | Training-load context, fitness/fatigue balance, TSB-related interpretation and phase-state context. Key literature: Fitz-Clarke, Morton & Banister, 1991 |
| Foster Training monotony & strain | Examines training-load variation and the accumulated stress associated with repeated or highly monotonous loading. | Monotony, strain and load-pattern risk used within broader load and recovery interpretation. Key literature: Foster, 1998 |
| Seiler Endurance training-intensity distribution | Describes how endurance training is distributed across low, moderate and high-intensity domains and the predominance of low-intensity work in successful endurance programmes. | Training-intensity distribution, aerobic-volume context and phase-aware interpretation of low versus high-intensity exposure. Key literature: Seiler, 2010 |
| Stöggl & Sperlich Polarised & pyramidal intensity distribution | Extends training-intensity-distribution research across polarised, pyramidal, threshold and high-volume approaches. | Helps contextualise intensity-distribution patterns rather than treating a fixed 80/20 split as a universal prescription. Key literature: Stöggl & Sperlich, 2015 · Stöggl & Sperlich, 2014 |
| Treff et al. Polarization Index | Provides a mathematical index designed to distinguish polarised from non-polarised training-intensity distributions. | Direct methodological basis for Montis Polarisation Index classification and intensity-distribution semantics. Key literature: Treff et al., 2019 |
| Critical Power / W′ Power–duration physiology | Describes the relationship between sustainable power and finite work capacity above Critical Power. | CP, W′, supra-threshold exposure, depletion behaviour and WDRM performance-intelligence context. Key literature: Poole, Burnley, Vanhatalo, Rossiter & Jones, 2016 |
| Durability research Physiological deterioration under prolonged exercise | Examines the magnitude and onset of deterioration in physiological and performance characteristics as prolonged work accumulates. | Durability interpretation, prolonged-session behaviour and ISDM structural performance context. Key literature: Matomäki et al., 2023 |
| Cardiovascular drift & decoupling Internal–external workload stability | Describes changes in cardiovascular response and the relationship between internal and external workload during prolonged endurance exercise. | Power–HR decoupling, repeated drift evidence, long-session stability and ISDM durability state. Key literature: Smyth et al., 2022 · Cardiovascular drift review, 2021 |
| Power-duration modelling Performance expression across duration | Models the relationship between maximal sustainable power and exercise duration to describe different performance domains. | ESPE progression across 1 min, 5 min, 20 min and 60 min anchors, curve shape, CP, W′ and related performance parameters. Key literature: Power profiling & power-duration review, 2021 |
| Allen · Coggan · McGregor Power-based training methodology | Practical power-based athlete profiling using FTP, power-duration characteristics and training/racing power data. | FTP and power-profile context, power-zone semantics and interpretation of longitudinal cycling performance data. Methodological reference: Training and Racing with a Power Meter, 3rd ed. |
| HRV / autonomic monitoring Recovery and adaptation context | Uses longitudinal variation in autonomic activity as one contextual signal of training response and recovery status. | HRV trend, HRV ratio and recovery guardrails alongside wellness, load and performance evidence. Key literature: Düking et al., 2021 · Manresa-Rocamora et al., 2021 |
| San-Millán & Brooks Metabolic flexibility | Relates lactate response, carbohydrate oxidation and fat oxidation to metabolic flexibility and endurance phenotype. | Provides methodological context for metabolic-efficiency and fuel-utilisation semantics when the necessary evidence exists. Key literature: San-Millán & Brooks, 2018 |
| Mujika & Padilla Tapering & performance maintenance | Describes taper strategies intended to reduce accumulated fatigue while retaining training adaptations before competition. | Taper classification, event proximity, load-reduction governance and projected event-form interpretation. Key literature: Mujika & Padilla, 2003 · Bosquet et al. taper meta-analysis, 2007 |
| Issurin Block periodisation | Structured concentration and sequencing of training objectives across defined training blocks. | Phase progression, transitions between training objectives and training-purpose governance. Key literature: Issurin, 2008 |
| Friel Practical endurance periodisation | Applied endurance-training methodology covering seasonal structure, progressive loading, recovery and race preparation. My reference God for 30+ years. | Practical phase organisation across Recovery, Base, Build, Peak and Taper states alongside the physiological models used by Montis. Methodological reference: The Cyclist's Training Bible |
| Sandbakk & Holmberg Integrated endurance performance | Examines how aerobic capacity, economy, intensity, technique and sport-specific demands interact in elite endurance performance. | Provides broader endurance-performance context for multi-signal interpretation rather than defining a single proprietary Montis metric. Key literature: Sandbakk & Holmberg, 2014 · Staff et al., 2023 |
| IOC Sports Nutrition Carbohydrate availability & fueling | Sports-nutrition guidance matching carbohydrate availability and fueling strategy to the duration, intensity and demands of training and competition. | Primary methodological basis for Montis fuel-availability and carbohydrate-demand matching. Montis estimates daily carbohydrate demand from recent training duration and intensity and compares intake with that day's requirement rather than accumulating an assumed multi-day glycogen deficit. Key literature: IOC Consensus Statement on Sports Nutrition, 2010 · Burke, Hawley, Wong & Jeukendrup, 2011 |
| ACSM · Academy · Dietitians of Canada Sports nutrition | Nutrition availability, timing and intake matched to training, recovery and competition demands. | Carbohydrate-demand matching, fuel availability and nutrition context used alongside current and projected training load. Key literature: Thomas, Erdman & Burke, 2016 |
| ISSN Protein & exercise | Protein availability supporting muscle-protein synthesis, repair, recovery and adaptation to exercise. | Protein-intake context during demanding training and recovery periods. Key literature: Jäger et al., 2017 |
| DFA-α1 / fractal HRV Aerobic threshold & physiological-status monitoring | DFA-α1 describes fractal correlation properties of the RR-interval time series during exercise. Research has explored two relevant applications: aerobic-threshold / intensity demarcation during controlled incremental testing, and longitudinal physiological-status monitoring during standardized low-intensity exercise. | Montis can retrieve AlphaHRV-derived DFA-α1, coverage, artifact rate, respiration, RRa1 and readiness observations from Intervals.icu when available. A DFA-α1 value around 0.75 may provide aerobic-threshold context during an appropriate incremental protocol. Repeated DFA-α1 measured during a standardized submaximal warm-up may also provide supporting information about physiological status relative to an athlete's individual baseline. Whole-activity DFA-α1 averages are not treated as direct LT1 or readiness measurements. Key literature: Rogers et al., 2021 — aerobic threshold · Schaffarczyk et al., 2022 — standardized warm-up / physiological status · Gronwald & Hoos, 2020 — fractal HRV review · Rogers et al., 2023 — fatigue effects |
Methodological principle
Montis does not assume that one framework completely describes an athlete. References provide physiological and methodological context; the athlete's measured and longitudinal evidence determines the actual coaching state.
Performance Intelligence · PI
How does fitness behave when stress is applied?
Traditional training load describes how much work was performed. Performance Intelligence asks what happened to capability under that work: whether high-intensity capacity remains repeatable, aerobic efficiency survives duration, and demanding sessions are clustering faster than they can be absorbed.
Supra-threshold repeatability
WDRM examines how deeply and how often finite high-intensity work capacity is being used.
- W′ and rolling W′ context
- Maximum depletion percentage
- Mean depletion percentage
- Repeated moderate and high depletion
- Joules above FTP
- W′ model divergence diagnostics
Question: is supra-threshold capacity being expressed repeatedly
and proportionately?
Durability
ISDM examines whether the athlete can preserve physiological efficiency as duration and accumulated fatigue increase.
- Signed power–HR decoupling
- Absolute decoupling magnitude
- Repeated high-drift sessions
- Long-session exposure
- Acute and chronic durability state
Question: does aerobic performance remain stable as work
accumulates?
High-intensity density
NDLI evaluates whether demanding work is being concentrated densely enough to raise the stress burden relative to the athlete's current structural capacity.
- Rolling joules above FTP
- High-intensity training days
- Intensity Factor
- Efficiency Factor
- Variability Index
Question: is high-intensity exposure being stacked faster than it
appears to be absorbed?
Energy System Progression · ESPE
Load tells Montis what you did. ESPE asks whether it worked.
ESPE compares rolling power-duration windows rather than judging adaptation from one isolated personal best. The engine evaluates how different parts of the power curve move relative to the preceding comparison period.
This allows Montis to distinguish a general upward shift from a change in the shape of the curve, such as stronger short-duration power without corresponding threshold or durability progression.
1 min
Anaerobic capability
5 min
VO₂-domain capability
20 min
Threshold capability
60 min
Aerobic durability
Power-duration comparison
Current vs previous window
1 min
5 min
20 min
60 min
Current vs previous deltas
Vertical curve shift
Curve rotation
Adaptation bias
Plateau detection
Curve quality / R²
CP · W′ · pMax · FTP
Adaptation state
No previous comparison window?
ESPE returns a baseline state. It does not manufacture a progression claim when valid comparative evidence does not exist.
Adaptive Decision Engine · ADE
What you can do is not always what you should do.
Montis deliberately separates short-term physiological capacity from training strategy. ADE establishes the operational decision first. Phase and event governance then decide whether that capacity should actually be used.
What can the athlete tolerate now?
ADE evaluates the immediate operating state before phase governance is applied.
- Load-accepting vs recovery-priority state
- Risk classification
- Future fatigue classification
- Projected load direction
- HRV guardrail
- ESPE adaptation state
- Event / taper context
- Supplementary nutrition context
What is appropriate at this point in the cycle?
Current capability is interpreted in the context of the purpose of the training block and the athlete's next target.
- Current and projected phase
- Recovery or deload requirement
- Build progression
- Taper state
- Days to target event
- Projected event TSB
- Target event-form range
- Controlled sharpening when appropriate
Governed action
An athlete may be physiologically capable of more work while the correct strategic decision is consolidation, recovery or taper. Conversely, a tapering athlete who is projected to become too fresh may justify controlled sharpening. The final instruction reflects both capacity and purpose.
Time matters
The same signal means different things at different horizons.
Montis deliberately separates recent high-resolution evidence from broader structural context. Acute training state and long-term adaptation are related, but they are not interchangeable.
What is happening now?
Weekly analysis uses the high-resolution recent dataset to assess immediate execution and response.
- Recent training load
- Recovery and wellness
- W′ depletion exposure
- Durability behaviour
- High-intensity density
- Current operational training state
What pattern is emerging?
Season and Summary analysis use a broader chronic dataset together with the recent acute window.
- Repeated stress behaviour
- Chronic durability state
- Longer-term W′ exposure
- Intensity-density patterns
- Power-curve progression
- Adaptation direction
Evidence hierarchy
Measured, modelled, derived and classified are not the same thing.
Montis keeps different levels of evidence conceptually separate. A direct sensor measurement has a different status from a platform model, and a derived coaching classification has a different status again.
Measured / recorded
Power, heart rate, duration, distance, activity data and athlete
wellness inputs.
Platform-modelled
Values supplied or modelled by the athlete's training platform,
such as FTP, CP, W′ and training-load state where available.
Montis-derived
Durability state, supra-threshold exposure, intensity density,
power-curve deltas and adaptation classifications.
Coaching decision
Operational state, ADE capacity, phase alignment, event governance
and the final training directive.
Missing evidence does not become invented evidence.
Optional evidence
Additional physiological evidence can sharpen context. It should not become a prerequisite.
Montis is designed to remain useful with ordinary training, wellness and activity data. Laboratory thresholds, DFA-α1 / AlphaHRV, detailed physiological testing and high-quality power data can add useful context when available, but their absence should reduce confidence or scope rather than create invented precision.
When additional evidence exists
- Laboratory LT1 / LT2 can provide measured threshold context and help validate training-zone interpretation.
- A whole-activity DFA-α1 average is descriptive context only. It is not equivalent to a DFA-α1 threshold crossing from an incremental test, nor to a standardized warm-up assessment of physiological status.
- Montis can retrieve AlphaHRV-derived fields including DFA-α1, DFA-α1 coverage, artifact rate, respiration, RRa1 and readiness when they are available through Intervals.icu.
- Reliable wellness data, including HRV, resting heart rate, sleep and subjective recovery inputs, can strengthen recovery interpretation when viewed longitudinally.
- High-quality power data enables stronger power-duration, Critical Power, W′ and ESPE modelling.
- Additional athlete-specific physiological measurements can be used as supporting evidence where their provenance and meaning are known.
How Montis limits interpretation
- A missing laboratory threshold is never replaced with a claimed measured LT1 or LT2 value.
- A whole-activity DFA-α1 average is descriptive context. It is not equivalent to the DFA-α1 threshold crossing obtained during a controlled incremental or staged assessment.
- DFA-α1 coverage and artifact information should accompany DFA-α1 interpretation because signal quality materially affects confidence.
- AlphaHRV readiness is treated as supporting evidence and is not equivalent to Montis ADE readiness.
- DFA-α1 / AlphaHRV fields are currently available as optional evidence but are not used directly by ESPE, ISDM or ADE.
- Missing or poor-quality power reduces the conclusions available from power-curve and Critical Power modelling rather than triggering estimated replacement values.
- Reduced evidence should reduce confidence, specificity or scope — never create false precision.
Evidence hierarchy
Optional evidence can refine interpretation, but it does not override the core Montis evidence hierarchy. Measured athlete data remains distinct from platform-modelled values, Montis-derived metrics and the final coaching decision.
Scientific boundaries
Knowing where not to make a claim matters.
Activity summaries and longitudinal athlete data support many useful coaching conclusions. Other questions require stronger evidence, laboratory testing or time-series analysis.
Strongly supported by the current pipeline
- Training-load and recovery context
- Longitudinal power progression
- Supra-threshold exposure
- Activity-level durability indicators
- Repeated drift behaviour
- Energy-system progression
- Phase alignment
- Event-form governance
- Practical training direction
Montis does not directly measure
- CNS fatigue as a physiological measurement
- Muscle glycogen concentration
- VLamax without suitable underlying evidence
- Laboratory LT1 or LT2 when no test exists
- Exact W′ recovery kinetics from activity summaries alone
- Fresh-versus-fatigued power curves after exact work thresholds without the required time-resolved data
- Stage-specific DFA-alpha1 behaviour when only whole-activity summaries are available
Science becomes useful when it changes the decision
Evidence → interpretation → strategy → action.
The value of the system is not the number of metrics it can display. The value is in resolving multiple signals into one coherent, explainable coaching direction.
Training looks tolerable
Load is stable, HRV is acceptable and acute Performance Intelligence does not indicate a major operational constraint.
ESPE shows mixed progression
Short-duration performance is improving while longer-duration progression has flattened.
Phase requires consolidation
The athlete may be able to tolerate additional intensity, but the current phase requires absorption rather than more overload.
Protect the adaptation
Maintain aerobic work, reduce unnecessary intensity density and preserve the stimulus required for the next progression step.
The Montis principle
Training is not governed by how much load can be accumulated. It is governed by how the athlete responds, adapts and where that adaptation needs to go next.This is why Montis separates measurement from interpretation, capability from strategy, and deterministic coaching logic from language-model rendering.
Want the implementation detail?
Science explains why. Technical design explains how.
The Technical Design page describes the Montis architecture, Unified Reporting Framework, validation tiers, data contracts, execution flow, APIs, MCP integration and semantic output.