As we start 2025 we have big news to share. Before getting carried away, however, we'd like to wish you and your loved ones Happy & Healthy New Year 🎉🍾 🚀 full of personal satisfactions spiced up with a lot of personal records! 🏆
At Spectro we are starting the year with a new hire, and not any hire: @Montis is our first virtual hire powered by Artificial Intelligence and we can safely say that he is the first Trail Running Expert AI Agent globally. 🌐
Montis was built specially for the elite trail runners and their coaches, however it adds a lot of value to the serious amateur trail runner looking to maximize the return of investment into every hour of training, beat of the heart, calorie of energy, gram of sweat and every hope they tie to their objectives. 🎯
What's in it for you
With Montis we are not just copying and improving what other Trail Running analysis software is doing, we are taking a completely new approach.
Some of the major areas completely reconsidered:
System thinking: cardio-respiratory, metabolic-muscular, mental-perception etc. systems and their interactions.
Objective based activity analysis and long-term performance progress. Objectives like target HR (uphill, downhill), effort, walk cadence and a lot more can be attached to an activity and measured with RAG (red amber green) status and actual numbers. Similarly higher level objectives can be set to the training cycle and performance profile so you can progress towards meaningful objectives.
Actionable Insights. We've seen experiments with AI where insights are not meaningful and actionable and we do everything to avoid such use of technology. Specifically sections that need consideration, metrics and trends against understood objectives and established benchmarks, warning and estimates of future Red Amber lane cross of the established objectives and ways to avoid that are just few of the insights Montis provides.
Signal trends like heart rate, effort, power, efficiency, decoupling etc. sliced and diced by segments, categorized by uphill/downhill, walk/run etc. with trends and normalized trends for direct comparison between activities of 3-4 hours to 7-8 or 10-12 hours and even longer.
Multi-dimensional analysis taking slopes, heart rate and fatigue into consideration, even looking at 3D or 4D data for better understanding and contextualization.
Data correction. If you look at various analysis software the effort (GAP) and pace signals are jumping few min/km up and down from minute to minute. Obviously runners don't run like that. That is a direct result of GPS errors. We minimize those errors to offer the most accurate and comprehensive insight onto your performance.
Vertical Ascent and Descent analysis done right, meaning only sections of an uphill or downhill with more than 85% of the slopes in the same category are considered only, removing noise that breaks VAM/VDM calculations.
Optimal decoupling for training. Decoupling is a measure of how much the body struggles to keep itself in balance; when the effort can not follow the heart rate anymore there is decoupling happening, When the decoupling is too large, the recovery from training is too large as well, limiting the amount of adaptation a runner can get from their training cycle.
Holistic approach looking at the performance profile of the runner across multiple activities on multiple systems, metrics and signals.
We know this is not an easy list. That's why we'll be covering these and some more in blog posts during the upcoming period.
For now to illustrate some of the features Montis brings to the world in 2025 we'll take a look at the best Romanian Ultra-Marathon performance of 2024, the winning run of Raul Butaci at Transgrancanaria Classic 2024. We'd like to thank Raul for being so kind to share some of his metrics from this run with the public and wish him good luck to the upcoming season.
Race Analysis, TGC Classic 2024, Raul
While the above is just the content for the Activity Analysis report, Montis can perform a full Performance Progress report on the runner's profile. We are not going to cover all the 50 pages of this report either, just some of the most important and publicly sharable aspects of the TGC 2024 win.
The stable basis for every data analysis is the right preparation of the data. As an example below, we categorize every uphill and downhill into how much actual uphill and downhill is within it. A D100 segment is a downhill where more than 95% is actually descent, a D90 is more than 85% while an U90 is more than 85% ascent. We consider U90 and U100 for Vertical Ascent (VAM) analysis and D100 and D90 for Vertical Descent (VDM) analysis.
This results in the correct VAM/VDM numbers and trends on each uphill/downhill and throughout the race. Taking into consideration the technicality of the track, the numbers are quite impressive. But even more impressive are the effort, efficiency and decoupling trends within Raul's winning performance.
But let's start with the technicality of the track. The Ground Contact Times average at a staggering 313 ms, which is a lot for an elite runner who usually hits the ground with less than 200 ms. The walk part averages a 446 ms and is very stable with 6.9% trend. We know that in such conditions it's less the run, but the walk part which makes or brakes the race, so this is exactly where Raul performed at its best. If we plot his walk cadence against objectives we have:
The cadence of 109 steps/min meets the 110 steps/min objective and even improves during the race with 3.2%. This performance puts Raul in the same league as Kilian and Francois during the UTMB 2017 which we'll analyze later, where Jim did not manage to keep up exactly due to lack of cadence during walk and needed to run where the others walked.
Now if we look at the effort Raul exerted during the race, we can say that it was exceptionally well paced. He was pacing using perceived effort and power mainly and was not even in the leading pack in the first climbs. We'd suggest him to hold back even a little bit more during the first 2-3 hours to be able to go even stronger in the last 2/3-rd. The target effort objective is set based on the optimal effort Montis calculated.
Looking at how steady the effort is and mostly within the Green lane of the objective, it is not surprising that the efficiency is also very well managed.
We calculate efficiency based on effort over heart rate. While we won't go into detailed heart rate analysis, we can show the decoupling on a high level which just shows that even with some heart rate dropping, Raul's system was almost perfectly in sync with his effort, meaning he was not really exhausted after the race itself, it was actually a bit more than a good training for him. A good training as the year's best run? Huhh 🤔
If we look at each slope and the efficiency on them it strikes us how evenly can Raul pace himself on every gradient and modality be it run or walk. The numbers line up almost perfectly, with a slight decrease in the 35-50% gradients and almost negligible differences between run and walk. Remember, that efficiency is displacement over heart rate beats, so while in absolute numbers the pace is higher, more energy is exerted as well, emptying the tank faster, so a correct balance is of utmost importance for any performance, let alone this winning one.
We could go on and on with the details, but we hope by this time you got an initial glimpse into what Montis can do for you and if you are interested reach out to us until we bring Montis to the public.
Creating a baseline of your best performance, understanding your worst performance, analyzing a sequence of runs in your training block or a number of races during your season or even just some regular runs to understand your progress, these are all very good use-cases and a lot of insights you can extract with the help of Montis for your future objectives.
Stay tuned for future blog posts on the topic and subscribe to our Facebook, Instagram, Youtube, LinkedIn, X and our Newsletter for timely updates on the topic.
Have Fun, Health and a Fruitful season ahead!
The Spectro.AI, Spectro.Life Team
PS. If you'd like to repeat the performance above, shoot for these numbers:
Comments