Have you heard someone boasting about there latest FTP result, which works out at over or 5w/kg, or a an amazing >20W/kg peak power result? Unfortunately, for most amateur athletes, misuse of data is an easy pitfall, while effective use of data is a much less utilised tool.
If you think you may be one of those people, cherry picking data results, then here’s is this article boiled down into simplicity: make sure you are using ‘in’ and ‘out’ data. Click on the links to see what I mean. if you train with power then have a look at 6 simple bits of data with little time or cost.
6 Data Streams for little extra time or cost:
- Heart rate data- Heart rate data is an essential in critically analysing other measurements.
- Speed – GPS data is not accurate enough for monitoring speed.
- Cadence- you power meter may measure cadence, but not always accurately particularly for high cadence. Make sure you have accurate cadence and speed.
- Consistent feel good measurements- sleep, feel good score in training, alertness and soreness.
- bike setup- I have heard of a number of people who regularly pay for bike fits but arent given an ongoing log of their geometry. The bare minimum should be a before and after log of the bike fit session, but bike setup can correlate to future injuries or loss/gain in performance, so why not have that information. New bike with a shorter stem and narrower bars, the measurements from your previous setups should be the starting place for your new one, not some random numbers based on what setup looks cool.
- Race weather- wind direction/speed, rain/dry, temperature
- Record of tactics- Your result, how the race was one, time gaps at the finish. The data should be recorded in a consistent way collecting information specific to the rider.
Most athletes that i have coached, i have to regularly take a critical look at their data. Anything from descrepancies between power measuring devices, to calming worries about unuasually high heart rates.
The data revolution is here! or is it?
Team Sky were apparently the first to use data, but its actually a staple of Elite track cycling. My favourite cyclist (and massively underated) Graeme Obree used power metering in its simplest form, using his turbo trainer with a speed sensor carefully monitoring for temperature and tyre wear to ensure a good correlation betweeen speed and power. At the same time Chris Boardman was training using early power measuring devices. Aside from power metering there were also, track times were a good bit of intrinsic output data. In road cycling data has taken longer to catch on
Data appears to dominate high level cycling. Team Sky attribute their success to a marginal gains approach in which, everything from weight management to, segment speeds are monitored and carefully managed. Without electronic data, this sort of monitoring would take reems of hand notes and calculations. So should you be using data better in your cycling? In short my opinion is it’t not the be all and end all, for self coaching if you want to use data, start with ‘out’ data (results), if you are coached, make sure data is not a replacement for coach-rider communication or impinges on individualism.
I want to go through a few questions, looking at how data is used in both Elite cycling and amongst amateurs.
- What’s cycling data?
- Why think about your performance philosophy before using data?
- What’s the benefit of data?
- What are the limitations?
- The data revolution, descriptive, predictive, presriptive.
Data is useful for managers working with teams of athletes, when there is a large number of variables which affect results. For beginner and intermediate cyclists (~1-5 years experience) your performance could be mostly reliant on one or two variables. Without the help of some guidance on using data, it’s easily to overcomplicate your preparation and possibly overlook the key insights which will help you to up your game. If you want to use data consider both internal and external data, and all the different variables which will be monitored.
What’s cycling data?
Data is information that can be systematically recorded in electronic form. Nowadays the first thing that comes to mind is power data, measurements made with a power meter on a bike, but it can also include:
- Physiological measurements, heart rate, VO2max, lactate levels, peak load tests, neurology
- Psychological observations
- Racing/training speed/time/distance/cadence
- Observations of technique and tactics (e.g. skill sports)
- Equipment set up (e.g. tyre pressures)
- Results and performance analysis
What are the benefits of data?
The main function pf data is objectivity, which makes it useful for professional cycling, results should be proportional to investment. In elite cycling data can be used to:
- Reduce error
- Simplify and constrain a complex model
- Team management
- Cost vs benefit analysis
- Performance prediction
- Quantify and corraborate existing knowledge
The common dominator in using data is scale. Professional teams are dealing with large numbers of individuals.
I define ‘in’ data as the data directly connected to the rider and their intrinsic ability, but disconnected from results and competition. (all the things underlined in the list below).
‘out data’ is the data connected the rider’s outcome goals, this data is usually a lot more complex and encompasess other variables. But some of these variables can be measured, e.g. ground conditions for an MTB race, tyre/equipment used, weather condition. Most importantly a underused data source is the competition.
One of oversight when peopple approach data is to not pay enough attention to external data. Data can measure both internal and external variables with respect to the athlete. Internal variables are those attached directly to the athlete, e.g. power, weight, sleep etc, external variables are things which might affect an athlete but are detached, race conditions, temperature, competition. People often try drawing conclusions from race results and how they relate to their phsyical fitness without looking at the data in fullness.
How do training tools fit in with this?
There are various online or computer programs which will process data uploaded from a bike computer, Training Peaks, Strava, Cycling Analytics, Golden Cheetah. If you use a bike computer you might as well upload your data to one of these, it allows you to quickly workout when you did certain training and can be a good overview of a long period. Training Peaks use various more advanced methods to process power data, which can give you a more stastically reliable idea of your power profile. However, the focus on these tools is analysing data from measuring devices, power, heart rate, speed, elevation, distance. Looking at the list above this excludes quite a few possible variables. One big one is the analysis of results. Despite the ease with which you can upload data, there are no tools to analyse competitions and results.
What are the downsides of data?
The flip side of error reduction is restriction of insight. (Gary Klein, ‘seeing what others dont’). I think this is largely the reason why there is hostility to Team Sky’s approach. Cycling is an emotional sport. Many of the cycling legends have been individuals fighting it out under their own steam. Its not likely football, indivuals are worth something. Team Sky are more like a Roman Army honed and repetitive. However in the case of Team Sky’s success it came from the insight of one person, who was willing to pool from the talents of individuals. Brailsford, saw the potential of talented staff and riders. On paper woulda team select someone like Froome now? he was incosistent.
Without insight data can lead to drudgery, there needs to be flare room for individualism.
What is your philosophy with data? keep it simple stupid? marginal gains?
The problem with the Team Sky marginal gains approach is that it is expensive. It requires 1000s of man hours, and high level expertise. Its not a something that can be DIYied!
From a coaching perspective, my philosophy is that power data can be useful for some analysis, goal setting and training prescription, but the data is not a predictive tool.
My philosophy with respect to my preparation, is that if im going to use data i need to use ‘in’ data and ‘out’ data. ‘in’ data are the measurements of my performance outside of results, ‘out’ data is results based data. My experience of data, is that i started off thinking that data was everything. Power tests, benchmarks. But then soon realised there is a limit on my time.energy and that there was much more useful data. Namely, what I am doing and how I feel. I have found my training record’s in a spreadsheet, invaluable. I can flick through 6 months of records and take note of when i was on holiday, or when i had a particularly cold/wet weekend coaching or riding, or even when i couldnt be bothered to fill in the record. I have learnt to take note of the things are specific to me, when i am getting tired in the afternoons and when i am getting back/knee/hip pain, or when cycling feels particularly effortless. I can see what contributed towards a good race and what was hard.
How do i use data as a coach
One of the ways i try to stand out from the average data user, is a good understanding of maths, statistics and probability. I have a strong foundation in computation and analysis methods from my degree and engineering work, which helps me to understand how certain data analysis models work.
These are some of the open ended questions that I feel i have a good judgement over most self-trained cyclists.
- is your power meter accurate?
- how much should i take from a single test result?
- Whats the reliability of a trend in race results?
Often I come accross one very simplistic assumption when it comes to the use of data: Phsyiological performance is measured by power, physiological performance is proportional to results. There are some disciplines in which this may work: individual road (TT) and track events. or for very experienced/succesful riders who have honed down their preparation such that most of it relies on phsyical preparation, and who are aiming to maintain performance rather than exceed.
What are the limitations of data?
Data is like painting by numbers. Constraints are laid on the picture in order to make to simplify it, whereas without data it’s like painting a water colour. So a data based model can be applied to a situation where there is a large number of repeated variables (such as a team), but in the context of an individual a model founded on data is like to be too cumbersome and possibly innacurate.