The Edge of the Envelope: Using a performance model week to week.
Understanding how to use a performance model to plan and track your training.
In my last “Edge of the Envelope” post, I discussed the parameter that governs performance models — Threshold. In my post on training plans, I allude to performance models as a tool for weekly planning. “Use a Performance Model” was number three on the list in my “what works” post after ride your bike and get a coach. In this post, I explore the question what is a performance model and how can we use it in planning and tracking our training?
While this post is primarily targeted at cyclists, the same concepts hold for runners.
A performance model attempts to represent short term training load in the context of long term training load giving an indication of freshness or fatigue.
A performance model takes the training load from every workout you do and adds it to two running averages — one long term and one short term. There are a number of terms™ used but for the sake of sanity let’s use the Training Peaks nomenclature here. The long term average is “chronic training load” (CTL), the short term average is “acute training load” (ATL), and the difference between the two is “training stress balance” (TSB). Every day, an estimate of training load is entered and the averages and difference get updated. If you want, you can estimate training load into the future and play around with what ifs about how your training might play out. Now you are a performance modeler… Neat, huh? [Find a sample performance model spreadsheet to play with here.]
In general, with structured training, you’ll see patterns like the one from my spring 2019 above. Hard training will spike your ATL and depress your TSB with minor rises in your CTL. Lighter training (also known as the work week) brings ATL back down toward CTL and TSB back toward zero. A common approach is to throw in a hard interval workout mid week when fresh then do a big weekend — shower, have a mineral water, and repeat. Clearly, this isn’t going to work every week and you might prefer a different rhythm, but it does a good job illustrating the kind of pattern you can track with a performance model.
In planning, the model can be used in two ways.
- To ensure adequate rest for effective training without losing fitness.
- To ensure training load results in steady gains without over doing it.
These two are fairly nuanced so I’ll break them down in some detail separately.
1: Balancing rest and work.
You just finished a tough Sunday ride, you are sipping a mineral water thinking about the training week to come. Should you do intervals on Tuesday, Wednesday, or Thursday? If you do the Tuesday night group ride, will you be ready to get a quality workout by Thursday morning? You have a race next Sunday — how much training can you do this week without hitting the race too fatigued?
These are the kinds of questions you can answer using a performance model.
In a typical week, you might go for a couple big rides on the weekend — say they are both double your long term average. In this case, on Monday, you’ll be feeling pretty fatigued. Your performance model will indicate this by showing that your TSB is negative and your ATL is kinda high relative to CTL. If you tried to achieve anything like peak performance in this state it wouldn’t go well. So you have to rest. Looking forward, you can estimate training load for each day until your TSB isn’t so negative and ATL is down more in line with long term average. This is the core function of a performance model on a weekly basis.
Without the performance model to guide you, you might plan your week around an ambitious assumption of recovery or go for too much of a “recovery ride” thinking you will recover anyways. On the flip side, you might also mistakenly rest a bit too much, meaning you blow some hard gained fitness getting super fresh for your weekly interval session. While this will always be error prone, model or not, a performance model provides information to aid decision making where none exists otherwise. You have to use it with an understanding that it is as imperfect as you are, but you can also use it knowing it’s the best you’ve got and may as well be relied on for guidance.
2: Steady sustainable gains
We’ve all done it. Gone out at the end of a hard week of training and ripped too much of a ride with too little physical durability and stamina to back it up. You come home with cramped calves, tender patellar tendons, and a lightly bruised ego. Alternatively, you might focus on hitting your workouts fresh to make sure you build top end fitness for race season only to get to races missing the fitness needed to make the front group.
Balancing these two realities is not easy — some data and a good predictive model can be very helpful.
Looking at a performance model when planning your week of training allows you to make sure the week adds up to maintaining or building relative to your long term load. It’s pretty simple — does CTL go up or go down from Sunday to Sunday? If you keep a performance model over the long haul, you can start to get a feel for how fast you can build your CTL without it being too much. If you care to look at things longer term, you can even plan weekly training load weeks and months into the future to get you to a CTL target a few months away. Given that, you can back out training time and design in periods of heavy or light training based on your life and/or competitive goals.
A performance model, if kept up over the long term, is a great source of data to understand your personal limits for both chronic and acute stress. That is, you can easily look back at your biggest hardest days and quantify how big they really were so you can design future training accordingly. In the same way, you can look back at big training blocks, see how the performance model looked then and plan future training blocks around what you learned. This could be accomplished with a basic training diary, but having the ability to quantify it and compare acute load to chronic load is a powerful tool.
Reality Check
Let’s be real, just riding your bike and maybe hiring a coach are way more important factors to success than keeping a performance model and using it to plan your weeks. This data-driven approach to tracking and planning training is just that, a data-driven metric you can add to your training diary tool kit. But if you are training with power or even just heart rate, and have some idea of your threshold, you’re collecting all the data you need to use a performance model and they offer insights you can’t get any other way.
I’ve prepared a sample Google sheet that you can play with to start your own performance model here. I left my data for this spring in there to get you started. Copy the sheet into your own user and go nuts.
And remember the infinite wisdom of LeVar Burton,
“But you don’t have to take my word for it.”
