The Skeptics Guide To Sports Science

Running the numbers before a game doesn’t save you

I spent years watching coaches pull out clipboards right before kickoff and pretend that some regression model would keep their running backs from breaking hamstrings. The math itself isn’t the problem. It’s the people who think a spreadsheet replaces film study. You can fit every metric into a dashboard and still end up with an ACL tear on third down because someone trusted the wrong variable. That’s why I started writing down what actually works when you strip away the hype and look at the raw data.

The Skeptics Guide To Sports Science

This isn’t a philosophy book. It’s a working manual for anyone who’s tired of hearing “trust the process” while the evidence says otherwise. I use it every season when the equipment managers send me load-monitoring spreadsheets that look impressive but mean nothing. The core idea is simple: question every claim that comes wrapped in jargon. If a study claims a supplement reduces injury risk by forty percent, I want to see the confidence interval, the sample size, and whether the control group actually did anything different. Most of the time they didn’t.

I learned this the hard way back in 2019 when a vendor pitched a neuromuscular training program that promised to cut non-contact knee injuries in half. The brochure had charts. I asked for the raw dataset. They couldn’t produce one. The “study” behind it had twelve subjects, no blinding, and a control group that also started stretching more because the researchers were excited. I walked away from that contract and went back to basics: progressive loading, adequate rest, and video review of technique. Six months later, our injury rate dropped by twenty-two percent without a single gadget. That’s the pattern you want to recognize.

How to actually use the framework

Start by collecting your own numbers instead of trusting a consultant’s slide deck. I keep a simple log for every athlete: sleep hours, RPE scores, jump height trends, and any pain reported on a scale from one to ten. You don’t need fancy software. A spreadsheet and a stopwatch will do. The goal is to spot deviations early, not to celebrate averages. When I see a player’s countermovement jump drop more than five percent over two weeks without an obvious reason, I pull them from contact drills until the number recovers. Sometimes it takes three days. Sometimes it takes two months. Ignoring the trend costs us more than sitting them out.

The hardest part is admitting when your favorite tool fails. I used to swear by blood lactate testing before competitions. The data looked precise. In practice, it barely predicted performance and caused more anxiety than advantage. I switched to rate of perceived exertion and simple sprint times. The correlation with actual game output improved immediately. Lactate thresholds vary wildly between individuals and even within the same person on different days. A number without context is just noise.

Common pitfalls that beginners miss

Most coaches focus on the wrong metrics. I’ve seen programs optimize for vertical leap gains while ignoring ground contact time. The athlete jumps higher but pushes off slower, which translates to zero improvement in acceleration. I track both numbers now and weight them equally in my evaluation. Another trap is over-indexing on lab-based VO2 max tests. The values look authoritative but correlate poorly with sport-specific endurance. A player can score eighty-five milliliters per kilogram per minute and still gash out in the fourth quarter because their energy system doesn’t match the demands of the game. I use field-based shuttle runs instead and compare them to historical baselines.

I ran into a particularly annoying edge-case last season when a wearable device claimed to predict muscle damage better than blood markers. The marketing was compelling. In practice, the sensor placement mattered more than the algorithm, and the “accuracy” dropped below random chance when tested against actual MRI findings. I stopped relying on the device for load decisions and went back to subjective soreness scales and range-of-motion tests. The workaround was simple: any metric that can’t be verified through direct measurement gets flagged for review. I kept a separate log for “unproven” devices and never let them influence playing time decisions.

When to walk away from a method

Not everything that promises results delivers. I once watched a program spend six figures on force-plate technology that claimed to prevent hamstring strains. The plates looked scientific. The data showed interesting patterns. In reality, the correlation with actual injury occurrence was weak, and the cost per prevented injury exceeded the budget for physical therapy alone. I pulled the funding and redirected it to strength coaching and video analysis. Twelve months later, our hamstring injury rate dropped by eighteen percent without a single plate. That’s the pattern you want to recognize: simple, verifiable interventions beat expensive gadgets every time.

If a tool requires more than twenty minutes to interpret, it’s probably not worth the overhead. I keep a strict rule: if I can’t explain the result to a player in two sentences, I don’t use the metric. Complexity without clarity is just a way to hide uncertainty. I recommend starting with basic tracking—sleep, RPE, simple performance tests—and only adding complexity when you have a clear reason. Most programs never reach that point because they confuse activity with progress.

Building a sustainable system

The goal isn’t to collect data for its own sake. I’ve seen programs generate terabytes of information and still make the same mistakes year after year. The key is to use numbers to ask better questions, not to replace judgment. When I see a player’s sleep score drop below five hours for three consecutive nights without an obvious reason, I adjust their training load immediately. Sometimes it takes one practice to recover. Sometimes it takes a week. Pushing through the deficit costs us more than resting.

I keep a personal anecdote about the time I trusted a predictive model over my own eyes. The software claimed to forecast ACL tears with eighty percent accuracy. I watched a player’s landing mechanics deteriorate over three weeks and ignored the model because the “risk score” stayed green. Two days later, he tore his ACL on a routine cutting drill. The model had weighted the wrong variables and missed the mechanical warning signs entirely. That’s why I verify every prediction through direct observation. Any tool that can’t be validated through practical application gets relegated to supplementary status. I keep a simple log for “unproven” methods and never let them override common sense. The framework works because it forces you to confront what actually matters instead of what sounds impressive.