January 14, 2026
Brains vs Gains: place your bets
Training my smartwatch to track intelligence
REM sleep wins, gym days lose, and the crowd fights over sleep trackers
TLDR: A chess fan built a Garmin watch app that predicts his mental sharpness from sleep and exercise, finding REM sleep and moderate stress help while recent workouts hurt short-term thinking. Comments split between “sleep trackers feel fictional,” “don’t let stats jinx you,” and “watches are uncomfortable—give us rings.”
Welcome to the brain-versus-body cage match. One chess-obsessed Garmin user trained his watch to predict “smart day” odds from sleep and exercise, then checked it against daily wins on Lichess. The surprise: REM sleep and stress (Garmin’s “stress” tracks low HRV—heart rate variability) boosted thinking, while recent workouts tanked chess performance. Deep sleep didn’t move the needle. Translation: short-term mental sharpness isn’t always buddies with long-term fitness, and your Garmin may rate “relaxation” while your brain wants that alert buzz.
The comments went full soap opera. Skeptics shouted that sleep scores “feel made up” and never match vibes. A cheerleader begged Garmin to make it official. A reality-checker warned the author’s morning stats might prime a loss—hello, self‑fulfilling prophecy. Then came the comfort wars: “How do people even sleep with a watch?” Team Ring chimed in, demanding a no‑subscription option. People joked about “algorithmic rebellion,” a “smart‑dumb dial,” and REM as “brain Wi‑Fi.” Hot take central: stress isn’t evil—it can be focus mode. Others asked if the model just predicts “don’t play chess after leg day.”
Verdict: an entertaining DIY brain meter, but the crowd is split between data believers, vibe loyalists, and wrist-haters. Plus, many want a Garmin ring.
Key Points
- •The author combined 1.5 years of Garmin physiological data with Lichess ELO changes to model daily cognitive performance.
- •Logistic regression predicted chess win/loss with ~60% accuracy, confirmed by cross-validation, outperforming a ~50% baseline.
- •REM sleep duration and Garmin stress metrics (inverse HRV) positively correlated with better chess outcomes.
- •Active calories and light sleep duration negatively correlated, while deep sleep and sedentary duration showed no effect.
- •A custom Garmin watch app was built to compute daily probability of ELO increase and update the model weekly.