research

training radar

The training radar is a compact public-profile summary. It is not a readiness score, coaching prescription, or ranking system. Each axis converts one observable part of a user's logged training history into a 0-12 score so visitors can scan training shape at a glance.

1. Axis definitions

AxisInputScoring rule
StrengthHighest estimated one-rep max from any weighted set.The estimate is capped at 700 lb, then scaled to 0-12.
ConsistencyLogged training days compared with expected active split days since joining logit.100% consistency maps to 12/12.
FrequencyAverage workouts per week across the last eight weeks.Six workouts per week maps to 12/12.
VolumeAverage weekly training volume across the last eight weeks.80,000 lb per week maps to 12/12.
VarietyDistinct exercises logged across public workout history.Sixty distinct exercises maps to 12/12.
ExperienceAccount age and total workout count.Half of the score comes from days on logit, half from workouts logged.

2. Normalization

Every axis is rounded to the nearest whole number and clamped between 0 and 12. This keeps the chart stable and comparable while avoiding false precision. Values above an axis cap remain at 12 rather than stretching the scale for everyone else.

3. Recent-window metrics

Frequency and volume use the most recent eight Monday-first training weeks. That keeps those axes responsive to current behavior without erasing longer history for strength, variety, consistency, and experience.

4. Profile consistency

Consistency is a profile-level attendance ratio. logit counts how many distinct calendar days have at least one logged workout, then compares that against the number of active training days implied by the user's public split and account age. This avoids rewarding someone for simply having an old account, while also avoiding the impossible standard of training every calendar day.

Dlogit=max⁡(1,  days since signup)D_{\mathrm{logit}} = \max(1,\;\text{days since signup})Dlogit​=max(1,days since signup)Aweek=active split days per weekA_{\mathrm{week}} = \text{active split days per week}Aweek​=active split days per weekDexpected=max⁡ ⁣(1,  round⁡ ⁣(DlogitAweek7))D_{\mathrm{expected}} = \max\!\left(1,\;\operatorname{round}\!\left(D_{\mathrm{logit}}\frac{A_{\mathrm{week}}}{7}\right)\right)Dexpected​=max(1,round(Dlogit​7Aweek​​))Dlogged=∣{workout dates with at least one log}∣D_{\mathrm{logged}} = \left|\{\text{workout dates with at least one log}\}\right|Dlogged​=∣{workout dates with at least one log}∣consistency=min⁡ ⁣(1,DloggedDexpected)\mathrm{consistency} = \min\!\left(1,\frac{D_{\mathrm{logged}}}{D_{\mathrm{expected}}}\right)consistency=min(1,Dexpected​Dlogged​​)
Interpretation. A five-day split on a 70-day-old account creates about 50 expected active days. If 25 distinct days have logged workouts, profile consistency is 50%, and the radar axis receives 6/12.

If no public split is available, logit uses seven active days per week as the fallback. That makes the score conservative until the profile exposes enough schedule context.