Why did your Tesla lose 6% overnight — and what will it lose tonight? Pick tonight's conditions and get the same forecast the app computes from recorded nights: a quiet-night baseline plus what each factor adds, learned from the nights themselves. Click any night below to see why it lost what it lost.
DEMO DATA — pretend car, pretend nights. This page never connects to a real Tesla.Method: baseline = median loss of quiet nights (no factor on) over the last 14 nights. Each factor delta = median loss of nights with the factor minus median loss of nights without it (both groups need 2+ nights, otherwise a transparent default: Sentry +1.5, cold +1.0, Cabin Overheat Protection +0.5, software update +1.0, plugged in −0.3). Forecast = baseline + tonight's deltas, clamped to 0–15%.
| Night | Bedtime % | Morning % | Lost | Factors |
|---|
Try it: the worst night is Oct 21 — Sentry on AND a cold night, 4.6% lost. Turn Sentry off in tonight's forecast above and watch the prediction drop by the learned Sentry delta.
The full app (github.com/rtmalikian/tesla-drain-forecaster)
records real nights from your Tesla (or ones you type in), keeps them on your own
computer, and exposes an OpenAI-compatible forecast API
(/v1/chat/completions, /v1/models, /v1/embeddings)
so any AI client can answer "predict tonight's loss" or "why did I lose 6% Tuesday?"
from your own numbers. It never wakes or commands a car and sends your data only
to the AI backend you configure — or nowhere at all, with the built-in rules engine.