Ashley
Candlestick Chart

Let's quantifies every single message, converting flirt, coldness, probes, conflict and repair into a relationship index, then aggregates them into OHLC candlesticks at multiple timeframes.

You can see how a relationship warms up, pulls back, probes, and repairs — and the follow-through and selling pressure after each move.

START

History

No analyses yet. Your first run will show up here.

Only analyses observable interaction signals · does not predict relationship outcomes · does not judge the other person’s true intent

← Back

Online analysis · Configure job

① Chat file

Absolute path to chat file File format

② Model selection

Provider Each provider uses its own SDK or an OpenAI-compatible endpoint. Once you pick one, the official base URL is filled in automatically. Model Custom model ID If you picked “Custom” in the Model dropdown or “Custom” provider, type the model id here. API Endpoint (base URL) Known providers auto-fill the official base URL. Override here if you use a proxy / mirror / self-hosted endpoint. API Key

③ Advanced settings (defaults are fine)

Initial index default 100 Starting height of the candlestick path. It only shifts the whole chart up or down; it does not change the shape. The model is multiplicative, so any positive value works — 100 is the default; pick higher (e.g. 200) if you want a chart that already starts warm. Batch size default 50 How many messages per LLM call. Larger = cheaper and faster, but one failure loses the whole batch. 50 is the sweet spot. Context window default 30 How many already-scored prior messages (with their scores and rationale) the model gets to see, so it can judge the current message’s tone relative to them. 0 = no history; the model only sees the current batch. Calendar mode Active periods only: skip days/weeks with no chat (finance-chart convention, compact). Every period: draw a candle every day/week, flat carry-forward when silent (useful to visualise long silences).

Analysing…

Stage: —

0 / 0

Job ID: —

Status: —