How to Read Divination Analytics
What the charts are showing, in everyday language and in more precise terms.
On this page
Readings and ratings are different counts What you are rating Reading Quality Index (RQI) Sample size tiers Accuracy Trends and the confidence band Stage Delta Trend Metrics The other charts Insight cards What these numbers are notOpen the live dashboard from Divination Analytics. A public description of the same workspace is on the feature preview. How to enter the ratings themselves is in Divination Ratings Help.
Readings and ratings are different counts
Total Readings counts the I Ching, Tarot, and Animal Archetype readings you saved. The accuracy charts count only the readings you later scored. A busy week of readings with no scores will raise the frequency chart and leave the rating charts empty.
Reading totals come from the reading tables. Accuracy, RQI, Stage Delta, moon-phase rating patterns, and trend slopes come from divination_accuracy_ratings inside the selected local-date window and divination type. The time window is applied in your saved timezone.
What you are rating
Each score is your own judgment, from 1 to 10, of how the reading landed. Right after a reading you score resonance, relevance, clarity, and overall value. Later check-ins can add whether it was practical, helpful, or wise in hindsight. These are reflection scores, not a hit-or-miss test against chance.
Immediate ratings store emotional resonance, immediate relevance, clarity and understanding, and overall value. The 2-week stage (short_term) can add practical application and event accuracy. The 6-week stage (long_term) can add helpfulness, life-guidance accuracy, and wisdom integration. The reading screens also offer a separate 6-month stage (six_month).
Stage Delta’s follow-up column uses only short_term and long_term. A 6-month rating stays on its own stage and is not averaged into that follow-up column. Charts that use every rating in the window, including Accuracy Trends and RQI, do include a 6-month rating when one was saved in that window. See how the stage buttons unlock.
Reading Quality Index (RQI)
RQI is one number, from 0 to 100, that blends your rating dimensions instead of looking at overall value alone. Overall value counts the most. On the Accuracy Trends chart the same score is drawn as a dashed line divided by 10, so it can sit on the 1–10 scale with the other lines. An RQI of 78 appears near 7.8 on that chart.
On the 1–10 scale the blend is 30% overall, 15% resonance, 15% clarity, 15% practical application, 10% helpfulness, and 15% wisdom. Multiply by 10 to get the 0–100 index shown on the insight card.
If a later-stage field is missing, that slot falls back to the overall score. Early ratings therefore sit closer to overall value than a fully scored follow-up does. The chart tooltip also shows how many ratings that day rests on, and whether that day’s sample is low, medium, or strong.
Sample size tiers
A striking number based on a handful of ratings can swing the next time you score a reading. The sample label is there so a thin result is not treated as a settled pattern. Low means fewer than 10 ratings. Medium means 10 to 29. Strong means 30 or more.
The insight card uses the count of ratings in the selected window and type. Stage Delta uses the smaller of the immediate count and the follow-up count for that divination type. A confidence band is drawn only when a day has at least two ratings. These tiers are display guidance, not a hypothesis test.
Accuracy Trends and the confidence band
Each point is the average score for that day. The solid lines are overall value, emotional resonance, and clarity. The green shaded band is a plausible range around that day’s overall average. A wide band means the day’s ratings disagreed, or there were only a few of them. A narrow band means they clustered.
The band is a normal approximation 95% interval: mean ± 1.96 × (sample standard deviation / √n), for the day’s overall ratings, and only when n is at least 2. It is not the Wilson interval used on the challenge analytics pages. The dashed RQI line is the daily index divided by 10. Days with no ratings are omitted, so the line connects rated days only.
Stage Delta
This table asks whether a reading still felt as strong after some time had passed. Immediate Mean is the average first impression. Follow-up Mean is the average of the later 2-week and 6-week scores. Delta is the later average minus the first. A positive delta means the reading looked better in hindsight. A negative delta means the first impression was higher than the later one. Delta % is that change relative to the first impression.
Rows are grouped by divination type inside the selected window. Follow-up is the pooled mean of overall_value where evaluation_stage is short_term or long_term. Delta = follow-up mean − immediate mean. Delta % = 100 × delta / immediate mean. The sample label uses the smaller of the two counts.
A 95% interval for the difference of means is calculated when both groups have at least two ratings, using the unpooled standard error. That interval is not drawn in the table. Read the delta together with both sample sizes. A large percent change on a low sample is a hint to keep rating, not a stable shift.
Trend Metrics
Slope / 30d says whether the daily averages have been generally rising or falling. A positive slope is an upward drift. A negative slope is a downward drift. A slope near zero is a flat stretch. R² says how steady that straight-line story is. Close to 1 means the points hug the line. Close to 0 means the line is a poor summary of a jumpy series. Points is how many rated days went into the fit.
The fit is ordinary least squares through the sequence of daily means for overall, resonance, clarity, and RQI. Rated days are spaced as successive points, not as calendar gaps, and the per-point slope is multiplied by 30. “Slope / 30d” is the change across 30 rated days, not across a calendar month that includes empty days. R² is the usual coefficient of determination, clipped to the range 0–1. The fit needs at least two rated days.
The effect sizes under the table are Cohen’s d for one divination type’s overall mean compared with your overall mean in the same window. The type is inside that overall mean, so the comparison is directional and exploratory. The panel shows the three largest absolute values. It is not a claim that one method causes better readings.
The other charts
Reading Frequency
How many readings you did on each day, split by I Ching, Tarot, and Animal Archetype. Quiet days stay on the chart as zeros, so a gap in practice is visible.
Moon Phase Patterns
Your rated readings grouped by the moon phase recorded with the rating. Use it to notice a personal pattern. It does not show that the moon caused the score.
The “best moon phase” card requires at least two ratings in that phase inside the window, then ranks phases by average overall value. Moon-phase effect sizes elsewhere use a minimum of five ratings and the same exploratory Cohen’s d comparison described above.
Divination Type Analysis
Which methods you used, and which ones you have been rating, in the selected window. A favorite type is the method with the most saved readings, which can differ from the method with the highest scores.
Question Themes
An automated grouping of the questions you asked. Treat the labels as a sketch and open the original questions when a theme looks off.
Personal Growth
A radar of your current average scores across the rating dimensions for the selected period. It is a snapshot of the window, not a before-and-after story. The growth-rate card is the before-and-after piece.
Insight cards
- Average Accuracy is the average overall score, from 1 to 10, across rated readings in the window.
- Growth Rate compares the average overall score in the second half of the window with the first half. Positive means the later half scored higher.
- Consistency is the share of days in the window that contain at least one saved reading. It measures how often you read, not how high you scored.
- Avg RQI and Rating Sample Tier are the index and the sample label described above.
Growth rate is 100 × (second-half mean − first-half mean) / first-half mean, using overall value, and it is omitted when the first half has no positive mean. Consistency is 100 × (distinct local reading days / window length in days). Changing the time period or the divination type recalculates every card.
What these numbers are not
Divination analytics describe how your own readings felt and how those feelings changed. They are not a test of hitting a hidden target. Challenge analytics, with baselines, uplift, and p-values, live on a different page and answer a different question. A high RQI does not prove a prediction came true. A rising slope does not prove a skill increased. Both are worth noticing, and both get more trustworthy as the sample tier moves from low to strong.
There is no chance baseline on these panels, because the outcome is a self-report rather than a match to a random draw. Intervals and slopes describe the rating series you produced. They are not a significance test against a null of no effect. For the trial statistics used in psychokinesis and telesthesia, use Analytics Help and Methodology & Evidence.