Three weeks ago you started walking for 20 minutes after dinner. The scale reads 0.6 kg (1.3 lb) less than it did then, but in between there were days at plus 1 kg (2.2 lb) and days at minus 0.8 kg (1.8 lb). You cannot read from those daily values whether the walks did anything.
Experiments and markers in Trendcurve are built for this. You record when a change began and how long it runs, and the app compares the weekly rate of your trend weight before, during and after that phase. Creating experiments and markers is free. The before and after evaluation is part of Pro.
Markers and experiments: two kinds of entries
A marker is a single day with a label: you started a medication, went on holiday, changed your diet. It appears in the chart as a line. An experiment also has an end. It appears as a tinted band over the period in which you kept the change going.
Every entry has a type: Nutrition, Exercise, Alcohol, Medication, Dose, Travel or Other. For each type the app suggests labels, for example "walk after dinner" under Exercise. You can overwrite the label freely and add a note. Markers from older app versions count as type "Other".
How to set up an experiment
Open "Settings & more" and then "Experiments and markers". When you create an entry you choose the type and the label, the start day, optionally a duration (2, 4 or 8 weeks, or your own end date) and optionally a note. Without a duration you get a marker. The "Experiments" card in the Analysis tab lists running experiments with their progress, for example "Day 12 of 28". The history shows measurements, markers and experiments together, each with an icon for its type.
On the morning after the end date, Trendcurve sends a local notification if you have allowed notifications. You can switch it off under Reminders. The next time you open the app, the result card appears.
How to read the result in the trend
The evaluation does not work with daily values. It uses the weekly rate of your trend weight, which is the slope of the smoothed curve. Why that curve is steadier than any single weigh-in is covered in trend weight instead of daily values.
The card divides time into three sections. Before is the 28 days before the start. During is the duration, at most 56 days; for a marker without an end, up to 28 days after the start count. After begins on the end date and exists only for experiments with a duration.
An example. You walk every evening for four weeks and create an Exercise experiment with a 4 week duration. The card shows:
| Section | Weekly rate of the trend |
|---|---|
| Before (28 days) | minus 0.1 kg (0.2 lb) |
| During (28 days) | minus 0.4 kg (0.9 lb) |
| After (14 days) | minus 0.2 kg (0.4 lb) |
Your trend fell 0.3 kg (0.7 lb) per week faster during the walks than before. After the end, the rate fell back about halfway. That fits an effect of the change but does not prove one. Over four weeks the old and the new pace differ by a little over 1 kg (2.6 lb).
When there is not enough data, the app calculates nothing. Each section needs at least 7 days of length and at least 4 days with a weigh-in. If something is missing, the card names the missing amount. Someone who weighed in only twice during the Before phase gets no falsely precise figure, only a note on what is missing.
In practice: weigh in several times a week for the 4 weeks before a planned experiment. Why daily values swing so much, and why that is normal, is explained in why your weight fluctuates daily.
What the evaluation cannot do
The result card carries a fixed note: "association, not cause". It describes exactly what is calculated. The app sees that your rate changed after your start. It does not see why.
A change in the trend may not come from your measure. You may also have eaten or slept differently in the same week. The Before phase may have been unusual, for example after a holiday or a party. The rate changes over time anyway, and in a short period a few days of water retention can shift the slope.
So change only one thing per experiment where you can, and let it run for at least 2 weeks. For the types Medication and Dose: the app gives no recommendations and is not a medical device. Questions about medication or doses belong with your doctor. On the share image and in the notification, these two types show only "Experiment", without the label.
A single experiment gives a hint, not proof. If you test the same change a second time later and the rate shifts again, the hint becomes more reliable.
What is free and what is Pro
Since version 1.5.0 you can create experiments and markers without Pro. Before, that was a Pro feature.
| Feature | Free | Pro |
|---|---|---|
| Create experiments and markers (type, duration, note) | yes | yes |
| Band and line in the chart, entry in the history | yes | yes |
| Notification at the end of an experiment | yes | yes |
| Before and after evaluation and result card | no | yes |
| "Restart trend" switch when creating | no | yes |
The monthly report, also Pro, lists the experiments completed in that month.
Sharing is free since 1.5.0. An experiment result can only be shared when the evaluation is visible, though, and that is Pro. The image carries trendcurve.app. With the Pro switch "Restart trend" you decide when creating an entry that the trend starts over from the experiment.
If you want to start an experiment tomorrow, create it today: pick the type, set the start day to tomorrow and choose 4 weeks. From now on, weigh in on at least four days a week so the Before phase fills up.