TL/DR; –

Examples of a quick start set up, leaving most settings at default. Let’s say you’re running a 50 mile race. You’re targeting a finish time of 12 hours. You think using a running average pace for the last 30 minutes of your run is a good average to predict your current finish time. The items that are bolded are critical, these HAVE to be set correctly to get valid data from the data field.

Open the Garmin Mobile Connect app on your phone, find the datafield in the watch section and go to Settings –

Predict using last XX minutes = 30
Update every XX seconds = 15
Select Race or [Use Custom] = 50
Use Custom distance if > 0 = 0

Custom dist is Kilometers = Doesn’t matter
Show timer when not moving = Check if you want this feature
Show Total Non-Moving Time.. = Check if you want this feature
… when > (finishtime * x%) = Necessary if you want total non-moving time. If you’re not sure what a good value is then just set it to Always.
Show too fast/slow warnings.. = Check if you want this feature
…(finish * %) or XX mins = Necessary if you want to see too fast/slow. If you’re not sure then set it to 3-6% as a general rule of thumb.
Target Finish Time – Hours = 12
Target Finish Time – Minutes = 0

On Pace Background Color = Your choice
On Pace Text Color= Your choice
Too Fast / Too Cold colors = Only used if you’ve turned on too fast/slow warnings.

Now let’s say you’re running a different race, it’s a odd race distance, 87 Kilometers, that you want to finish in 11 hours and 20 minutes. You want warnings if your current predicted finish time is more or less than 30 minutes off your target finish time. You want walks to filter out of your estimated finish time faster and a faster update cycle. Items in bold are critical to setting up this race for you.

Predict using last XX minutes = 15 (this lets walks ‘drop out’ of the calculations faster)
Update every XX seconds = 5
Select Race or [Use Custom] = 50 (doesn’t matter)
Use Custom distance if > 0 =
87
Custom dist is Kilometers = Checked
Show timer when not moving = Check if you want this feature
Show Total Non-Moving Time.. = Check if you want this feature
… when > (finishtime * x%) = Necessary if you want total non-moving time. If you’re not sure then just set it to Always.
Show too fast/slow warnings.. = Checked
…(finish * %) or XX mins = 30

Target Finish Time – Hours = 11
Target Finish Time – Minutes = 20

On Pace Background Color = Your choice
On Pace Text Color= Your choice
Too Fast / Too Cold colors = Only used if you’ve turned on too fast/slow warnings.

And now for something completely more detailed –

On this page you’ll find the instructions for how to set up the UFTP datafield I’ve written for my own and Bunny’s use. This datafield is something that didn’t exist so I wrote something to fill my needs. It uses a longer rolling looking back window from the present to calculate the finish time for a race than currently exists. It’s primarily focused on longer races so it doesn’t have the granularity that someone might need for something like a 5K. It’ll work but only down to the minute type predictions. There are other finish time predictors that work better for this type of high intensity race.

It does this by recording an array of elapsed time and distances instead of using an array of average pacing. This gives a more useful average for my needs to accommodate walking, aid station/bathroom breaks etc.

In addition it can provide warnings if your finish time is too far above or below your target time which is useful for us to tell us to slow the eff down.

Non-Moving time is another problem for ultra marathons. It can sneak up on you if you’re not paying close attention. 2 minutes here, 3 minutes there and by the time you’ve run 100 miles you’ve got 3 hours of non-moving time. So as added features it can display a timer when it detects you’re not moving and as well show total non moving time. It’ll flip between these data points in the lower third of the display if they’re relevant and active.

  • 1 – Set this field to the number of minutes you want to look backwards to use as an average for the finish prediction. *see below
  • 2 – How often you want the prediction to update. This may be adjusted internally if it’s too often for a long average time to reduct memory usage. *see below
  • 3 – Select from one of the more common race distances OR
  • 4 – Set this to a non-zero to use a custom distance. It’s important to note that if this is not 0 then the Select Race value is ignored. 
  • 5 – Is the custom distance in kilometers? Check this box. 
  • 6 – Check this box to show a count up timer when the datafield detects you’re not moving. This will disappear when you start moving again. 
  • 7 – 8 – Show a running count of non-moving time over the entire race… But only when the total non-moving time is > than this percentage of the finish time. You can select Always instead of a % to have this always shown. 
  • 9 – 10 – Change the colors (if set) and display a notification if you’re too far ahead or behind your target finish time…when your finish time is either > or < this % of your target time or your finish time is > or < this many fixed minutes. 
  • 11 & 12 – Set the target finish time for your selected in hours and minutes. 
  • 13 – 18: These set the colors of the datafield. On pace is the default coloring when you’re on target to finish in the time you set. Too Fast is the coloring for when you’re ahead of target. Too Slow is the coloring used when you’re behind your target finish time.

(*below) – There are some technical limits on the values that can be used for the Prediction and Update values. The higher the Prediction and the shorter the Update values are, the more memory that is consumed by the datafield. Older Garmin watches have a very finite amount of memory that is availalble for datafields.

As a result in order to help insure the datafield will work on older watches the datafield will change the Prediction and Update values if it thinks they’re too high. These values are adjusted as below. In general these adjusted values should update the prediction with enough frequency to make the values not lag too far behind reality while keeping the memory usage to between 22K- 24K typically and under 24.5K in even the worst situations. All the supported watches reportedly have at least 28.6K of usable memory. All memory usage is pre-allocated at activity start so there should be no increase in memory usage over time even for the longest runs.

  • Prediction is 120 minutes – Update is adjusted to no more than once every 45 seconds
  • Prediction is 60 minutes – Update is adjusted to no more than once every 20 seconds
  • Prediction is 30-50 minutes- Update is adjusted to no more than once every 15 seconds.
  • Prediction is 20 minutes – Update is adjusted to no more than once every 10 seconds

You can of course set the update cycle to higher values than these if you want to save some memory or you don’t need it to update so frequently. This will also in practice save a fractional amount of battery as the less often the datafield has to compute the prediction the less CPU cycles it uses and thus the less power it draws.