The VeloViewer Score is a composite measure providing a single value that represents your best Strava achievements. This value can then be used to benchmark yourself against any other VeloViewer user around the world. In this post I’ll explain how it is calculated and where your Score fits in against other VeloViewer users.
The VeloViewer Score is the average Position Score of your top 25% of your non-descending, scoring segments maxing out at 100 segments. A Score close to 100 is good, close to 0 is bad!
Segment Position Score… eh?
Your segment Position Scores have been in in VeloViewer for quite some time now and provide a value that represents your relative position to the other riders on each segment. This is using the exact same scale as the VeloViewer Score so close to 100 is good, close to 0 is bad. The formula used to work out this score is:
Position Score = ( [K/QOM Athlete Count] + 1 – [Your Position] ) * 100 / ( [K/QOM Athlete Count] + 1 )
Because we add 1 to both of the athlete counts in the equation it is impossible to actually get a value of 0 or 100, this is because not all KOM’s are worth the same. If you are in 1st position on a leaderboard then your score will vary depending on the number of athletes: 50 when you are the only person to have completed the segment as you are actually equally in last place as well as first. The same applies at the other end of the leaderboard, being last of 3 people isn’t as bad as being last of 1000. Here are a few example scores:
Note: being 1st out of 100 gives the same score as being 100th out of 10,000.
Why use only 25% of my segments (up to 100)?
I don’t take into account all of your segments as I just want to use the scores from segments where you’ve actually had a proper dig at getting a good time. This also lets you focus on the segments that suit your strengths as well. e.g. if you’re a Rouleur then just target those flat segments and your slow climbing times don’t affect your Score.
To stay on the right side of the Strava API usage policy I’m going to steer clear of encouraging competition on descending segments. Strava already take a small step in this direction on their own site by only allowing the setting of goals on a segment if the average grade is greater than -0.25%. I’m using that same criteria to limit the segments taken into account for the Score.
You’ll see just below the score something along the lines of “From 100 of 921 segments.” These are the two numbers determined by the above two factors.
How does my Score stack up against others?
UPDATE 21st Nov 2013: Just rechecked the distribution chart with my current data (21,679 athletes) and the result is remarkably similar to look at to that above. The updated, key stats are as follows:
- Lower Quartile – 74.60
- Median – 85.46
- Upper Quartile – 93.81
Here is a frequency distribution chart for each Score (rounded to nearest whole number). It is clearly very top heavy as the Score is weighted towards people’s top scoring segments. The average (median) is the shown by the red line and the inter-quartile range is also shown by the white area. This should give you a good idea of where you sit in the grand scheme of things but for those not up on their stats speak:
If your score falls:
- in the left grey area (below 76.8) then I’m afraid you are in the bottom 25% of VeloViewer athletes. Move direct to “How do I improve my score?” section below, do not pass go, no not collect £200.
- in the white area to the left of the red line (above 76.8 but below 87.6) then your score is in the range of 25-50% of VeloViewer athletes. Pretty good going but still some ground to make up. Maybe time to upgrade your wheels?
- in the white area to the right of the red line (above 87.6 but below 94.5) then your score is in the range of 50-75% of VeloViewer athletes. Proof that you are better than average even if it is in something fairly meaningless!
- In the right grey area (above 94.5). You can direct your partner to this graph as scientific proof as to why spending that excessive portion of the kid’s inheritance on those ridiculous looking/sounding aero wheels was justified. Congratulations! (note: if anyone wants to give me a set of those ridiculous aero wheels I’d happily take them and blog about how much faster they make me go.)
How can I see what segments make up my score?
Go to your Segments list and pick the “Scoring” option from the “Config” drop down. A filter will be applied removing your downhill segments and the data will be ordered by the position score. Make sure you remember how many segments your Score is being made up of (in my case it is 100). The top 100 (or less if that is the case for you) segments will be the ones making up your VeloViewer Score.
How do I improve my score?
Go faster more of the time? Does that help? A couple of slightly more practical tips might be:
- Go to the “Scoring” config for your segments but then reorder by the “Total” column (total number of athletes). The segments at the top of your list will be the ones that have the greatest potential position score. Now filter these segments to pick the ones most suited to you (e.g. long climbs, short climbs, long flat sections etc.) and look to see which have the best opportunity for you to increase your current placing using something like the “Behind leader %” value.
- The other way would be to find segments with lots of athletes that you haven’t actually ridden yet. For the time being you’ll have to do that on Strava’s Segment Explorer.
What’s not so great about the VeloViewer Score…
Like all things, nothing’s perfect, and the VeloViewer Score has a number of limitations which is why it should be used as a tool to rib your mates that you are beating but not something to lose any sleep over. The issues that stand out for me are:
- Duplicate segments – as you can see in the screenshot above showing the segment list filtering, my top 3 scoring segments used in the Score are just about all the same bit of road. I could go and hide 2 of them in Strava but then down goes my Score (and my KOM count!) If I was a real knob I could create a couple more duplicates here and give my Score a boost! Don’t do that, unless you really are a knob.
- Limits to individual position scores due to low athlete counts – even if my wife was QOM on all of her segments her Score would be lower than mine despite a large number of my scoring segments not being KOM’s, this is because each segment’s position score is limited by the number of athletes and round these parts there just aren’t that many women Strava’ing their rides.
- Now that I’ve plotted out the distribution of Scores it is obvious it isn’t very normalised having a distinct lean towards the higher Scores. What this means is that where the highest number of athletes are scoring, the distinction between their Scores can probably get lost. Potentially I could attempt to apply some kind of multiplier to everyone’s Score in an attempt to normalise the distribution but I’m not sure how well that will go down with people! It might be worth playing around with if only from a “isn’t maths fun” perspective. If anyone knows how I would go about doing this then please let me know 🙂