BHPC Events - DNF handling for fixed distance races

DNF handling for fixed distance races

Printed From: BHPC Forum
Category: Public: Open to anyone
Forum Name: BHPC Events
Forum Description: Come along and meet the club!
URL: https://forum.bhpc.org.uk/forum_posts.asp?TID=7567
Printed Date: 27 March 2026 at 1:56am
Software Version: Web Wiz Forums 12.07 - https://www.webwizforums.com

Topic: DNF handling for fixed distance races

Posted By: Kim
Subject: DNF handling for fixed distance races
Date Posted: 18 July 2023 at 1:54pm

Heather's DNF on Saturday raised an interesting question regarding the handling of DNFs (I appreciate that she's unlikely to care about the points, but we should work out how to do this in future):

In a criterium race, we fudge a DNF rider's last lap time to the finish time of the last finisher. This gives the DNF rider a virtual last lap equivalent to waiting and then limping over the finish line at the end of the race, so they can gain a finish position (and therefore points) in accordance with their overall average speed.

However, this doesn't make sense when the race is a fixed distance, as with the 4-lap race we did on Saturday (but equally applicable to time trials). I haven't knowingly encountered this problem before, though it's possible that someone didn't make it round a time trial at some point and we treated as DNS.

As I see it we can:

Leave them as DNF without any fudging. They won't get ranked, but SeriesMgr can optionally allocate a fixed number of points to DNF riders (I propose 5 points, as this is where all the points allocation serieses asymptote).
Set them to NP (or DQ or DNS). They won't get ranked, or appear in the points table.
Leave them as a finisher and let CrossMgr extrapolate times for their uncompleted laps. This is extremely unfair as you could rank quite highly by sprinting for one lap and then DNFing.
Set them to PUL and allow CrossMgr to rank riders by average speed. Equivalent to the above, with the same risk of being gamed.
Set them to PUL and allow CrossMgr to rank riders by laps completed and then finish time (the default ranking algorithm, which we decided was unfair for combined criteriums, but shouldn't be a problem when everyone is riding a fixed distance).
Give them the last finisher's time for a virtual last lap and let CrossMgr interpolate the rest. This seems obvious, but it's equivalent to giving them the same rank as the last finisher, which is unfair as they haven't completed the distance.
Give them a fictitious time for the last completed lap to ensure they are ranked below the last finisher, and let CrossMgr interpolate. Involves manual arithmetic.
Give them the last finisher's finish time for their last completed lap and let CrossMgr extrapolate subsequent lap times from that, for a low average speed. This is probably the closest functional equivalent to what we do with criteriums, but would make for some very confusing results, with times for laps that were never ridden.
Something else, likely that isn't currently supported by the software.

I'm inclined to say that we leave the DNF-fudging for time-based races, and treat a DNF in a distance-based race as a true DNF, ie. Option (1) above. The reason we fudge DNFs in the first place is to remove the incentive to wait and limp over the line at the end of a time-based race. You can't do that in a distance based race, as you'd have to limp the entire race distance, and if you manage to do that you're a normal finisher.

Failing that, option (5) seems the best way to give riders credit for what they've ridden in a way that doesn't involve bistromathics or fill the results table with too many fictional numbers.

Thoughts?

Replies:

Posted By: Andrew S
Date Posted: 18 July 2023 at 11:09pm

Option 2 unfairly penalises the DNFer whereas Options 3-8 all seem to confer an unfair benefit. So I agree that the fairest solution is Option 1. As there is no practical way of measuring the effort expended, unlike in a fixed-time event, I like the idea of giving a flat score of minimum points if that's reasonably doable.

Interesting that we're still discovering undiscovered quirks in the scoring after all these years.

Posted By: Andrew S
Date Posted: 18 July 2023 at 11:12pm

Actually Option 5 does seem to reward effort proportionately, if I've understood it correctly, so maybe that's slightly preferable to Option 1.

Posted By: Kim
Date Posted: 21 July 2023 at 1:36pm

Andrew S wrote:

Actually Option 5 does seem to reward effort proportionately, if I've understood it correctly, so maybe that's slightly preferable to Option 1.

Having gone away and done something else for a bit (namely, made CrossMgrVideo display the times of the selected frame for each rider in its list view, rather than just the time that the capture was triggered), I think that's the better option.

I've made the relevant changes and re-generated the results and points serieses. Heather now gets 27 points!

Posted By: HeatherF
Date Posted: 21 July 2023 at 5:36pm

Woo-hoo!

Posted By: Dan Bergonzi
Date Posted: 21 July 2023 at 11:16pm

Off thread
Thanks to whoever put the finish line photos on the 2023 race page - great to relive those close finishes

Posted By: Kim
Date Posted: 22 July 2023 at 1:23am

Dan Bergonzi wrote:

Off thread
Thanks to whoever put the finish line photos on the 2023 race page - great to relive those close finishes

Yes, I've now made sufficient refinements to CrossMgrVideo that publishing the finishes isn't a great deal of extra work, assuming you've been through each ~5sec video and selected the correct frame for each rider's finish.

Originally you'd have a bucket full of trigger times to wade through, with no obvious clues (beyond the timestamp) if it were an intermediate lap, a finish or someone with a helmet wandering around near the timing tent. As such, it was only worth doing if there were actual close finishes or missed reads to investigate, and you had to export every photo individually (or the entire database).

Now you merely have to establish the leader's finish time, and it'll select what should be everyone else's finish after that, so it's a bit of mindless mousewheeling on n videos to line each bike up with the finish line to generate a list of precise finishes, which is useful enough that it's worth doing before fixing the race times. You can then bulk-publish the selected photos into a web page with a few clicks, and a bit of FTP to upload them (which could also be automated).

Posted By: Dan Bergonzi
Date Posted: 22 July 2023 at 1:24am

Thanks Kim

Posted By: Kim
Date Posted: 22 July 2023 at 1:25am

Glad someone appreciates it.

One point to note with the photos page is that CrossMgrVideo gets the rider details and stores them in its database each time CrossMgr tells it someone's passed the finish line. Which means that - unlike in CrossMgr and SeriesMgr - subsequent changes to the sign-on sheet (or changes to the finish order or race names) aren't reflected in that version of the list.

(It's also possible to trigger CrossMgrVideo to record directly, in which case it won't have any rider details at all, unless they're manually added.)