Ticket #88 (closed defect: fixed)

Opened 9 years ago

Last modified 9 years ago

Leading Coef. too low when nobody makes goal

Reported by: danielvelezbravo@… Owned by: ste
Priority: major Milestone: 1.2.3
Component: FsComp Version: 1.1.6
Keywords: Cc:

Description

I have taken the liberty of re-scoring the Santa Cruz Flat Lands Hang Gliding Meet, using FS 1.1.6 (1.2.1. is not reading the kml files), and using Leading Coef, as in GAP 2002 + CompeGPS configuration. The actual meet was scored (I believe) using SeeYou? Software and a scoring script made by Davis Straub, that uses the OZGAP2005 formula.

Nevertheless, I was hopping on seeing some differences on the scores, because of the leading coefs, to figure out if it really worthy to lead agressively during the task or if you should keep a low and conservative profile until you are close to goal, and only then, run for it to get arrival time points.

The leading coef on some of the tasks was actually working nice, giving a small difference but as I remembered, on the days that I actually pushed a little bit forward, it gave me a better score spread (not total score) on relation with the ones that did not lead.

The issue that I found is that actually on two of the tasks nobody made it back to goal, still there where some agressive flying on the first places. Still, the leading coef. scores for those two days was as big as 1 point for the best, and 0,5 average for the other pilots. So it seems that on days that nobody makes it to goal, there´s no leading coef incentive to get the pilots pushing in front. I´m guessing that this is actually a bug, as the Lead. Coef. explanation from gap 2002 is supposed to give an score incentive for leading on every case.

Attachments

Santa Cruz Flat Lands Day 2.zip (1.3 MB) - added by danielvelezbravo@… 9 years ago.
Santa Cruz Flat Lands 2008 - Day 2

Change History

Changed 9 years ago by danielvelezbravo@…

Santa Cruz Flat Lands 2008 - Day 2

comment:1 Changed 9 years ago by ste

  • Owner changed from somebody to ste
  • Status changed from new to assigned

I see your point Daniel. Will look into it on Monday. We're having our first comp for the season here in Norway so a bit busy.

Stein-Tore

comment:2 Changed 9 years ago by ste

Just read the GAP 2002 document ("GAP02_en.pdf") that comes with RACE.

Can't see that it says anything about LC if no one makes ES.

However, I do belive CompeGPS/RACE will give more points for leading when no one finish speedsection than GAP2007 in FS does. I do not have a description on how CompeGPS calculates LC.

Stein-Tore

comment:3 Changed 9 years ago by ste

  • Status changed from assigned to closed
  • Resolution set to fixed
  • Milestone set to 1.1.7

comment:4 Changed 9 years ago by ste

Found bug and fixed in FS 1.1.7 (and FS 1.2.1).

At best one should get 25 leading points for task with no one at ES (or open dist. task).

Stein-Tore

comment:5 follow-up: ↓ 9 Changed 9 years ago by Eduardo

  • Status changed from closed to reopened
  • Resolution fixed deleted

Actually, the maximum leading points is 18 in tasks with no one in goal (in GAP 2002) CompeGPS + RACE assigns a maximum of 17,5 points but the rounding gives 918 pts to the winner (900 distance + 18 leading) if he is also the best leading pilot. I am talking of tasks with Day Validity of 1.00 of course.

I have no idea how Stein reallocated distance and leading points when he created the new option of giving a maximum of 1000 pts to open distance tasks or tasks with no pilots in goal, can you tell us Stein ?

If we want to keep almost the same scores in FS as CompeGPS + RACE when GAP 2002 formula is used and the option of 1000 points is not active, then the 18 max leading points should be used.

comment:6 Changed 9 years ago by danielvelezbravo@…

Actually, my intention was not to start a disscusion on Leading Poins according to GAP, as that was a discussion held and decided by the GAP designers. Nevertheless, i did not knew the actual behviour os the leading points whithin GAP 2002.

Now i think its important to re-evaluate the behaviour of leading points.

Their philosophy was to reward the pilots that actually lead out during a race, so that the ones that only followed the leader, and at the las momment go for the finish line get somehow penalized for their flying. OZGAP was somehow a step back as it intended to simplify the lead calculations and the limitation to one scoring software that was able to do it. There is not a good explanation as why a task that nobody makes goal should have less leading points available than a task when they actually do, as its more common that on tasks that nobody makes goal than the followers get more distance than the leaders, as they can see how bad the leaders are strugling and held back to pass above them when they land. I personally think that now that we can use a free software to verify the leading coefficient, we should give more important to it. The leading to turnpoint, as they do on long cycling races, and that was suggested by Gerolf (i think) on another ticket, is actually a nice step forward related to OZGAP, but then you will need to have a few extra turnpoints to make it work (something like a turnpoint each 20 or 30 km) so everybody would be wanting to race torwards them within the task, taking their chances and leading out, because it pays, but if you could measure and score the leading during all the race, it would be really good. If the cyclist could measure the leading coef on their racers they will probably see that the winners dont lead at all, but they team members thar are doing all the aerodinamics jobs to keep the stars from getting tired... Those guys would love to have some score recognition for their hard work. On flying competitions we have the means and the tools for knowing who was leading at any moment, and no followers can say that they did not get an advantage out of it (dived in for a thermal, slowed down on a bad zone, choosed a nicer line than the leader was showing and so on), so giving the propper score to the pilots that decide to lead should consider a nicer leading points score. Go to the first GAP proposal that had early bird and arrival points that all summed like 100 points. Now we can forget about those two but give the leading coef. the same importance of 100 points. If noones makes goal, then the longest distance would be taken into consideration as the distance to measure the leading coef., but the leading points would go to every pilot instead as to the pilots that make goal on a goal task. Giving 18 points does not seems to be worty at all. As a personal story, on the last Santa Cruz champs i was leading to a goal with another pilot, and we landed 6km short. Then like 15 minutes behind a nice pack of 6 or 7 gliders passed above us and made a few km closer. Nevertheless the leading was not awarded at all and the distance, those last 6 km were 65 points worthy. Im not saying that one should win as the distance should be prevailing, but that somehow it should be worthy to at least not loose to one pilot that landed 900 mts farther than me but 15 minutes later, and got 20 points more than me...

comment:7 Changed 9 years ago by Eduardo

I share your concerns that leading points are meager in tasks with no pilots in goal, but I was only pointing out what should be taken into account if we want to score GAP2002 with FS.

And there is the definition of 18 leading points in the GAP02_en.pdf I quote

"If nobody completes the SpeedSection? there are a maximum of 900 points available for Distance plus 18 points for Departure but, of course, no points for Speed and Arrival."

comment:8 Changed 9 years ago by danielvelezbravo@…

Hello Eduardo.

You are absolutelly rigth about it. If the FS software is intended for implementing the GAP 2002, then it should stick to it´s bonus values, meaning only 18 points for departure bonus. Now if we have a newer tool that can measure not only the departure time but the leading during all the race, I would have to disagree that we can assume that the GAP writters would think or measure that leading during a whole race would have the same score as taking the earliest start.

On the same basis, when originally written, the GAP 2000 stated the following philosophy about Early Bird:

"Pilot Departure Bonus Known as the "early bird bonus" but better defined as the "leading bonus" is provided to encourage fast pilots to take off early and rewards the risk involved in being in the leading group... ...Keep in mind: Departure points are a percentage of your own speed points."

And then the same GAP 2000 document concluded, after explaining the arrival bonus the following:

"Keep in mind: The departure and arrival bonus system is there to reward the pilot who leads out. If you lead out, you make decisions yourself, and you take more risks of bombing out or being slower. If you follow other pilots, you might get to goal with a fast time, but your points will be less than the pilot who flew with the same time (or maybe a slightly slower time) but who flew in front of you. The GAP formulas are designed to reward the pilot who makes the decisions."

Probably this is not the place for this comment, and we should start a new discussion on the forum perhaps, and I´m also not sure about who has the autority on the FAI to elaborate a newer GAP for scoring, but as we can now measure the leading during all the way, and not assuming that the starting and ending would be the same as the leading (starting at the same start gate but a few minutes behind, and following all the way until the last termal on wich you could find a nicer line to goal and beat the leaders), I´m wondering if we could propose a GAP 2008, that uses a more agressive leading coef, regardless of the goal. Probably if you lead all the way to goal but don´t make it inside the goal line, you should get all the leading points, even when you don´t get any speed points, so this way you will probably beat the slower pilots that makes it all the way to goal, or at least be too close, and the pilots that were following you and made goal won´t get all the leading points because they actually did not lead at all but followed you and took advantage of your last bad decision in order to overfly you and beat you to goal.

GAP 2008 should have the two possibilities: The more agressive leading coef. or no leading coef but a "leading to turnpoint" bonus.

About this last leading to turnpoint bonus, I´m starting to like it better than the full leading coef, but I have a suggestion that I´m about to post on the subsecuent ticket.

comment:9 in reply to: ↑ 5 Changed 9 years ago by ste

  • Status changed from reopened to closed
  • Resolution set to fixed
  • Milestone changed from 1.1.7 to 1.2.3

Replying to Eduardo:

Actually, the maximum leading points is 18 in tasks with no one in goal (in GAP 2002)

You're right Eduardo,

for some reason I typed the wrong number when coding this little bit (4-5 years ago, never checked it since) resulting in 25 points instead of 17.5 points added to the 900 as max leading points when no one reach goal.

If we have pilots in goal, max leading points is exactly same as in the GAP formula used in RACE.

I think we'll leave GAP2007 as it is with the 25 points.

GAP2002 in FS 1.2.3 will be corrected so it has the 17.5 points.

I have no idea how Stein reallocated distance and leading points when he created the new option of giving a maximum of 1000 pts to open distance tasks or tasks with no pilots in goal, can you tell us Stein ?

Simply adding what is missing to max distance points. Leading points stays the same.

Stein-Tore

comment:10 Changed 9 years ago by ste

Both GAP2002 and GAP2007 in FS 1.2.3 will have the original max of 17.5 points for Leading points when no one reach goal.

Now these formulas (as well as GAP2000) should give pretty much same results as they would do with same flightdata in RACE (&CompeGPS).

Stein-Tore

Note: See TracTickets for help on using tickets.