US contest scoring formula is broken

On Jun 22, 11:08*pm, Steve Koerner wrote:

Then a bizarre compression comes into play for slower finishers. *The
seventh place pilot flew a mere 67 miles at 23.33 mph and received 76%
of the winners score. *That’s absurd. *He flew only about one third of
the speed and distance that the second place finisher flew yet scored
a mere 70 fewer points.


This in part results from a change in 2007 that changed Maximum
Distance Points (MDP) from 400 to 600. The relevant rules (for MAT and
TAT - similar for AST) a

11.6.6 Maximum Speed Points:
MSP = STF * (600 + 500 * SCR) (but not greater than STF *
1000)
11.6.7 Maximum Distance Points:
MDP = MSP * (0.8 - 0.2 * SCR)
11.6.8 Points for Finishers:
POINTS shall be equal to the largest of the following
three quantities:
MSP * SPEED / BESTSPD
MDP + 30 + MSP * 0.2 * ((SPEED/BESTSPD) - 0.4)
MDP + 30

As I understand it, the basic idea is that the rule makers (with pilot
input) are trying to make sure that pilots who have a landout can stay
somewhat competitive (See: 11.6.8 - dropping 600 points on a day is
hard to make up - 400 is still hard, but less so). There are also
provisions for devaluing tasks with lots of landouts (more than 20% -
see 11.6.6). Lots of landouts are thought to correlate to a higher
"luck factor". Short tasks are also devalued under similar logic.

The simplest scoring formulas would be to set Maximum Speed Points
(MSP) to 1000 and Minimum Distance Points (MDP) to zero and to get rid
of Scored Completion Ratio (SCR) and Short Task Factor (STF). What
this would mean is that if you finish, you get a score that is
directly proportionate to your percent of the winner's speed - no
matter how slow you go. If you land out you get zero. An alternative
is to to keep the idea of MDP (but we need to pick a number - 200,
400, 600, 800?) and relax the constraint that the best landout has to
get fewer points that any finisher. This allows us to keep the idea of
strict proportionality for any speed finisher. Under this scenario you
could see your score drop by a lot if you you are slow and finish
rather than landing out (because MDP exists, finishers whose speed as
a percent of the winner's speed is less than MDP/1000 could score less
than a long landout). This may be a bad idea as it encourages landing
out.

A third alternative is to pin any finisher's score that is less than
MDP to MDP - but this introduces the possibility that the bottom of
the scoresheet is populated with pilots who have identical scores
because they can't get above MDP. The lower you make MDP, the less
likely this is, but to avoid it for the Montague example described by
Steve MDP would need to be 300 points - which is less that it has ever
been in the rules. Obviously, that specific situation doesn't come up
often, and when it does its usually a funky day or a situation where a
pilot had something odd happen on course. It can also result from
cutting short at MAT (particularly if no turnpoints are assigned) or,
to a lesser extent, a TAT course.

I agree that the Montague example seems extreme, so it may warrant
review. It seems to me that all the alternatives have some tradeoffs
though.

9B

On Jun 23, 6:18*am, Andy wrote:
On Jun 22, 11:08*pm, Steve Koerner wrote:



Then a bizarre compression comes into play for slower finishers. *The
seventh place pilot flew a mere 67 miles at 23.33 mph and received 76%
of the winners score. *That’s absurd. *He flew only about one third of
the speed and distance that the second place finisher flew yet scored
a mere 70 fewer points.

This in part results from a change in 2007 that changed Maximum
Distance Points (MDP) from 400 to 600. The relevant rules (for MAT and
TAT - similar for AST) a

* * * * 11.6.6 Maximum Speed Points:
* * * * * * *MSP = STF * (600 + 500 * SCR) (but not greater than STF *
1000)
* * * * 11.6.7 Maximum Distance Points:
* * * * * * *MDP = MSP * (0.8 - 0.2 * SCR)
* * * * 11.6.8 Points for Finishers:
* * * * * * *POINTS shall be equal to the largest of the following
three quantities:
* * * * * * *MSP * SPEED / BESTSPD
* * * * * * *MDP + 30 + MSP * 0.2 * ((SPEED/BESTSPD) - 0.4)
* * * * * * *MDP + 30

As I understand it, the basic idea is that the rule makers (with pilot
input) are trying to make sure that pilots who have a landout can stay
somewhat competitive (See: 11.6.8 - dropping 600 points on a day is
hard to make up - 400 is still hard, but less so). There are also
provisions for devaluing tasks with lots of landouts (more than 20% -
see 11.6.6). Lots of landouts are thought to correlate to a higher
"luck factor". Short tasks are also devalued under similar logic.

The simplest scoring formulas would be to set Maximum Speed Points
(MSP) to 1000 and Minimum Distance Points (MDP) to zero and to get rid
of Scored Completion Ratio (SCR) and Short Task Factor (STF). What
this would mean is that if you finish, you get a score that is
directly proportionate to your percent of the winner's speed - no
matter how slow you go. If you land out you get zero. An alternative
is to to keep the idea of MDP (but we need to pick a number - 200,
400, 600, 800?) and relax the constraint that the best landout has to
get fewer points that any finisher. This allows us to keep the idea of
strict proportionality for any speed finisher. Under this scenario you
could see your score drop by a lot if you you are slow and finish
rather than landing out (because MDP exists, finishers whose speed as
a percent of the winner's speed is less than MDP/1000 could score less
than a long landout). This may be a bad idea as it encourages landing
out.

A third alternative is to pin any finisher's score that is less than
MDP to MDP - but this introduces the possibility that the bottom of
the scoresheet is populated with pilots who have identical scores
because they can't get above MDP. The lower you make MDP, the less
likely this is, but to avoid it for the Montague example described by
Steve MDP would need to be 300 points - which is less that it has ever
been in the rules. Obviously, that specific situation doesn't come up
often, and when it does its usually a funky day or a situation where a
pilot had something odd happen on course. It can also result from
cutting short at MAT (particularly if no turnpoints are assigned) or,
to a lesser extent, a TAT course.

I agree that the Montague example seems extreme, so it may warrant
review. It seems to me that all the alternatives have some tradeoffs
though.

9B

Sorry - I meant to say MDP is Maximum Distance Points, not Minimum.

9B

On Jun 23, 9:18*am, Andy wrote:
On Jun 22, 11:08*pm, Steve Koerner wrote:



Then a bizarre compression comes into play for slower finishers. *The
seventh place pilot flew a mere 67 miles at 23.33 mph and received 76%
of the winners score. *That’s absurd. *He flew only about one third of
the speed and distance that the second place finisher flew yet scored
a mere 70 fewer points.

This in part results from a change in 2007 that changed Maximum
Distance Points (MDP) from 400 to 600. The relevant rules (for MAT and
TAT - similar for AST) a

* * * * 11.6.6 Maximum Speed Points:
* * * * * * *MSP = STF * (600 + 500 * SCR) (but not greater than STF *
1000)
* * * * 11.6.7 Maximum Distance Points:
* * * * * * *MDP = MSP * (0.8 - 0.2 * SCR)
* * * * 11.6.8 Points for Finishers:
* * * * * * *POINTS shall be equal to the largest of the following
three quantities:
* * * * * * *MSP * SPEED / BESTSPD
* * * * * * *MDP + 30 + MSP * 0.2 * ((SPEED/BESTSPD) - 0.4)
* * * * * * *MDP + 30

As I understand it, the basic idea is that the rule makers (with pilot
input) are trying to make sure that pilots who have a landout can stay
somewhat competitive (See: 11.6.8 - dropping 600 points on a day is
hard to make up - 400 is still hard, but less so). There are also
provisions for devaluing tasks with lots of landouts (more than 20% -
see 11.6.6). Lots of landouts are thought to correlate to a higher
"luck factor". Short tasks are also devalued under similar logic.

The simplest scoring formulas would be to set Maximum Speed Points
(MSP) to 1000 and Minimum Distance Points (MDP) to zero and to get rid
of Scored Completion Ratio (SCR) and Short Task Factor (STF). What
this would mean is that if you finish, you get a score that is
directly proportionate to your percent of the winner's speed - no
matter how slow you go. If you land out you get zero. An alternative
is to to keep the idea of MDP (but we need to pick a number - 200,
400, 600, 800?) and relax the constraint that the best landout has to
get fewer points that any finisher. This allows us to keep the idea of
strict proportionality for any speed finisher. Under this scenario you
could see your score drop by a lot if you you are slow and finish
rather than landing out (because MDP exists, finishers whose speed as
a percent of the winner's speed is less than MDP/1000 could score less
than a long landout). This may be a bad idea as it encourages landing
out.

A third alternative is to pin any finisher's score that is less than
MDP to MDP - but this introduces the possibility that the bottom of
the scoresheet is populated with pilots who have identical scores
because they can't get above MDP. The lower you make MDP, the less
likely this is, but to avoid it for the Montague example described by
Steve MDP would need to be 300 points - which is less that it has ever
been in the rules. Obviously, that specific situation doesn't come up
often, and when it does its usually a funky day or a situation where a
pilot had something odd happen on course. It can also result from
cutting short at MAT (particularly if no turnpoints are assigned) or,
to a lesser extent, a TAT course.

I agree that the Montague example seems extreme, so it may warrant
review. It seems to me that all the alternatives have some tradeoffs
though.

9B

It's the use of MDP in formulas 2 & 3, without regard to how many
miles any given pilot actually flew, that's causing problems.

Agreed that our scoring rules are getting a little dense. I'm
curious: how are European comps scored? Anyone got a link?

-T8

Agreed that our scoring rules are getting a little dense. *I'm
curious: how are European comps scored? *Anyone got a link?

-T8

In Europe, most "classic" contests use the scoring formulas from the
Sporting Code Section 3, Annex A (International competition rules), §
8.4. Fairly complicated too!

On Jun 23, 7:51*am, T8 wrote:
On Jun 23, 9:18*am, Andy wrote:





On Jun 22, 11:08*pm, Steve Koerner wrote:

Then a bizarre compression comes into play for slower finishers. *The
seventh place pilot flew a mere 67 miles at 23.33 mph and received 76%
of the winners score. *That’s absurd. *He flew only about one third of
the speed and distance that the second place finisher flew yet scored
a mere 70 fewer points.

This in part results from a change in 2007 that changed Maximum
Distance Points (MDP) from 400 to 600. The relevant rules (for MAT and
TAT - similar for AST) a

* * * * 11.6.6 Maximum Speed Points:
* * * * * * *MSP = STF * (600 + 500 * SCR) (but not greater than STF *
1000)
* * * * 11.6.7 Maximum Distance Points:
* * * * * * *MDP = MSP * (0.8 - 0.2 * SCR)
* * * * 11.6.8 Points for Finishers:
* * * * * * *POINTS shall be equal to the largest of the following
three quantities:
* * * * * * *MSP * SPEED / BESTSPD
* * * * * * *MDP + 30 + MSP * 0.2 * ((SPEED/BESTSPD) - 0.4)
* * * * * * *MDP + 30

As I understand it, the basic idea is that the rule makers (with pilot
input) are trying to make sure that pilots who have a landout can stay
somewhat competitive (See: 11.6.8 - dropping 600 points on a day is
hard to make up - 400 is still hard, but less so). There are also
provisions for devaluing tasks with lots of landouts (more than 20% -
see 11.6.6). Lots of landouts are thought to correlate to a higher
"luck factor". Short tasks are also devalued under similar logic.

The simplest scoring formulas would be to set Maximum Speed Points
(MSP) to 1000 and Minimum Distance Points (MDP) to zero and to get rid
of Scored Completion Ratio (SCR) and Short Task Factor (STF). What
this would mean is that if you finish, you get a score that is
directly proportionate to your percent of the winner's speed - no
matter how slow you go. If you land out you get zero. An alternative
is to to keep the idea of MDP (but we need to pick a number - 200,
400, 600, 800?) and relax the constraint that the best landout has to
get fewer points that any finisher. This allows us to keep the idea of
strict proportionality for any speed finisher. Under this scenario you
could see your score drop by a lot if you you are slow and finish
rather than landing out (because MDP exists, finishers whose speed as
a percent of the winner's speed is less than MDP/1000 could score less
than a long landout). This may be a bad idea as it encourages landing
out.

A third alternative is to pin any finisher's score that is less than
MDP to MDP - but this introduces the possibility that the bottom of
the scoresheet is populated with pilots who have identical scores
because they can't get above MDP. The lower you make MDP, the less
likely this is, but to avoid it for the Montague example described by
Steve MDP would need to be 300 points - which is less that it has ever
been in the rules. Obviously, that specific situation doesn't come up
often, and when it does its usually a funky day or a situation where a
pilot had something odd happen on course. It can also result from
cutting short at MAT (particularly if no turnpoints are assigned) or,
to a lesser extent, a TAT course.

I agree that the Montague example seems extreme, so it may warrant
review. It seems to me that all the alternatives have some tradeoffs
though.

9B

It's the use of MDP in formulas 2 & 3, without regard to how many
miles any given pilot actually flew, that's causing problems.

Agreed that our scoring rules are getting a little dense. *I'm
curious: how are European comps scored? *Anyone got a link?

-T8- Hide quoted text -

- Show quoted text -

Has anyone thought of getting back to basics - by that I mean getting
rid of TAT and all of the follow on tasks to POST - and fly only AST -
Assigned Speed Task - start gate, finish gate, all competitors fly the
same turnpoints, finish gate - fastest pilot wins?

But then what would happen to US Sports Class the best bureacraticly
created catch-all class in the world...

Her's to real racing, with a minimum of complicated rules... and bring
Club Class tot he U.S. so we c;ub class pilots can actually fly AT
speed tasks in official contests.

EY

I would urge that instead of looking at this in terms of ‘how can we
fix the Montague problem’ lets look at this much more fundamentally.
The Montague example shows that year after year of elaborating the
scoring formula has resulted in a mess.

It is time to step way back and start over. Let’s get rid of the
alphabet soup that now exists in the rules for scoring. All (or at
least most) of the accumulated complications in the rules were
undoubtedly made in the interest of improving fairness. Yet at the
most elemental level it is really unfair to be using a set of scoring
formulas that cannot be comprehended by an intellectually
sophisticated competitor who applies reasonable diligence in studying
the rules. That is what we have now. It is especially irksome that
competitors should put up with the unnecessary complication and still
not have a scoring system that produces reasonable results.

So here’s a simple proposal to get this discussion going in the right
direction: Let’s give 500 points for speed and 500 points for
distance. Period. Speed points are proportioned to the fastest
finisher’s speed; if you don’t finish you get zero speed points.
Distance points are proportioned to the best distance achieved.

Yes, on TAT and MAT style tasks, my formula would change the game plan
a bit. It would introduce an objective to go forth and fly far (like
OLC) as well as fast. Wouldn’t that be fun?

On Jun 23, 9:31*am, Steve Koerner wrote:
So here’s a simple proposal to get this discussion going in the right
direction: *Let’s give 500 points for speed and 500 points for
distance. * Period. *Speed points are proportioned to the fastest
finisher’s speed; if you don’t finish you get zero speed points.
Distance points are proportioned to the best distance achieved.

Yes, on TAT and MAT style tasks, my formula would change the game plan
a bit. *It would introduce an objective to go forth and fly far (like
OLC) as well as fast. *Wouldn’t that be fun?


This idea would be a terrific start... or we could just score
according to the FAI rules and use SeeYou to score cotnests like the
rest of the world. You can definitely overcook the rules and we are
definitely in that kitchen.

EY

I also dislike the complexity of the scoring formula. But every bit of
this comes from pilot demand!

You basically can't get less than 600 points, no matter how slow you
go, because a) pilots want a landout not to be a catastrophic end to
the contest, as it would be if you got 0 points for landout and b)
pilots do not want a "finisher" to receive less points than a
"landout". (FAI rule fans note, the change from 400 to 600 points, by
the way, was made in part to harmonize US with world treatment of
landouts vs. speed points.)

I happen to disagree with the latter objective, especially on MAT and
turn area tasks. If you fly 400 miles and land 10 miles from the
airport, I think you should get more points than the guy who does 61
miles and lands 2 hours early. The guarantee puts too much emphasis on
coming home very early to "get a finish." But that's just me, everyone
else likes the current system.

If you want "simpler", so that a very slow finisher always gets his
fraction of the winner score, tell the rules committee which of a or
b above you want to give up.

The big complication with the rules involves day devaluation, all that
scored completion ratio and other stuff. I'd be happy to get rid of
that too, but pilots want it. If every day is 1000 points, so the mass
landout day where the best pilot goes 60 miles, or the 1.5 hour dump
task day counts exactly as much as the other days, then contests will
all be won and lost on those weak short days. That used to happen,
pilots felt it wasn't fair, so we have our current system. Do you
really want to go back to the old system where every day is 1000
points?

Then we get "use the FAI rules" as in the quote below. Have you read
the FAI rules? "Simplicity" and "use the FAI rules" are two separate
ideas! If you don't like our day devaluation formulas, look at
theirs! The FAI devaluation formulas also lead to much more gaggling.
Since the "lone wolf" doesn't gain much by being the only finisher,
everybody sits in one big gaggle until sunset and then lands out
together.

Again, I dislike the complexity and the fact that it's very hard to
check your score just like everyone else. But the scoring formulas in
US contests evolved over the years responding to specific and valid
complaints from pilots. Simplicity is important, but fair and fun
contests are also an important goal! So it would be much more
productive to understand those goals and suggest a simpler system that
addresses them, or makes a clear case for abandoning them

John Cochrane BB.


This idea would be a terrific start... or we could just score
according to the FAI rules and use SeeYou to score cotnests like the
rest of the world. You can definitely overcook the rules and we are
definitely in that kitchen.

EY

John Cochrane BB

On Jun 23, 12:56*pm, John Cochrane
wrote:

If you fly 400 miles and land 10 miles from the
airport, I think you should get more points than the guy who does 61
miles and lands 2 hours early.

I agree. What I would suggest is replacing "MDP" in the scoring
formula for finishers with MDP*DIST/BESTDIST.

-T8