Facebook

Login

Lost password?

Support Sailonline

If you haven't already - join the SAILONLINE YACHT CLUB!

Please also consider making a donation - all amounts are greatly appreciated!

Board » Sailonline Race Committee » Championships » Discarding ziggy points or… not.

Good morning SRC and all.

There are very good reasons for the following suggestion to SRC on future Series editions in a form of two race account options with cumulative effects:

1 - Don’t discard Series race points, and/or,

2 - give (in fact, remove) extra points to reward SOLers regularity in the Series (ex.: 1 point removed per race fully sailed and an extra #3 points for a full raced Series in the end of the season).

The second option has also other beneficial implications.

If you prefer a Statistic language, discarding an “outlier” in a population with a size of 30, or bigger, doesn’t have the same impact as doing it in a population of size 6.

If your intuition doesn’t help you now, please have a look on the following numeric example - in the attached files - where “Boat 2” is favored in both cases after discarding the worst race points.

Maths brings us to Earth, but judge it yourself.

João /psail

P.S.: A quick explanation on the above race results/points. They were firstly generated with the Excel Random Number function to be considered inside the classification of an aleatory experiment and avoid any bias to specific conclusions. After, only #2 numbers in #2 different columns were changed to give equal Sums (*) and Averages (*) per example.


--- Last Edited by JB at 2017-10-02 10:46:31 ---
Sail Fair.
Attachments
I'm not SRC, but I think I'm qualified to talk about this. I can join the discussion, but not make the decisions.

I don't understand your argument João. You're saying one discard in 6 races has a larger impact that 1 discard in 30 races? I agree. Question is, is it a good or a bad impact? Your suggestion of not discarding Series points seems to indicate that you think discards are a bad impact, but I don't understand why you think that is the case.

I like to believe that discards are there to eliminate the element of "bad luck" from the racing. Having bad luck one more time than the opponent shouldn't give you a 100 point disadvantage, so that's why you allow skippers one bad race every 5 (or so) races. If you have a bad race more than once every 5 (or so) races, you deserve to be punished.

In a race like the VOR, where we have 7 competitive boats, the punishment for a DNF (because of damage for example) is relatively small, but in a big race on SOL, with 200 entrants of which 100 competitive, the punishment if a DNF (because of connection failure for example) is much larger. So without discards, bad luck had a massive influence of the rankings on SOL compared to some IRL races. The VOR doesn't have discards (at least not last time) because the penalty of bad luck is not too bad meaning there is a reasonable balance between doing well in normal races and managing the risk of having a bad race.

Anyway, this is an interesting topic, I might do some experimenting like you did in your spreadsheets but on a larger scale. I'll keep you posted when I do.
Viva Huib.

More than qualified, namely for this discussion. Thank you for your input.

I’ll start by reinforcing and clarify the suggestion I’ve made:

it should be applied to “Series” with a short number of races, like the 40’ ones, where you only have, precisely, #6 races / quarter and each race can be sailed (in the present edition) with a maximum of #3 different boats.

In relation to the remaining SOL Series, where most of them have more than “30” races / year accounting for the overall classification, I’m good with the actual applied criterion and the main reason behind it, as you well explained.

For now, let me please point this:

1 - With the exception of “40’ Series” all SOL races, namely VOR ones, are sailed with the same boat, in spite of having different legs;

2 - “The “40’ Series”, no;

3 - Any race discard on the “40’ Series” is bad as it can induce not only a big distortion in the final account as also "kills" the Series strategy (finding the best boat for each race);

4 - I kindly suggest you to read my last race report on the Hawaiian 40’ Series race that I think is about to be published (I sent it yesterday to Finn).

The full reasons can be easily understood there.

After the race rep publish I’ll wait for your further comments/questions/discussion.

Big Hug
João/psail
Sail Fair.
Good afternoon SRC, Huib and all.

I bring here some additional thoughts about this subject now the Hawaiian race report has been released.

The fundamental Principles of Game Theory are:

1 - Each player makes the best possible move;
2 - Each player knows that his or her opponent is also making the best possible move,

where here the “best possible move” means each skipper, for every race (#6 in total) chooses the boat he thinks/knows has more chances to win.

I’ll add a third Principle specifically for the SOL 40’ Series:

3 - For each race, there’s only one boat class (from the three available) that is the best choice for wining.

As per condition of the actual 40’ Series rules you’ll have to choose twice the same boat to sail two different race fields for a total number of #6 races.

The optimal Series strategy comes from the “maximin” criterion, which consists of selecting for each race the boat that maximizes the minimum possible “payoff”, i.e., the race outcome.

When you discard races/points you introduce:

4 - A downstream artificial “leveling” that, not only roughly distorts the final Series account (for all boats), but mainly

5 - impacts differently the top positions,

by subverting the Game strategy.

In a “full cooperative” world/game, ideally each race would have been sailed only with one of the three possible boats, obviously the one under Principle #3.
But we know the “full cooperative” it’s utopic, especially in a community where “secret” exchange is a taboo. It’s a fact.

IMHO, and to what concerns SOL/SRC, the right steps to be taken are the following:

6 - Don’t discard races in the 40’ Series;
7 - Give a premium to SOLers regularity for each race and in the overall Series;

Both 6 and 7 actions will:

8 - Minimize not only individual races DNF’s but the Series in all;
9 - Increase the competitive edge by approximating the different levels of SOL sailing expertise and also,
10 - bring & keep more SOLers inside.

In spite of some technicalities I think the issue is more “political”.
That’s why SOL management should also have a word on this, me thinks.

Big Hug
João/psail
Sail Fair.
This is not a rigorous scientific result, but I believe it makes a convincing argument in favour of having one race discarded in a 6 race series.

The 40' series is a different thing all together. The whole limited boat choice makes it a very different type of race and discarded races have side effects that are not observed in normal race series. The following analysis is not applicable to the 40' series.

I have simulated 200 series of 2000 races each with 200 boats. I have used to final ranking in the series as the "true" ranking, and after that I have only looked at the first 40 races and compared them to the "true ranking". I have modelled the boats as follows: in a race they have some performance 0-100%. 100% means they couldn't possibly have sailed faster, 0% means they didn't finish. Each boat is defined by 6 parameters:
- the mean performance when all goes well (90 to 100%)
- the standard deviation of performance in a good race (0.4 to 2%-point)
- the mean performance in a bad race (something breaks/goes wrong, but boat can continue racing and finishes): values between 50 and 90%
- the standard deviation for performance in a bad race (2 to 10%-point)
- the probability of having a bad race (between 5 and 15%)
- the probability of scoring a DNF (between 2 and 8%)

For every simulated series I "created" 200 new boats by picking for each parameter a value uniformly at random in the specified range.

As a measurement of how good a ranking is, I took the Spearman's footrule distance from the "true" ranking (obtained after 2000 races). The larger this distance, the larger the "ranking error". When this distance is 0, the ranking is exactly equal to the "true tanking".

Attached is a figure (linear scale and log scale) showing how this "ranking error" decreases as the number of races increases (as you would expect). Up to about 5 races in a series, we see that ranking with no discards gives on average the best ranking. Between about 5 and 12 races in a series, one discard gives a better ranking. When there are 24 or more races in a series, allowing 3 discards yield better rankings than allowing 2 discards.

How I modelled the boats may not be very accurate. Changing the model and the number of boats in the fleet will likely have an effect on where these "switch-overs" are exactly, but I am confident that the general behaviour is consistent: as the number of races increases, the number of discards should increase as well, but slower.
Attachments
Viva Huib,

Good afternoon.

Excellent work, Huib!

Let me see if I understood it well, starting with the generated “population” size for the race results.

#Races = 10 races / serie x 200 series = 2.000 races.
#Results = 2.000 races x 200 boats/race or results/race = 400.000 race results.

Reaching the “results sample” size (#40) I didn’t understood how did you chose it (that precise “sample”). This is important.

Did you pick #40 results randomly from the overall generated “population” data (race results”)?
Or, did you just pick the first 40 list results from the population data?

Some comments now.

Otherwise clarified for this purpose and, as in RL, a “DNF” can’t be a “good race”, so I’m inclined to think there’s a “cross” probability for having a “bad race”(5% - 10%) and a “DNF” (2% - 8%) simultaneously, as you didn’t mention they are (?) mutually excludable.

Mind you there’s other trivial situation like “before you were sailing a perfect race till you crashed the boat in the rocks”, or to say: a “bad race” doesn’t necessarily implies a “DNF” while, as above mentioned, the contrary holds true.

The “Spearman’s” method is very clever. Chapeau Huib!

About the results.

I confess I’m puzzled about the results in the lower range of races numbers.
Concretizing: “Between about 5 and 12 races in a series, one discard gives a better ranking.
When there are 24 or more races in a series, allowing 3 discards yield better rankings than allowing 2 discards.”.

Intuitively it is not expectable that, with a small set of races, increasing the number of “discards” also increases the reliability of “grading” the rankings, if you understand me.

Big Hug
João/psail
Sail Fair.
You make some good points, let me clarify.

200 series. In each series I picked 200 boats and had them sail 40 races (for each new series, I picked a new set of 200 boats, but within a series, I used the same set of boats for the 40 races).

To determine the "true ranking", simulated 1960 more races in each series, totalling 2000 races in each series, but only the first 40 are used in the analysis.

Why 40 races? because I thought series with more than 40 races were not too interesting to look at. But yes, my results for a series of x races and that for a series of x+1 races are not at all independent.

About DNF and bad races, you are correct. Let me give a more accurate description of the probabilities, solving the problem:
- the probability of having a bad race assuming the boat finishes (between 5 and 15%)
- the probability of scoring a DNF (between 2 and 8%)
The actual implementation works as follows: flip a (weighted) coin to determine if the boat scores a DNF, if the boat doesn't score a DNF, flip a different weighted coin to decide between a good and a bad race.

Although I'm not really surprised by the results, I cannot really explain what's going on either. I'm continuing to investigate.
Cheers João & Huib!

Ment to write earler, unfortunately no time, but maybe even better now, with current 40'Series situation.
Before the last race, and without discarding the worst points, SimeMali (Class40) has a safe margin of 37 places behind Kipper1258 (Class40) and 38 behind psail (SOTO). With discarding, that margin is only 3 and 8, respectively. (btw. just posted race#5 xls at the usual place). Thus seems reasonably me vote for no discarding. But...

We all know that over a 6 month period there's a GREAT probability to have at least one race ruined because of work/health/family/other reasons. João, I entirely support your goals: 'Increase the competitive edge by approximating the different levels of SOL sailing expertise' and 'bring & keep more SOLers inside', and think discarding 1 bad race does exactly that: no one is eliminated by just one misfortune.

You bouth have right on influence of number of events and number of competitors, and I'm afraid we have here the worst case scenario: small number of races and large number of SOLers, so every point has great specific weight.

Because of that I'm thinking about a new way of scoring in 40 Series, and realy be glad to have your opinion.

My basic idea is a sum of normalized race times. In each race, winner's time should be considered as 1.000000, and all other race times calculated relatively to that (so we have +0.000012 for each second delay in 24h race, but +0.000006 per second in 48h race).

Example attached is for a few boats in Refeno Rally. As you can see, boat with 0.22% (not a mistake, it's 0.0022) slower time get +28 PTS, then for further 10% 17 PTS more, and further 0.3% 6 PTS more. That distribution doesn't seems fair.

With normalized race times, all races are equally weighted, the worst result can be discarded, and there's no cumulative delay for slower boats in longer races. This way also minimize influence of occasional, not in-Series racers on overall Series ranking.
(This model maybe can be applicable on TR Series too)

I'd like to hear your thoughts on this idea (and from other SOLers, oc :) ), also considerations how to handle DNF and DNS/DNC boats.

I'll try to do a simulation of that method on current 40' Series results, ASAP (but welcome any volunteer :))

Be well, stay well, FW!

Mladen / SimeMali

--- Last Edited by SimeMali at 2017-11-08 16:53:33 ---
Ibis redibis numquam in mare peribis
Attachments
As an example of proposed method, comparison on results of Refeno Rally Race, with graphical presentation.

Note two big, short-time, jumps and the influence of not-in-Series racers on SPTS, vs linear result distribution with normalized RT.

(non-Series PTS = not SYCM or has less than 4 races finished, thus not in position for overall 40' Seies high Ranking)

FW all!

Mladen / SimeMali
Ibis redibis numquam in mare peribis
Attachments
Viva Mladen and all,

Good afternoon.

I’ll make an introductory exposition as a simple SOLer.

One of the BASIC rules for racing in the 40’ Series was/is to use the same boat ONLY TWO TIMES per Quarter, in consecutive races or not.

This rule was/is well explained in the NOR for every Series race.
Also during the Series all SOL “departments” in different SOL “media” issued warnings about it: not complying with that rule would mean a DSQ for competitors.

In spite of all, and if I recall it right, there wasn’t ONE (!) Series race where we didn’t had any DSQs applied from infringements to this rule - I repeat, a BASIC one!

Seems quite linear till now but, the problems come later.

Even with just one non-comply case with this rule, there is necessarily need to carry out a race results redress after all competitors cut the FL.

Those corrections are made by only one person and “by hand”, who takes not only a lot of time to be properly done, patient as also a lot of their personal time, but above all, it isn’t neither reasonable nor bearable on the long run, by any means.

Before knowing in detail how the procedure for redressing race results was/is made, my recent suggestions to SRC for the 40’ Series were basically:

1 - No race discards (reasons already explained here in my previous posts);
2 - Modernize the actual fleet with new boats, faster than the actual used ones (Ker 40, SOTO 40, Class 40), in order to bring,
3 - more close competition and,
4 - more Skippers to the Series.

Afterwards, my thought was very different so my last suggestion to SRC was to change the actual Series format.

Sad and somehow frustrated to say it, but the actual status is not at all viable and, once more, few of us - the rules non-compliers - condition the vast majority of the remaining SOLers.

I don’t speak for SRC but probably the 40’ Series will continue in 2018, with a new format using only OD fleets, and for sure to be sailed with more exciting 40’ boats (some of them being prepped in the “SOL yard”).

SRC will have more flexibility on making the SOL races calendar, something we normally get for acquired but not easy to plan.

If we are going to use the “normal” process for (the potential future) 40’ Series, all the discussions related with redressing results function of the different boats used, are meaningless.

Anyhow, I shall analyze your suggestion and will come back with comments to it.
A final note. Other way to solve the issue in discussion is by means of “TCF” calculated alongside boats “Ratings”.

A Big Hug to all.

João/psail
Sail Fair.

Please login to post a reply.

Races

Next Race: 00d 00h 00m


Current Races:

Yates Cup 2017
RAYC welcomes Sailonline to the 2017 running of this Category 3+ race from Auckland around the Hauraki Gulf and challenges us to pit our skills against the real fleet who use the course as training for larger offshore events.
Race #1093
INFOfrom brainaid.de
Beneteau First 44.7 Particulars
WX Updates: 0430 / 1030 / 1630 / 2230
Ranking: SYC
ALT. CLIENT:Classic
Race starts: Nov 23rd 21:00 Registration Open!
GO TO RACE

Lake Superior Lights 2017
Sailonline loves lighthouses and our classic race around the famous lights of Lake Superior is a grand tour of 572nm . With fickle weather worldwide, this could be interesting!
Race #1092
INFOfrom brainaid.de
First47.7 Particulars
NAM_AWIP WX Updates:
0245 / 0845 / 1445 / 2045
Ranking: SYC
PRIZE: SMPF
ALT. CLIENT: Classic
Race starts: Nov 17th 18:00 Registration Open!
GO TO RACE

Elbetico's Fun Run TIMED Race 2017
This ever popular annual 20nm TIMEDRUN returns to our calendar slightly later in the year than usual. This year, our craft of choice is the SOTO30. Like last year’s Class40, the SOTO30 is a close-winded boat and has the additional upwind feature of a constant optimum TWA when the wind goes above Force 4. Chances are these aspects will stand to you, weaving your way up the mud flats of the upper reaches of the Huelva delta.
RE-REGISTER HERE to race again after finishing a run
Race #910
INFO by brainaid.de
SOTO30 Particulars
WX Updates:
0430 / 1030 / 1630 / 2230
RANKING:
TRCH-SUPSOL-TRQ4-SYC
ALT. CLIENT: Classic
RACE CLOSE: Sunday,
19 November at 2300utc
Race starts: Nov 10th 16:00 Registration Open!
GO TO RACE

SWR 17/18 Leg 2 - Lisbon to Cape Town
The second of eleven legs of the 2017/18 Sailonline World Race takes us from Lisbon to Cape Town!
In this leg we must catch the Trade Winds, avoid the Azores High, pick the thinnest spot in the Doldrums to go through, then avoid the St. Helena High and catch the Westerly Storm Track that will get us to Cape Town
The southern race boundary has been set at 51°S to mark the Antarctic Ice Exclusion Zone
Race #1095
INFOfrom brainaid.de
OD65v3 Particulars
WX Updates: 0430 / 1030 / 1630 / 2230
Ranking: OCCH-OCQ4-SUPSOL-SWR-SYC
ALT. CLIENT:Classic
Race starts: Nov 05th 14:00 Registration Open!
GO TO RACE

Go to race archive

SYC Ranking

  1. Sailonline Yacht Club Member bonknhoot
  2. Sailonline Yacht Club Member Alexandria
  3. Sailonline Yacht Club Member SimeMali
  4. Sailonline Yacht Club Member rumskib
  5. Sailonline Yacht Club Member rafa
  6. Sailonline Yacht Club Member longreacher
  7. Sailonline Yacht Club Member Kipper1258
  8. Sailonline Yacht Club Member Billy
  9. Sailonline Yacht Club Member sassy63
  10. Sailonline Yacht Club Member Dingo

View full list

Series

Mobile Client

SYC members have the benefit of access to our mobile/lightweight web client!

The mobile client