Facebook

Login

Support Sailonline

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

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

Board » General Discussion » Proposal for Performance Handicapping

Page: Previous 1 2 3 4 5 6 Next

I've had a go at developing 76t's handicap / performance system. The attached spreadsheet has several worksheets.

This is only based on 8 races held this year so far, and the top 150 in each race. This would need populating with more races, but hopefully this gives a good idea of what I've been playing around with.

The performance / handicap rating is initially established through 5 races (in this case 441, 443, 282, 437, 442). A better developed system would take your first 10 ten races (or whatever number we decide on) as the starting handicap. I just did it this way to have something to get going with. So some boats have average of 5 results, others have less.

The per race rating is based on the ratio your result (finish time) is slower than the winner. i.e. the winner has rating 1. 1 is the best you can achieve. Winston has this as he won all the races in the initial rating. Once the initial rating has been established, future races can have a handicap result ranking and a boat performance rating.

Look at sheet 438 (scroll to the top if it opens lower down). Column A is the race result ranking (line honours). L is the race result as a ratio of the winner. O is the handicap based on the average of the initial races. The performance is determined by comparing your race rating to your handicap. A higher number is better.

P is the adjustment based on your race performance. If you beat your handicap, then it is increased - see column Q (I've just chosen 10% of the difference for this example). Your handicap does not decrease for a bad performance. This avoids DNF or cruising affecting the long term handicap.

Finally, the handicap results can be determined, which is shown on column T.

I'll keep playing with this, but have a look and let me know what you think. If we are keen on this system, then we'll need a hand with some of the programming aspects.


--- Last Edited by NZL Scotsman at 2012-02-21 11:10:36 ---
Attachments
I've been doing more analysis of the race results from last year. So far I've analysed the first 20 races out of about 90. The aim of this is to establish the baseline handicaps for going forward. I thought I'd share some thoughts as I've been doing this.

1) How many races should make up the baseline? I'm assuming 10 at the moment. From last years results, if a boat did more than 10 races then I've averaged the top 10 results.

2) For boats that are new or don't race as much, should the handicap be provisional until a certain number of races are complete, for example 15 or 20?

3) Should we ignore results below a certain time, for example outside 25% of the winning time?

4) Should certain races be considered more important, for example ocean races?

Any comments welcome.
First let me state that I have no experience with handicapping systems and the like but in principle I support its aims.

As for weighting the results I think the number of boats (that actually race) and the composite strength of the participants in the field should be factored in.
None so blind
I think A2 has the right idea when it comes to weighting races and adjusting vs boats sailing.

I've played a bit with a couple variations of what I posted earlier, but have recently been thinking about (but not seriously trying to work out the details of) the following:

ELO - style system where for each race your time-on-handicap is compared to that of every other boat racing and an adjustment made based on relative performance.

Only boats that improve their rating get taken into account. In order to rate faster you have to beat a boat that has also rated faster...

I'm not quite sure of the implications of the second rule. Maybe performances near-handicap could be included? We need a way to exclude cruisers, unlucky sli etc or else the handicaps become a dis-incentive to race more frequently.
Regarding unlucky SLI's, missed Wx's on TWA or whatever may completly ruin a race. I had a couple of thoughts how to handle that for a Handicap.
But i didn't get any idea how to judge a wrong but continously and ambitious sailed Strategy from a SLI, an oversleped Wx, a lost connection or even cruising mode.

Leaving bad races out might lead to frustrated Beginners with one luckily achived good Position unable to get down to more suitable ranking while it simultaniously leads to suitable Handicaps for skilled but lazy Navigators.

Counting all races will have similar more or less unwanted Sideeffects.

And at least there will be two Groups of People; one that is willing to see how they are doing against very skilled Sailors with high-power Computers and very expensive Software in one specific Race and another Group willing to get the lowest possible Handicap in order to look like beeing skilled. ;-)

So i got to the conclusion that there is no right or wrong Formular and no Formular better than others. I would suggest to compute both and let Folks decide which they will use for themselves.
Most of the systems I've considered have relatively slow response rates. A boat that "got lucky" would have to do it 10 or more times in a row to bias their stat too much. At that point I'd start to argue that getting lucky is one of their skills/strategies :-D

There will be variations due to different approaches. Boats that "swing for the fences" mix big wins with big losses, even if they aren't cruising. What I'm proposing would handicap their potential, rather than the average. Boats that always do about the same are more likely to handicap out near the middle of the fleet more often.

In both cases, the SPH tells you quite a bit about how you did vs yourself for each race. The comparison vs the fleet is interesting too, but I see improving your own stat to be the primary goal, rather than winning on handicap. For a top 20 boat to win on handicap would require a stunning race (it's been done).

A top-half result on handicap lets you know that you sailed a good race. A race that changes your handicap lets you know that you sailed a great one - from a personal best pov.
I did think of weighting the performance handicap rating based on number of sailors. However, this is a bit tricky as we need to consider:
a) number of starters
b) number of finishers (this depends on the cut off, typically about 50% actually have a finish time)
c) does it make a difference if the top ranked players turn up?
To explain a bit more how the rating might change based on results: I mentioned 10% before. This would mean that the rating would increase 10% of the race result.

For example, say your rating is 0.9. This would mean that you typically finish about position 100. Say you have a blinder and win the next race on line. The race result would be 1.0 so the difference is 0.1. 10% of that is 0.01. So the new rating becomes 0.91.

You would need to win 8 races to get up to 0.95 and 23 races to get to 0.99.

Based on the first 20 races last year, our top ranked player winston has a rating of 0.999. Karmafish (#10 in rankings) has 0.989. The wonderful rainbowchaser has 0.970.

We may decide to vary the 10% rating improvement factor based on type or race or number of sailors. Just using this as a starting point for discussion.
If I understand it the performance aspect of the raw data is based on a time comparison expressed as a percentage.

76T & Scotsman certainly are more qualified to deliver a final product that suits this venue, all things considered.

I do have experience with another approach to the dataset that I would like to develop in parallel to see if it merits further consideration.

Namely converting the elapsed times into an Indexed rating. The slowest finishing time in the sample would be given a value of zero, the winning time a one and those in between a corresponding proportional value.

This then indexes the margin of victory as a basis of comparison rather than the overall elapsed time.

Perhaps it will yield the same results but I think that it would be a good way to evaluate races of varying lengths.

Comments welcome.
None so blind
I've been trying to do that, but picking a lower point gets tricky quickly. Different races have different numbers of boats, different strengths of boat, and different "difficulty" of course.

I'm messing about with the "use everybody that sailed at or above handicap" to try to define the fleet used for calculations.

Page: Previous 1 2 3 4 5 6 Next

Please login to post a reply.

Races

Next Race: 00d 00h 00m


Current Races:

Moheli PYOC 2025


For our first sprint of Q2, we will re-visit the African archipelago nation, L’Union des Comoros, and its most southern island, Moheli, in the Mozambique Channel. Created by MustangMark in 2014 this is a “Pick Your Own Course” round the island, approximately 20nm in our First 27. Expect winds to be light and variable; we are in the Tropics.
Race 1907
INFO by brainaid.de
First 27 PARTICULARS
WX Updates:
0430 / 1030 / 1630 / 2230
Ranking:
SPRQ2 - SPRCH - SUPSOL – SYC
Race starts: Apr 05th 09:00 Registration will open soon
▶ Flash
GO TO RACE

Africa by Sea - Tangier to Abidjan


Whatever about pharaoh Necho II’s Phoenician expedition of the 7th C BCE, the Carthaginian Hanno the Navigator with more certainty sailed south from Tangier to explore the West African coast two centuries later, but how far he got is a matter of more conjecture. However, if he did the 2300nm to today’s Ivory Coast he would have found little there other than good shelter in the vast lagoons behind the dunes, much the same as today, excepting the highly industrialized port of Abidjan. Not a great haven for a stop-over but let’s take the 90ft Monohull anyway.
Race #1855
INFOby brainaid.de
90ft Monohull PARTICULARS
WX Updates:
0430 / 1030 / 1630 / 2230
Ranking: SYC - ABC
Race starts: Mar 26th 17:00 Registration Closed
▶ Flash
GO TO RACE

Sea Lion TIMED Race 2025

Come on down South for long days of sunlight and strong bracing winds to race our Finngulf 43 as many times as you like - 117nm as the albatross flies - through some of the islands of the Falklands archipelago. This is a TIMED race, so you may RE-REGISTER HERE to try again, after finishing a run. You will have 13 days and 11 hours to test your skill and decision making after the race opens.
Race #1903
INFO by brainaid.de
Finngulf 43 PARTICULARS
WX Updates:
0430 / 1030 / 1630 / 2230
Ranking:
TRQ1 - TRCH - SUPSOL - SYC
RACE CLOSE: Saturday,
05 April at 23:00 UTC
Race starts: Mar 23rd 12:00 Registration Open!
▶ Flash
GO TO RACE

Melbourne Osaka Cup Double-Handed Yacht Race 2025

Welcome once again to what these days is Sailonline’s almost annual virtual Melbourne to Osaka Yacht Race. In real life, this double-handed 5500 nm race between these two sister cities, one deep in the southern hemisphere, the other high in the northern hemisphere, is run every four to five years, and this year is such a year. So, this year’s online version is in sync with the Melbourne Osaka Cup 2025 organised by the Ocean Racing Club of Victoria (ORCV), and the Sandringham (SYC) and Osaka Hokko (OHYC) yacht clubs. We’ll be racing the well-known First 40, a popular size of boat which should be a good match for many of the boats entered in the real race. With the doldrums unavoidably lying across our course, you can expect to be at virtual sea for at least a month!
Race #1669
INFO by brainaid.de
First 40 Particulars
WX Updates:
0430 / 1030 / 1630 / 2230
Ranking: SYC
Race starts: Mar 16th 06:00 Registration Closed

▶ Flash
GO TO RACE

Go to race archive

SYC Ranking

  1. Sailonline Yacht Club Member Patrick70119
  2. Sailonline Yacht Club Member WRmirekd
  3. Sailonline Yacht Club Member CriticalHippo
  4. Sailonline Yacht Club Member vida
  5. Sailonline Yacht Club Member KaSToR
  6. Sailonline Yacht Club Member rafa
  7. Sailonline Yacht Club Member FreyjaUSA
  8. Sailonline Yacht Club Member Kipper1258
  9. Sailonline Yacht Club Member bonknhoot
  10. Sailonline Yacht Club Member rumskib

View full list

Series

Mobile Client

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

The mobile client