Monday, 23 January 2017 09:49

Regatta format

Richard Hart explains a new system for the allocation of random groups in large fleets. This has so far been used in both 2015 and 2016 Finn World Masters.

In 1961, my club ran a weekend regatta for the new OK Dinghy Class. As the day approached, it became clear that the number of entries wouldn’t fit on the start line on our Pond, so I was asked to produce a list of four groups who raced ‘heat’ races in pairs: A against B concurrent with C against D, then AC/BD, then AD/BC. On Sunday afternoon there was a final race for a Gold Fleet and a Silver Fleet. I ‘seeded’ a likely top four into different groups, then a second four, then just wrote down everyone else to make the numbers right. I thought the system worked well, perhaps because I won. I hasten to add that the allocation of the ‘seeds’ into groups was done at my parents’ dining-room table, with the aid of a die. Our family cat was a witness.

At the Masters’ 2003 in Schwerin we used the same system, except that, so far as I know, there was no ‘seeding’ so no cat was needed. Using that system, there should be three, six or nine pairs of heats followed by a Gold Fleet Final and a Silver Fleet Final. The weather had other ideas – we had five sets of heats and no finals. 25% of the Fleet raced against Eberhard Bieberitz in every race (to be fair, 100% sailed against him at least once). I didn’t hear anybody complaining about the system: we just realised that it was the best way anybody had developed to handle large numbers of entries. Nowadays the Masters’ can expect fleets so large that even two races for a heat is not enough. If we have 285 boats then each of four races for a heat (I’ll call them ‘flights’ to avoid confusion) will still have 70+ boats on the line.

What should be done about all this (it’s an issue for many other classes as well)?

Limitation of Entries
At the Finn Gold Cup in 1963 we had over 160 boats on the starting line at Medemblik, and since then we have limited entries for the Finn Gold Cup and the Europeans. In fact the starting lines weren’t so difficult as at some recent regattas with far fewer entries, because at each end of the Line there was a good big Committee Boat that you could see. For various reasons, ISAF has gone down this standard path of limiting entry by using qualification systems. Most of the Masters seem to dislike the idea of limiting entries if it can be avoided, and I think that’s right: Do we promote the sport of dinghy racing by stopping people from going to regattas?

What are the alternatives?

Separate Age Groups
One is to race the age groups separately. On balance, it seems that most of us don’t want that solution: we all want to race against Maier and Budzien (although perhaps not in every race). Because of different numbers in different age groups we’d just have another lot of problems anyway.

Opening Series and Final Series
Perhaps the most frequently used system for large fleets is to have a selection series during the first half of the week, using some form of arbitrary allocation for the first day, then re-allocating daily using some system based on daily overall standings. For the second part of the week, the fleet is divided into Gold and Silver Fleets.

I think that there are huge disadvantages with this type of system, some in particular for our FWM Championship, and some that have been demonstrated at a high level and in major championships recently.
1. Traditionally, championships have been raced over as long a period as practicable – usually a week – in order to test the skills of sailors under various weather conditions, and so on.
If we split the regatta into two series, each of half a week,
• there is a greater risk that the regatta will be spoiled by days lost due to weather,
• It will be more difficult to reschedule around days lost from one or both of the shorter series,
• We always have the chance of a ‘light weather regatta’ or a ‘heavy weather regatta’, but we add the chance that the weather conditions will be (for example) light during the opening series and strong during the final series.
2. After the split part way through the week, there is no way to change Fleets. For the Masters’, this means that among the age groups it is likely that some leaders will be separated from each other, irretrievably, half way through the week.
3. I have an over-optimistic opinion of my abilities. When I end up in the Bronze Fleet after hoping to be in the Silver Fleet, maybe I’ll go home, or maybe I’ll just not bother too much about being over the starting line from then on.

Full Series with daily re-allocation
In our search for a fair and enjoyable FWM, we have developed to the present system where we are allocated to one of a suitable number of ‘flights’ for the first day, and then re-allocated on a daily basis for each subsequent day of the week.

For the first day, the Finn Masters have already agreed that allocation into flights should be done at random, and it seems to work.
For subsequent days we’re not unanimous about what is best. We’ve tried allocation by some arbitrary system such as #1 to Red, #2 to Green, 3G,4R, 5R, 6G, 7G as for the ‘Opening Series’ scenario. For most of us, this is effectively a random allocation. Right at the top, it may be fair because it is arbitrary, but should the system separate the first two if they are both on 7 points and 50 points ahead of the third boat? What if one is on 7 points, the second on 8 and 50 points ahead of the third boat? What if the second boat is 50 points behind the first but 3 points ahead of four other boats? I believe the best ‘arbitrary’ allocation would vary, depending on the points standings and on whether we are near the beginning or the end of the week. In reality, we have to choose a system and accept that it is a compromise.
The re-allocation process has to be done in a hurry during the evening, perhaps after protests have been heard.

We have already tried daily allocation at random, and it was a great disappointment that some sailors felt that this resulted in ‘flights’ of different levels on the last day. One of several reasons suggested was that the random number generator didn’t.

A first attempt to make the random system more credible was to look for a better generator, and here I was helped by Jiri Outrata: we served together on the Technical Committee, and he is known to many of us as a sailor competing at Finn Masters’ events. Lukas Adam, then one of his students, prepared a programme to allocate boats into groups as required.

Full Series, Pre-Allocation
A second step in this process is to suggest that all the allocation be done and published before the first race. So far the Masters’ have not taken this second step – this letter is a promotion for it.

What’s this ‘random allocation’ anyway?
It’s a grand way of saying ‘throw a die’. For two groups, if it comes up with an odd number, allocate to red. If it comes up evens, allocate to green.
    
Simplification
While thinking about how to operate the system in a fair and transparent way, I realised that we don’t need to produce a new lot of random numbers for each day. We only need to do it once ever. All we need is a very long list of ‘records’ each allocated at random to one of two flights, another list for three flights and so on. (I’ll call each one an ‘Allocation List’ and we can tell the Computer to do that sometime during the Winter). Then we need to have a lottery arrangement to find where to start.

Bending the Mathematics
So far I’ve been staggering uneasily through mathematics (not my forte). The random list we’ve just made has produced a collection of sequences, such as R,G,G,R, which come round again quite often. We accept that easily, until a sequence comes up such as R,R,R,R or G,G,G,G. We can tell the computer to discard records like that. The remaining Allocation List may not be random, but it’s made before the competitor’s names are attached, so it’s still fair. (The adjustment doesn’t address the chance that two people will be in the same flight all week and that two people will never race together during the week, but the other allocation methods won’t either).
    
To do it
Long before our regatta the Allocation Lists are prepared, one for regattas using two flights, one for three, one for four. (No sailors’ names are involved at this stage.) This only needs doing once ever.

When the number of competitors (and committee boats) is more or less known, Andy decides how many ‘flights’ we will race in during the week.

At the regatta
1. The Regatta Office have a List of Entries, arranged in order of Sail Number in use, from ALG1 through to ZIM999.
2. After Registration, the Regatta Office remove ‘no-shows’ from the Entry List to give a List of Competitors.
3. At the Opening Ceremony the local Mayor thinks of a number (say 21). The Regatta Office puts the first boat on the List of Competitors against Record 21 on the Allocation List, to show her Flight for Monday.
4. The next boat goes to Record 22 and so on.
5. When all the Boats have been allocated for Monday, the next Record is for Boat 1 (ALG1) on Tuesday, and so on.
6. If the allocations put two boats more in one flight than in another for some day, then one boat gets switched. The scoring system in the SIs defines which.
7. A list goes up on the Notice Board, telling us all which colour flight we are in, for each day of the week.
There are no hang ups during the week, waiting for protest decisions or for the Jury to use existing results, or just waiting for the list to go up.
    
So far as I can see, this system would be transparent, robust and easy to administer. There are two elements of chance (practically) that govern the application of the Allocation List:
• The number the Mayor thinks of, which governs the Monday starting point on the list.    
• The number of boats registered. Once the Monday starting point is fixed, this governs the starting point on each subsequent day.

The Medal Race
This letter is not taking an opinion about whether to have a Medal Race. The ‘Full Series, Pre-Allocation System’ discussed here could accommodate the Medal Race as easily as can single-group racing. Equally, it could stand alone.
If we have a Medal Race, then on the last day the ‘Top Ten Plus’ people do their thing. The rest of us do our flight races as programmed at the beginning of the week. We are encouraged not to pack up and go home because, with the top ten elsewhere, we might finish a few places higher than we did earlier in the week.

The Small Print
There are some places where the Scoring System can make people happy or sad. One is the Medal Race Qualification, where we have already adopted the ‘Top Ten Plus’ arrangement, where if several of you (not me this time) are on equal points to #10, you all go to the medal race (no need for the scoring system to choose between a first in Green Flight and a first in Red).
Another is the arrangement for dealing with flights ‘orphaned’ because a concurrent flight has been thrown out or not sailed. The SIs should define how to sort this out without removing the scores for the flights that did succeed (these are always where you had your best result of the week).
    
Conclusion
Writing all this, I’ve been rather grand in saying that ‘We’ or even ‘I’ have done or decided this or that. Not so: I’ve made a lot of noise, but the decisions have been undertaken by the Masters’ Committee led by Fons and then Andy Denison, also by the Race Committees who have been making the developments work so far.

What’s suggested here is very little different from where we have arrived already. The main differences are that
• Odd-looking allocations were removed fairly at source
• Most of the sorting work is done before the regatta
• Right at the start of the regatta, we know what flights we’re in for the whole week
• it seems easier for everyone to work with, and
• it’s more transparent.

I’ve tried to work all this out because something is needed. I’m not a mathematician or a computer person. If you see blunders or places for improvement, please let Andy know.

Acknowledgements
Lots of Finn Friends. Jiri Outrata and Lukas Adam.
Alen Kustic, our PRO at the Masters’ in Split 2010, who pointed out a total failure in the logic when I was trying to incorporate the Medal Race into the Main Points Table.
Henk de Jager, for inspiration. Robert Deaves for inspiration and careful scrutiny. Andy Denison for support and entrusting the regatta to this system.
 

Quick links to news

schwerin2019

portzelande2020

SEARCH THIS SITE