What would the rating be without the spinnaker?

Download the calc rountine from here and enter the values from the site and play to your hearts content!

So how does SCHRS/Texel look like along side all the different PY systems? are they close or poles apart?

2- 3 % would be fair <img src="http://www.catsailor.com/forums/images/graemlins/crazy.gif" alt="" />

just putting weight difference into schrs formula wouldn't be fair ,

true weight difference doesn't transfer to weight in boat as crew weight is dynamic and not as big a penalty as a fixed weight

In ideal world if fattys could get this sorta allowance I would put into extra power to drive my bulk rather than look for handicap advantage , who wants to win that way .

Depends on the cat. I did a comparison between PY (simple DPN number), Texel and SCHRS based on the following assumption: A race where an F18 sails 45 minutes real time. What real time needs to sail a H16, an A cat (Flyer in SCHRS) and a Tornado to get the SAME corrected time in the different rating systems?

Diff_F18 PY2006 Texel SCHRS
Hobie 16 588.5 427.7 378.8
Tornado -147.1 -187.1 -188.1
A class 95.2 0.0 -69.9

(difference in real time with respect to F18, negative means he has to finish earlier, positive later, units is seconds)

I.e. in PY a Hobie 16 has to cross the line 588 seconds after the F18, whereas in SCHRS he has to finish only 379 seconds after the F18. The difference is 210 seconds or 8% which is the biggest deviation in this table. Between Texel and SCHRS the deviations are smaller, a H16 has 50 seconds more with Texel which is 2%.

Of course one should better apply the BF numbers for PY, but this calculation gives an idea about the ballpark. At say 10 knots boat speed which is 5 m/s 200 seconds are 1000 meters!

Not a specific reply to anybody.

Amazingly, there are literally thousands of clubs using Sailwave; that's why I thought about using the available data for a rating system. However, perhaps more importantly, I intend to make the data itself public. Obviously this won't be done w/o club's permission but, since it's personally anonymous data I don't foresee a problem. I intend to include facilities to easily record/report wind speeds, course, sea state, region and a very crude sailor ability(*), which I realise will all have to be optional. Essentially a centralised mass of data that can be mined by anybody... A simple click of a button is all it will take to send it to the Sailwave server (it has to be easy to get participation). There are some issues like consistent class names and configurations etc but they can be solved. Essentially I figured, why waste all that information when it's sat in a common piece of software in so many clubs - let's grab it and see what happens; it's irresitable given the numbers...

I get a lot of emails from clubs in Countries that do not have their own rating system asking what to do, which is another reason for the plan, which was in fact formed years ago on the SUG... I get the feeling these clubs would be more than happy to be involved in somehting like this... They may not have a MNA like USSA or RYA but they have broadband...

(*)Something like this:-

1. Occasional racer
2. Regular racer
3. Has participated in a weekend regatta
4. In top 3 at a weenend regatta
5. ...
6. ...
10. Has won an international event...

Crude but potentially useful data....

PS: there is a new version available that hides infrequently used stuff.

http://www.sailwave.com?page=download

Edit yet again: Codblow: and something for weight bands... and for the MarkP-type sailors (including myself), something about beer intake during events... <img src="http://www.catsailor.com/forums/images/graemlins/smile.gif" alt="" />

Colin - I'm an absurdly grateful fan of Sailwave. If you add the functionality you've decribed, I forsee a ratings system with world-wide acceptance.

Hi John,

Thank you for the encouragement; it is appreciated. There are a few common topics in the emails I receive and ratings and especially personal handicaps are 'up-there'. Since each club has it's own PH system I envisage a procedure whereby they report the data to the server then suck it back en mass to mine themselves using scripting... Grass roots club sailing is the heart of the sailing community, but absurdly there is a current imbalance in Sailwave that needs addressing in that direction...

With reference to Claus response, it seem that there is a big imbalance between what is happening on the water ( PY ratings) and what can be predicted using simple formula. I think this is why a lot of clubs/ sailors are reluctant to jump into measurement rules if they know they will be at a disadvantage before the gun goes. This may come from a number of areas: One design sails, old hull design, basic deck gear, too light, too narrow etc. How do you address this problem in a formula rule? Don't get me wrong I believe in the formula route but obliviously it needs constant adjustment to even the playing field and still holding on to the single number rating. This is a tough nut to crack.

Maybe the best solution is to have both systems in one, so you have the information from Colin sailwave program and it pumps out an PY adjustment (fudge value) number that's added to the SCHRS or Texel rules and you should have a more precise rating system, the best of both wolds. Each system keeping the other in check. what should happen is with all the returns to the sailwave system the fudge factor will get tighter and tighter. Okay Wouter and Simon I'm ready to get flamed.

On a different tack which is giving better results on the water between Texel and SCHRS, with both being formula rules they should be coming out close, from Claus attachment you get big difference the further you go away from the F18 base line ratios, the Hobie 16 is old and a OD and the A-Class too light. What you are seeing is type forming, it happens in every formula rule, it's the nature of the beast.

Maybe it is actually the PY numbers that are wrong and not the measurement system ratings ?

For some reason people always believe that RAW statistical results are more in line with reality then mathematical models smoothing the raw statistical data into properly balanced tendencies.

This is a very grave error. An understanble error maybe but a grave error nonetheless.

Compare it to this. Take you GPS track log and remove any smoothing (averaging) of the speeds, you are very likely to see more then 1 or 2 spikes in your speed data. There is simply no way you can accellerate from 10 knots to say 25+knots and back again in 10 seconds. Any real life system will be disturbed by such factors, many systems will be HIGHLY disturbed by such factors all the time, in that case we say the data is disturbed by noise. Compare the last situation with old AM radio's and walky-talky's; the cracking and hissing sounds making speach almost indelligiable. Modern devices you a score of mathematical smoothing and post processing technics to filter out the speach from the background noise/disturbance. And the measurement based ratings system actually so a very similar thing.

Measurement based rating systems force a smooth function (or graph) to be fitted to heavily disturbed measured data. This graph is then the best fitting for that dataset given its enforced limitations on smoothness. So yes, it may distort the pitch opf the "speach" at little but it is still a [censored] load better then the old way of trying to hear what somebody is using over the screams and clicks of the static (background noise).

This graph below gives such a example of fitting a very simple mathetical model to a very disturbed set of measured data.


Raw data and the real underlaying system




Smoothed data and the real underlaying system





Now any rating system that is entirely based on raw statistic data will have to work with a cloud of distributed datapoints where no simple assumptions with respect to averaging can be made.

An example : If all data is taken from fleets that have active group of skilled racers then simply averaging the raw data will produce smoothing. But when some fleets are raced much more actively (F18, A-cats) then others (Prindle 16, supercat 17) then simply averaging the raw data will results in relatively large biases with respect to the real (yet unknown) underlaying system (situation).

Of course, in our field of interest, sailboat races, we can simply not assume that the skill level of racing crews is the same over all racing fleets (boat types). Actually we know for certain that very large differences exist EVEN inside a single class ! Example; Mischa Heemskerk coming over to the USA grabbing a crew from the beach and putting in a line of straight bullets in any US F18 regatta while he can't be that chavalier about it in any EU F18 race.

Now with sailwave we are proposing to mix this EU data with US data and expecting improved accuracy ?

I will tell you what will happen, you will loose accuracy as you will be averaging the relative difference in fleet skill between EU and USA and deduct that from the REAL F18 designed in performance which will at least be the same or just above the EU level.

Having more data simply does not garantee higher accuracy. There is such a thing as poor and good data.

Next points made in my follow-up posts.

This "system identification" stuff is actually a standard skill in my line of work.

Wouter

"Old sails" should never become part or a rating system. A crew keeps their craft in racing order or not. Factoring in "old sails" is a nightmere. It leads to many new questions (what is old for dacron what is old for pentex ?) that are very hard to answer and requires detailed compliance checks at events that nobody really wants.

Other factors like width can very easily be factored in into a measurement based system and these element has been factored in recently in Texel.

Again, making a measured based rating system reflect such differences is not a difficult task. Getting the race committees to actually accept and use the new (multiple) rating numbers is however a very difficult task.

Ergo, the problem is not technical in nature but human (as most problems in the world are)





Ohh that is very simple, there are standard numerical tools for such a thing.

Basically you load a large dataset of race results into your computer memory with data about crew skill level, wind conditions and boat specs and do a regression.

Regression means that you first select a limited number of influence factors, say length, width, height, weight, area, boarded (yes/no) and crew makeup. Then you choose a model structure, most often a linear relationship in such cases. And then you have the numerical algorithms find those weighting factors that make the inherently smooth "graph" approximate the race results the best as is possible given this limited number of influence factors.

Why is this advantagious ?

Well, say you have regressed the system on a dataset that is composed only of actively raced fleets that are spread evenly over the range of beach cats; from 14 foot to 23 foot. The produced results is pretty accurate in rating these fleets. Now when a new class is formed that shares some aspects with a 23 footer and other aspects with say a 16 and 14 footer then its rating will be darn accurate the first time around. Because the regression will tune the describing formula to each individual influence factor on a well conditions and dependable dataset. It will therefor give a pretty accurate weight to each, making any new combination of specs a sum of well known factors.

This means that accuracy achieved by regression on other fleets is passed over onto fleets and designs that are not actively raced. This is an advantage that can NEVER be had with methods based solely on pure statistics. Here only actively raced fleets will get accurate numbers while all others will just swing about some incorrect rating that is more reflective of the skill level inside a class then of the design itself.


Again similar aspects are used to accuraty descibe systems in engineering despite strong disturbances.





All rating systems suffer to the same exend when forced to predict performance over a wide range of conditions and course by a single numbers. Being "Yardstick" or "Measurement" has nothing to do with it.

Also it is actually the Yardstick systems that need to continiously adjust themselves when gaining more data and NOT the measurement based systems. Measurement based systems, as described above, are very succesful in getting a rather accurate rating the first time around by using "knowlegde" gained when rating older more active classes. When the initial rating is already rather accurate then future adjustments are negligiable or very small indeed.





Not a bad idea, but the very best solution is to have Sailwave create a very large pool of DEPENDABLE race data that is then used for a new regression of Texel/ISAF. Of course a re-evaluation of the influence factors is then also in other. One of the more interesting ones being the factor where a spinnaker is added to a sloop rig making the jib much less effective on downwind legs.

I'm against Fudge factors as they are tuned by human beings with convictions and believes that are often counter scientific. I would just have a human being or a committee decide on the influence factors and model structure and then have a numerical algorithm produce the actuall regression.


]


A heated discussion does not equal personal disrespect !

So don't worry here.





There are some differences between both systems at the fundament. Texel is a pure regression executed on a very simple modelstructure where human inspired modifications were later added to tweak the number. SCHRS uses a far more elaborate modelstructure (or used to, prior to changes made in 2007) with a less pure regression, it also has less afterthough modifiers.

In my opinion both systems should just merge into one system and do a new PURE regression on a larger set of influence factors. The new resulting system will then be more accurate, more transparant and simplier then both.


Wouter

Wouter, I said One Design sails, not Old sails

same thing <img src="http://www.catsailor.com/forums/images/graemlins/grin.gif" alt="" /> (joke)

By the way you actually mean SMOD sails as OD still allows you to purchase modern sails that aren't that much slower. I think that is just one of those drawbacks you have to accept when sailing in a SMOD class where the monopolized builder is having a crap quality control system in place.

There is no reason why a SMOD builder can't sell up to date class compliant sails other then his desire to maximally milk class members for money by sourcing our the production of sails to some Chinese sweat chop.

Wouter

Nice post Wouter, I need to go off and have a think about this one, but here's one for you "Under SCHRS/Texel Gavin Colby is racing himself in a F18 Tiger and a Hobie 16 in a W/L race, if the measurement system was perfect then he should have the same corrected finishing times for both boats, my money would be on the Colby F18 correcting out the winner"

NO, other conditions play an important part as well.

BUT, if he did each of the following races in the following places:


1, W/L race
2, Triangle races
3, Squares
4, mixture of all the above

In the following wind conditions

A, F1
B, F3
C, F5
D, F6+

And in the following places

I, Small lake – so small courses, less chance to really fly the F18 kite.
II, Large lake – smoother water
III, Seaside – smooth water
IV, Open sea – lumpy when windy

So 4x4x4 = 64 races.


Then hopefully a measurement rule would, overall, score the boats the same.

All boats have sweet spots, the 16 might do better in the big conditions in big waves and will also fare better in the square courses as the F18 will have the kite up less.

I actually think the situation is a little bit more nuanced.

In light and medium winds Gavin on the F18 will win more often then loose against himself on the H16,; that is with the current single number measurement based handicap.

In strong and very strong winds Gavin on the H16 will win at least as often or more often then loose against himself on the F18, again with the current handicap ratings.

This is mostly because of the "need" to rate performances that can change from condition to condition into a single number. Often this means that there is a cross over point somewhere, at some wind speed and seastate combo.

We can make the measurement rating systems almost as accurate as we want, but only if we are all willing to put in the effort to measure more specs of each boat, get measured for checks at events and put enough pressure on the organisors to accept the new system.

Like I said, when push comes to shoove we can even train an artificial neural network on the carefully gethered and labelled race data and use that.

Wouter

Wouter
You have to come to the F16 Global Challenge this August. We just wont accept any excuses. You and Scooby can have a great event discussing your handicaps!! I'll have to pass thou cause I've got to concentrate on my drinking problem. <img src="http://www.catsailor.com/forums/images/graemlins/grin.gif" alt="" />

Bollox to that, I'll concentrate on helping you research your drinking problem. I may declare myself officially off duty at certain times, Maybe I'm "on Duty" between 18:00 and 18:01 each day.

I fear that I will just have to disappoint all you guys and myself. I really wanted to do Mumbles because of all the positive reports last time, but I really don't see how I can given my personal circumstances at this time.

Wouter

What's that saying.........a quitter never wins and a winner never quits!!