Hi all.

The True Rankings are pleased to announce that Dragon Boat Canada has agreed to host the rankings and that we will be working to develope a Canadian Ranking system if possible.

The new home of the rankings is of course:
http://www.dragonboatcanada.org

PD

Yaaaaaaaaaaaay!

Good job!

And they said nothing good could come of this forum... :-D

With all due respect (and thanks), have you been able to do anything about the blatant "Pickering Bias" from last year?

It seemed like this one regatta was the basis for the entire system.

I'm totally happy to answer questions, but please try to have some kind of objective criteria pointing to the data instead of just coming up with an unfounded critique.

Hmm...Pickering Lane Bias throwing off the rankings. Let's see, G&G is worth a discussion, but lets look at the rest of the field.

Team Ranking Time - Pickering
CDN. Senior DB Club 2:02.1 - 2:02.1 (Same)
Mayfair Predators 2:02.2 - 2:02.0 (Same)
Scotia Rouge 2:05.2 - 2:04.9 (Same)
Hanalei 2:06.7 - 2:05.3 (Slowed down at GWN, faster in PIC)
Hammerheads 2:07.4 - 2:07.3 (Same)
Imperial Dragons 2:07.6 - 2:07.6 (Same)
Verdun DBC Adrenaline 2:08.2 - 2:08.1 (Same)
SRS Rouge 2:09.1 - 2:09.1 (Same)
Collingwood Sidelaunchers 2:09.3 - 2:09.1 (Same)
Mofos 2:11.6 - 2:12.6 (Found legs at GWN, slower in PIC)

I'm mystified as to why one might think that Pickering is throwing the times off while the crews performed very similarly at the other festivals they did? The comment that somehow the Pickering data is corrupt seems highly unfounded.

G&G however, is a real question, is it possible that having only done two festivals, and coming in a whopping 4secs ahead of the competition in PIC could have been through lane bias?

Let examine:
G&G Lanes: 3,4,4
Mayfair Lanes: 6,3,6
Seniors Lanes: 2,4,3

So the seriors had the same lanes for all but one race. G&G had a 4 while the seniors a 2. Given they finished 4secs ahead of them in Pickering's true ranking time, and given this was averaged over three races, then the lane advantage of lane 2 over 4 must have been 3*4seconds, or 12 seconds.

Given a random sample of teams, I really can't see much of a clear pattern of lane bias at all - if anything, 2 looks faster.

4 Hanalei Mixed 02:05.6
2 Hanalei Mixed 02:05.0

2 Shogun Warriors Mixed 02:10.0
4 Shogun Warriors Mixed 02:13.0

2 CDN. Senior DB Club Mixed 02:03.4
4 CDN. Senior DB Club Mixed 02:05.8

2 A Long for the Ride Mixed 02:19.2
4 A Long for the Ride Mixed 02:17.6

4 OHDBC Premier Women Women 02:11.6
2 OHDBC Premier Women Women 02:10.4

2 pdbc Pink Sensations Breast Cancer 02:40.6
4 pdbc Pink Sensations Breast Cancer 02:38.5
2 pdbc Pink Sensations Breast Cancer 02:38.5

2 Pickering DBC B Mixed 02:14.1
4 Pickering DBC B Mixed 02:15.8

4 Pickering DBC C Mixed 02:34.9
2 Pickering DBC C Mixed 02:32.1

2 Ryerson University Rye-D-Boat Mixed 02:18.6
4 Ryerson University Rye-D-Boat Mixed 02:21.5

4 York University Seawolves Mixed 02:28.0
2 York University Seawolves Mixed 02:25.8

Therefore, I would wager that the faster G&G time in Pickering was legit and that at this early point in their season they were ahead of the curve or simply a better team.

Cheers,

Power D

Whoa, hold on there big fella.

I never said anything about a Pickering lane bias. I only inquired if you had found a way around what I called the "Pickering Bias". It just seemed that you based everything in the Power Rankings on the results of the Pickering Dragon Boat Festival, and that every other result was adjusted to fit in with how Pickering turned out.

Buddy. That's how statistics work. You need a base case.

Get a fricken clue

Hopefully PD can explain it better than that.

Surely the "how statistics work" isn't that you take one set of results and then wedge all your other results into that mold and throw out what doesn't fit.

Wait a minute, maybe it does!

I'll take a stab at it

It's not that Pickering is weighted more heavily than other regattas.

The "base case" is used to determine if one festival is slower or faster than another. It's more accurate than just averaging all of the times because as we know some race courses are faster/slower than others.

If there are two festivals A and B you can look at all the teams that went to both festivals and from that you can tell which regatta is faster. You can adjust the fast regatta to make all of the times slower so that they are comparable to the slow regatta.

Holy smokes. Using Pickering as a base case is irrelevant. The "official time" is irrelevant, these times are just used to rank each individual team and, as someone mentioned, to adjust the respective regatta for speed biases. So if we use the island, Pickering, Stratford or GWN... it just doesn't matter.

Thanks G. That is pretty much bang on.

The actual base case (and maybe the reason I got thrown off of the question asked) is Toronto Island. This just sets the benchmark for times (for example a fast crew in TO goes 2 minutes – then the fastest rankings time will be somewhere around 2min). If we had chosen Welland as the base case, with the very slow boats last year, then the fastest time might now be 2:30 minutes. However, the teams would all still have the same ranking order and relative spreads. Thus, it is pretty much irrelevant as to what festival is chosen as this just sets the baseline for the times shown and won't influence how teams rank.

Cheers,

PD

Sudden thought came to mind.

So we already know that some regattas are faster than others, and of course, some are slower. But this increase/decrease of speed is not an absolute factor. Would there be a logarithmic or even an exponential factor which takes place?

Yes, the ranking would stay the same, and so would relative speeds, but the spread would increase.

So instead of the Beast being .4 seconds behind Mayfair, in the example above with the Welland boats, they might be .6 seconds behind Mayfair.

PD

Well my first suggestion to is to have DBC stop referring to these as "national rankings" as I didn't see any Western crews on the list at all. I'm all for a national ranking system if you can come up with one, but right now these rankings ARE NOT NATIONAL RANKINGS! If they were, then you'd see FCRCC, Gorging Dragons, Pacific Reach, Pemberton Eagles, and many others listed amongst these teams. It's a big country people...

The way I read that was they are planning to include western, central and east coast teams for this year. The data on their site right now looks like it was copied directly from:
http://www.dragonboat-truerank.blogspot.com/
and
http://dragonboat-truerank2005.blogspot.com/
which were for Ontario/Quebec based regattas, before DBC association.

If that's indeed what happened then that's understandable. They may just want to make note of that so nobody out west gets their noses bent out of shape.

I hope they can make a TRUE national ranking system, but there will be many issues, especially with the need to include Alcan in the mix of festivals. I'll be curious how they adjust for different tide conditions in the calculations, as well as what has always been a rumoured "lane bias" in that race. Harbour racing can be much different than lake or canal racing.

Read the section under Methodology. It explains in great details about how regression takes care of the issues you mentioned.

http://www.dragonboatcanada.org/rankings/Methodology.htm

It's summer in Ontario and that means one thing: road work. After reading some of the strings, we know where to find extra pylons .....

Shut up and paddle.

In the immortal words of the Master of Wit and Repartee...

"F#ck off, Clown."

Hey Mr. May 14 2007, 2:42 PM.

So typical of the forum: talkin' wannabe, too busy sitting on the couch, trying to find the remote, and picking the Doritos off their Randy River track suit.

Less posting, more paddling chunky!

Oh look, you wasted 30 seconds of PERFECTLY GOOD PADDLING TIME on that ridiculous and hypocritical post. Better do some extra chin-ups tonight, pal.

Really, anyone who wastes their time by going on the forum to type "shut up and paddles" is an imbecile. Do you not see the irony here? I'll spell it out: you're not paddling yourself but you're yelling at OTHERS to "shut up an paddle?" Shouldn't YOU be paddling? Quick, grab your paddle. Go. NOW!

I'm wasting my time here too, but at least I'm honest about it.

Why does there need to be an allowance for conditions? Every paddler races in the same onsite conditions at any given festival.

Of course the locals will have the benefit of detailed inside knowledge about the course itself, but that's no reason to skew the results. I'd be willing to bet that you can find differences between lanes in EVERY dragon boat course. Unlikely that the depth of water and the direction of underwater flow is consistent between each lane. Now if you happen to get the shaft and draw the "sh!tty" lane for more than your fair share of the festival, that obviously sucks. But them's the breaks. Conversely, you could draw the premium lane more than your fair share of the time and end up placing (and therefore being ranked more highly than your team should have been).

If you simply average festival times, you wouldn't be accounting for differences in boat speeds, course length, weather conditions, and course depth.

By taking these into account, the only thing you are really not adjusting for (at least to some extent) is lane bias - which tends to wash out as teams do more and more races.

Just want to congratulate PD for the amazing he's done with the Ranking, despite some harsh (and I feel undeserved) criticism.
I think it's safe to say that it must be ZE biggest positive thing that ever came out of this forum
Would be nice if there would be more positive things coming out of this forum.
The Dragon Boat community at large, needs more positive output and ideas than all the trolls are seeding on this forum

Thanks Montrealer!

Since their inception in 2005 I've had a tremendous amount of positive feedback from the community which I greatly appreciate. The small amount of negative feedback usually result from a misunderstanding of how the rankings work or a fair critique of the approach.

Cheers,

PD

Which races will be added?

Question for you. How will DQ's be factored into the rankings. So crews with 5 or 10 second infractions will count. What about the victimized crews who lost their races? It kind of benefits an A crew who gets DQ'd but has no time.

As many festivals as can be added will be. Thanks to help from the Dragon Boat community in collecting the data, this should result in the best coverage yet.

Last year's coverage:
http://www.dragonboatcanada.org/rankings/2006_byFestival.htm

Races with time penalties or DQs get dropped as they are not considered to be a fair assessment of boat speed. The team with a DQ simply gets one less race averaged into their ranking.

PD

Cool! The rankings now list websites and locations.
http://www.dragonboatcanada.org/rankings/Annual.htm

HEY PD,

do you need the lachine festival results ? I have an excell file for that .......

When are the 2007 rankings going to be posted? Or are you waiting until after the Islands are done?

I would imagine after the islands or Ottawa, as he would have to have a few races in hand to run a better regression.

GTA area will get done after Toronto Island.

The Montreal area usually after Montreal.

Stouffville has been added.

PD

Thank you!
Keep up the good work.

Michael

I can't wait for the post-Ottawa update.

Bye bye Rouge!

I bet Mayfair is ahead of Events.

The kids were racing in Ottawa as "WRCC Masters A."

What's up with that? They were obviously not a masters team.

Ya, it was a young team. Not all juniors but very young. Alot different from the WRCC team who was at Pickering or will be at GWN. I don't know they were called Masters.

Whatever. First you say you would have come in 3rd at the island and then you get royally smoked in Ottawa. Now you say the team is really different than the Pickering team. Newsflash superstar, everyt team that travels generally is a little different because not everyone can go. So all of your competition except for the local teams (which is just Events Alive) were in the same position as you.

You got pwned. Enjoy the re-ranking

the last poster sounds like a bitter G division paddler. One day you will be able to paddle a race under 3 minutes, one day.

The local teams were only Events Alive?

...I guess the other 189 teams were out of area.

Actually, I've probably won more events than 99% of the people on this board. I just hate it when teams show off. I never talked trash to others when I win and still don't. But I have zero tolerance for those who toot their horns at the first whiff of success. These people are my motivation to keep going. As for the other guy, I was referring to teams who beat WRCC in Ottawa last week. Off the top of my head I know Events Alive did since they were the only local team in A division.

A division? WRCC was at the bottom of B division.

Then you forget Cascades, Paddle Demons, and Gung-Ho. All ahead of WRCC in the B.

Good try though.

I dont get it. They were called WRCC Masters, so shouldn't this just be a different entry in the rankings? Why do other have problems with this? Especially without much proof?

Question for PD:

Will you be doing a separate regression for each of the days for Ottawa? I say this because there seemed to be about a 3-4sec difference from one day to the other due to the head wind

Great question. I'll think about it but I believe the correct way to do it would be to treat each day of the festival separately.

PD

June 27 2007, 9:34 AM <--- Sounds like WRCC trying to get their Ottawa results dropped.

Best thing to do is to contact PD and tell him. BUT, I point out that the loud mouth WRCC paddler who broadcasted their attendance to Ottawa and not the island (and also said the one going to Toronto was essentially a "B" team [what a nice guy]) posted on other threads stating the good WRCC would be going to Ottawa rather than placing THIRD at the island.

Hey PD, in light of the great work you have done so far with the rankings what about factoring in a value for each paddler? Some teams change rosters (slightly) and if the change in bodies is accounted for then these ranking would really make sense. Come to think of it, the ranking would be so good we would not have to race anymore, just play with the rosters!
Sun Microsystems has a new computer with a 1.7 petaflop processor that should help you with this.
Thanks for listening.

That was not a masters boat. The WRCC team were all young 'uns.

Re: True Rankings 2007 May 15 2007, 11:57 AM

If you simply average festival times, you wouldn't be accounting for differences in boat speeds, course length, weather conditions, and course depth.

By taking these into account, the only thing you are really not adjusting for (at least to some extent) is lane bias - which tends to wash out as teams do more and more races.

On the west coast, it is bit harder, ignoring lane bias. The main problem would be tide with most coastal festivals. Another is river current mixed with tide, for the Richmond DB Festival. Finally, the best one yet, some festivals in BC have alternating types of boats between heats. For example, Alcan DB Festival heats alternated between modified Geminis and BuKs, with a 5 second different between boats.

I still haven't figured out how PD will address that issue.

boat speeds, course length, weather conditions, rosters, course depth and true dragon boater vs flat water paddler. WHO CARES?

Why do WE continue to diminish a good thing?

WE did the same thing to the Dragon Boat World magazine. A good thing created by a fellow paddler and WE all had something negative to say. The last time I checked this is (was) the only dragon boat magazine.

PD stepped up to the plate and created a system that works. Over the years you all may have noticed that the system is expanding and improving. If you are able to create a better system then create one yourself.

It seems that ever time someone in this sport try something new to support the paddlers WE just rip it apart.

It’s time WE as a community step up and make a difference by supporting the new pioneers of OUR sport instead of constantly pulling them down.

Michael

I concur.

If you have any negative criticism about the rankings, then create a better system. If you can't, then shut the hell up. Everyone has their opinions on how it can be run better, but they are just opinions. I bet they wouldn't even know how to run a regression if their life depended on it.

"On the west coast, it is bit harder, ignoring lane bias. The main problem would be tide with most coastal festivals. Another is river current mixed with tide, for the Richmond DB Festival. Finally, the best one yet, some festivals in BC have alternating types of boats between heats. For example, Alcan DB Festival heats alternated between modified Geminis and BuKs, with a 5 second different between boats.

I still haven't figured out how PD will address that issue."

There is no doubt that things are a little different out West. Tide is a real issue and I really can't think of a direct way of addressing. It may be the case that like lane bias, it will tend to come out in the wash. The only real way to tell would be to put together some numbers and see if they are reasonable. This issue may or may not make the rankings feasible out West.

The alternating boats is not a huge issue as long as you can properly identify which heats have which boats. You would then simply regress the change in times by the same crews from one set of boats to the other. For those who remember I had to do this for the Montreal festival two years ago.

PD

Hi all, what do the following columns in the True rankings mean?

Adj. Chg. & Move??

adj. chg. = the delta from last year's times vs this years
move = the rating on how the team is doing this year compared to last. ie. hhh means > 4sec change, hh means > 2sec change

When is Ottawa being addded??

Bump

PD Ottawa?

Ottawa will be added after Montreal (end of July) and with Lachine. The additional data points gained from Montreal are required to perform the regression mapping these crews to the GTA crowd.

Sorry for the delay, but given the methodology this is the only viable approach.

I have also reviewed the Welland results and they cannot be added at this point due to low sample size (only 3 currently ranked teams attended). This may change as the season moves on in which case it will be added.

PD

Have you thought about adding some sort of legend on the site? You know, for those of us non-statistics people in the audience?

What is i, hh etc?

The i,ii,iii and h,hh,hhh is something I need to fix. Under most versions of microsoft office, these letters will show up as 1,2, or 3 up or down arrows, indicated movement from last year. However, it seems that not all computers read them as such, showing letters instead.

If anyone has any ideas on how to insert arrows that would be more universally recognizable (regardless of what software is being used) please feel free to drop me a line. This is something I would really like to clean up.

truerankings@hotmail.com

Cheers, PD

Ottawa???? What about Ottawa.
please add the results.

read the thread dumbass!!

Ottawa will be added after montreal.

Leçon vous dumbass anonymes. J'ai oublié ainsi embrassez mon Ass. velue. Mais merci du rappel.

To Paddle Demon: please note that the Sudbury race time results are incorrect in some cases on the "2007 Competitive Overall Rankings" page on the Sudbury Dragon Boat Festival website for round 1. For accuracy's sake when times are tabulated please refer to the actual race time results found on the other pages of the site (maybe they will issue a correction). Because of the terrible changing conditions during the 2nd heats which were on average 6-8 seconds slower how are you going to work with the race times? An interesting challenge to say the least.....

Sudbury is up. I will review and try to make sure I have the right times (for simplicity I dropped all races that were reported as 10:40.0). The slower times in the second heat simply figure into the average and are then adjusted.

PD

Why bother with the times from the preliminary races? I assumed that if Sudbury gets included, it will only be the results and times from the final (when conditions were the best, anyway).

Read the methodology. All races are used - not just finals.

But in the case of the conditions and variability in times at Sudbury, should all results be included? That may be the standard approach in the methodology, but I would argue that an exception could be made for Sudbury.

Variable conditions WOULD ONLY affect the rankings to the extent that only a subset of all boats were affected (for example, for the second race of the day it was calm for half the boats, but then windy for the rest).

Usually weather effects are gradual (wind slowly rising) - without large discrepancies between races.

The reason all races are included in the rankings is to help wash out these effects. The more races recorded, the less likely it is for a team to continually get favorable weather for a subset of their races, while the other teams face worse conditions for the same race.

Therefore, Sudbury will be added as per the standard methodology. Honestly, you aren't going to find any big discrepancies as a result of weather as the regression will remove the average weather effects for each set of races.

The Sudbury results are up so feel free to look for any large biases and let me know what you find.

PD

For a competitive Dragon Boat Crew the final is the only race that counts, the only race where every paddler gives an all out effort. So why are the preliminary rounds included in calculating the true rankings? Preliminary rounds are used to advance - the slower you can go and still advance to the next round the better you are as a crew. In my mind it should all be about the time in the final.

Comments?

Bobby Moore = New College

During the development of the rankings there was much discussion as to the number of races. The final decision was to include all races in order to help eliminate condition effects such as lane bias and weather.

Imagine that you include just the final race, but the wind has been on the rise all day. By the final race, all boats have a two second advantage compared to the first final. Here we have now introduced an inaccuracy of 2 seconds (which can be nearly 20 ranking places), plus additional innaccuracy of wash riding (as the slower boats in the heat will get pulled along - a common feature of all finals as the boats are often pretty evenly matched).

If we include all races on the day, this inaccuracy drops significantly - i.e. by the number of races (and other biases introduced should be random and therefore tend to offset each other).

It was also felt that most cometative crews nearly always race at 100% capacity - since we do 2 minute races there is no rational advantage to trying to save up glycogen or "taking it easy" during early races. However, the use of subs may have a more significant impact - but all teams are likely to have this effect.

In a nut shell the rankings reflect the times of a crews average race. Those who can ramp it up for the final will be more dangerous than the ranking predicts, and those who can't will be less so.

Based on my own observation over the last three years, the early races are a very strong prediction of the final - so I wouldn't get too excited about the approach, as it really won't make a significant difference in how you rank.

100% Bobby Moore paddles for New College

I understand the rationale - and I think I speak for (almost) everyone when I say that I really appreciate the time and effort that has obviously been put into this. I think any reasonable person would agree that your system works very well and that no system will be perfect unless every team raced at every event. I agree that the early rounds are a good predictor of the final, but I also think that for the top crews there is certainly less than a 100% effort early on, I'd even venture a guess to say that there could be several seconds difference between round 1 and the final. Perhaps this all comes out in the averaging, perhaps not.

Anyway, keep up the good work.

Is a similar system in place to rank other sports where conditions can have a large effect on time? I'm thinking of Athletics (long jump, 100m...). If not have you ever tried to apply your system to another sport? It would be an interesting statistical adventure that any math geek would love.

hey Fake Bobby Moore, get your own name

Thanks Bobby. Both approaches definetly have their own pro's and con's but all races was the road we choose to head down.

PD's methodology is the best that I can think of.

Hea