The Canadian Dragon Boat Rankings (CDBR) team is excited and pleased to present: 2009 Rankings

The CDBR team aims to perform quantitative analysis on regattas throughout Canada to provide insight into the relative competitiveness of each race and provide an integrated ranking system for teams in Eastern Canada. We are committed to delivering up-to-date and relevant information about race results in Canada. Although the CDBR system currently only encompasses regattas in Eastern Canada, as we continue to grow the rankings, our goal is to eventually provide coast-to-coast rankings coverage.

We are looking to continually develop and strengthen the system. Your comments, insight and feedback are welcomed and highly appreciated!

In addition we would like to ask for everyone's help to improve the quality of the rankings on a few fronts:

(1) A common problem is for the same team to use slightly different names at different regattas. We tried to catch all of these instances but I'm sure we missed some. If you see your team ranked more than once please contact us.

(2) If your club uses the same name for more than one team please inform us so we can separate their ranking. For example Center Island comes to mind where I believe I saw an ADBC Women's team in the Womens Premier and Womens Sport division. For now I am assuming it was the same team but I question this. Clarification would be greatly appreciated.

(3) We are interested in adding more regattas to the dragon boat rankings. I suspect there are regattas especially in the Ottawa, Montreal region that we can add to the existing rankings. We just require that the additional regatta have a significant number of teams that are already ranked, and preferably ranked fairly highly.

(4) At this point the major time consumer is entering the race data into our model. If there are any volunteers willing to enter race results into a spreadsheet and send them to us we would love to hear from you.

If anyone is curious about the details of our model we have a methodology section. Feel free to read it and offer any feedback.

Finally, we should mention that Power Demon did a fantastic job developing and maintaining the rankings for years. The reason we donated the time and effort to put together a 2009 rankings is only because Power Demon did not have the time to do it himself this year.

Thanks!

Nice job. Thanks for sharing your findings.

What about those 1-hit wonders like Calisse and Level 6? Should they be on the rankings even though we'll never see them again this year?

ADBC womens and Premier womens were relatively the same
the one in sport was the Senior Women I think

Alliance Dragon Boat Club Open had a time of 1:59.09 in the 500 Meter Final at GWN Sport, not 1:59.90 as shown in the rankings.

I've been waiting for Power Demon's rankings for a while now. Great job mofos!

I still hate you.

if i remember it correctly
Alliance Dragon boat had a time of 1:55 or something...

Based off my team's results, I think the listed times are adjusted times.

Why not include the Welland regatta? It is a CWCC qualifier.

Scotia Momentum at Pickering was also West Rouge at Centre Island (7th place finisher).

I like how the website explained the methodology. Good job. Keep up the good work whoever is running the rankings!

Nice job Mofos. Way to pick up the ball for PD. Congrats also on the flashy new website and a resurgent team.

Achievement is limitless when you harness the fantastic powers of ugly men surrounded by attractive women.

Keep it up!

Thanks for putting this together! Awesome job folks! Just a quick comment, OHDBC Shockwave was not present in Montreal, that was another Montreal based team.

Thanks!

Now I've seen it all.....geeky Mofos, who actually do things other than flexing LOL!!

Will the team that everyone loves to hate get a bit of a reprieve from the forum? In the short term yes, long term I hope not :P

Its great you're setting this up but atleast review the results. I mean Piranhas were 4 secs faster than Chaos at the island yet you have Chaos ahead of them...you put the effort in just put a little more in and you'll be ok.

Yes Piranhas were 4 seconds faster than Choas, but Choas also has a really fast Milton time while Piranhas have some slower results (Welland in particular).

More paddling less number crunching.

Yeah the common sense indicator re: Chaos vs. Piranhas throws a wrench into an otherwise very legitimate system. Perhaps its something that should be accounted for in the analysis? Not sure how since I don't have the math background but when we see a team that was beaten pretty soundly head to head rank ahead of another crew this should be cause to look for some improvement in the system

All I can say is wow!!! Ok, I completely dropped the ball on the rankings but I've pretty much had to drop the ball on most things in my life with the arrival of a new little one. In fact this is my first time on the forum since the summer began.

Anways, I'm just incredibly impressed that the Mofos were able to pick this up and make it happen. The community owes them a big thanks. It is amazing what reading a few statistical books between reps can do

A few recommendations:

1. Give each team a number identifier so if the team name changes a bit you can still keep track of them from year to year but update the name (I didn't do this and regret it).

2. In response to this "What about those 1-hit wonders like Calisse and Level 6?" If the team was predominantly made up of other teams already ranked I recommend leaving them out. If they are all new paddlers would just don't show up much include them (i.e. Level 6 which is really Event's Alive).

3. Anytime you need an extra hand feel free to drop me a line

PD

In terms of festivals to consider, Lachine is a must.
http://www.22dragons.com/site/?q=en/node/49

Other than that Sherbrooke or possibly Rawdon are sometimes well attended by Montreal crews.

There is not other big festival in Ottawa but many of the teams do Carleton Place in September.

500m is the basis for the Mofo ranking but Lachine is 200m. I don't remember how you did it PD but if it was a simple extrapolation that's a bit of a disservice since 200m average speeds are always higher than 500m average speeds.

I would urge you not to even bother trying to rank the entire country. Unless you get a large enough cross sampling it's really way to subjective.

Unless many teams race against each other and cross pollinate many regattas you're not going to get it close when including a larger sampling based on a few connections.

Even is FCRCC had come here and another team along with Jetstart had gone to Alcan, the simple fact that it's such a huge distance almost guarantees that the rosters participating are not what you would see locally. You just can't take Jetstart's time at Alcan and normalize it against their times in local regattas and then blow that out to cover teams in an entire region or visa-verse.

If you go to Dragon Boat West, they don't bother trying to rank eastern teams. You won't see Mayfair up there. I'm not complaining, I think this makes sense. If you want to do separate (Central/Eastern) and Western ranking and people can play at comparing times to see where they think they would rank then that's all good. I guess if FCRCC came here and beat everyone then you could put them on top, but realistically to me that would still be a one off and I think you really should only be ranking teams that come out to play at least 2 or 3 times. Otherwise to me they would be the same as an all-star team or something. Not a true club we would be likely to compete against at any given regatta. Events Alive/Madawaska/Whatever and MCC would fall into this category as well from an Eastern Rankings perspective.

It works pretty well for Quebec and Ontario because so many teams face each other on several occasions at different events that while still not perfect, the deviation or error range is much smaller due to the large sample size.

This is also why I think GWN's ranking are a waste of time and a marketing gimmick designed to make some teams feel good about themselves and participate in events. The US teams ranked on their make no sense to me and I doubt they would fair as well as their rankings in head to head matches with many teams they are ranked above. Just take DCH for example. All this BS they say about exact lane depth and course length and conditions blah blah blah. Doesn't work. If you don't have enough multiple head to head matchups, you just can't do it with any degree of reliability. The error rate on the GWN rankings must be like +/- 5-10 seconds 95% of the time.

In any case, that's my two cents worth. Thanks for taking this on and I think you will have plenty to keep you busy in this region without getting two ambitious.

Awesome feedback, thanks everyone. This kind of feedback will make the rankings more accurate which is what we all want. We also have some emails pointing out duplicates and different teams with the same names. After work today we will make these corrections and produce a slightly better ranking. We will also look into the Piranhas & Chaos rankings to see if we can uncover a problem.


Power Demon, I like your suggestion regarding one off teams. So teams like LEVEL 6 will stay, but hybrid teams like "ON S'EN CALISSE", "LANE 6 7 8 9 AND 10" will be removed. We considered adding a team ID to handle changing team names but I thought it wouldn't be useful. It's great to hear from your experience.


I just want to raise a point regarding why the times in the rankings do not match the exact times recorded at a regatta. The overall times and individual regatta times listed in the rankings are adjusted times. The model does include the Alliance 1:59.09 time as input but this time gets adjusted.

The reason the model needs to adjust the times is because not all race courses/conditions are equal. For example if we didn't adjust the times at Montreal then the Philadelphia mixed team would have an average time of 2:07.8 putting them close to 10 seconds behind a team like Mayfair from GWN Sport regatta. A sound statistical framework is used to determine the adjustment. The theory behind how the model determines what adjustment to use is the heart of the rankings system and is fully explained with illustrations here .

For people who do not care about the details then roughly speaking if your team shows up to a generally slow race course (like the Philly team did) then the model will adjust your time to make it faster. The opposite occurs if you go to a fast race course. All adjustments make your time roughly on par with Center Island.

In addition to this we include all races including heats and semi-finals which may account for some of the observations made above regarding timing discrepancies.

With that said if you see something in the rankings that you know intuitively just seems wrong then we are interested to hear about it so we can use it as a starting point to look for problems.

Great work! Must've taken some time.

Two suggestions:
1) Would it be possible to add a filter by region? (GTA based, Montreal based, etc)
2) Similar to what other people have said, can you make an option to get rid of teams that have only participated in one regatta? (One hit wonders?)

Really awesome work. You put some well served time into the "methodology" write up.

I see a slippery slope trying to morph 200m and 2000m into 500m times, but they are so common now as to literally be a part of a lot of events. IE. There are more and more sport regatta events.

I would think that other than adding pages to the site and a few lines of code to your program that you should be able to include 200m and 2000m times. Maybe the deviation adjustment needs a little tweak for the shorter and longer distances compared to what is done for 500m, but that would only need to be figured out one... And not by me, so all good.

Then you would have a lot of groupings and would probably want a second layer of drill down from the main page. IE. Choosing Mixed > gets you (Overall /200 / 500 / 2000) choices. The "Overall" tab would include each mixed teams average time per distance and that times overall ranking in each distance along with a rolled up average ranking over the three distances.

Just an example:

Team.......Overall/Avg Ranking.....200m Ranking/Avg Time.....500m Ranking/Avg Time.....2000m Ranking/Avg Time
Team A...........1/1................1/44.5.....................1/1:58.6......................1/9:03
Team w...........2/2.33.............3/45.9.....................2/2:00.5......................2/9:08
Team x...........3/3.33.............2/45.2.....................3/2:01.2......................5/9:21
Team y...........4/4................4/46.8.....................5/2:03.2......................3/9:12
Team Z...........5/4.33.............5/47.0.....................4/2:02.2......................4/9:18

This type of ranking would give an idea how teams would fair against each other at a Sport Regatta or say, Nationals or Club Crews. You could even call it Sport or Club Crew Rankings summary or something.

I thought about maybe doubling the weighting for 500m ranking distance, but that isn't the way it works in real life at sport regattas so forget that. I would suggest that if two teams have the same combined ranking total for the three distances, that the team with the higher 500m distance would get the higher rank.

So for (200m / 500m / 2000m), a team with rankings of (2 / 1 / 3) = (avg rank of 2) would rank higher than a team with (3 / 2 / 1) = (avg rank of 2) because the first team had the higher 500m rank.

Only teams that have at least one 200m and 2000m would be included in this overall/sport/Club Crew ranking (obviously). I would be tempted to say that you needed at least two times at each distance to qualify, but I'm not sure enough teams would are doing more than one until more events offer them.

Maybe the "sports/overall/club crew" tab should be a separate link from the main tab that just links directly to this master ranking and the other classes stay the way they are organized now.

Anyway, there are a couple more of my pennies.

I actually like the idea of seeing where any of the one-off hybrid teams would rank because the potential is there that they re-form and kick some ass.

But I think if they can't be dynamically shown and hidden, then it's best to leave them off so they don't dirty-up the rankings. So that you can use this ranking to categorize true top-5, top-10, top-20, etc teams.

One inherent flaw I find is basing the rankings on all times including heats and not just finals, take the Montreal results for instance. Verdun and Philly are "equal" when Philly beat Verdun head to head twice, Verdun is ahead of Hydros who came in second, BMO ahead of Philly seniors who beat them head to head in the final, etc. The final is generally where you really see what a team has. No big deal just my own bias, thanks for the rankings.

Doesn't make sense to only include finals. If you only use finals then you're going to make a lot more problems!

wow terrible ranking system...

wow amazing ranking system...

wow. a ranking system.

We managed to find a little bit of time this evening to work on the rankings.

Woodstock
Woodstock results have been added (Mayfair Warriors were fast!)

Team name changes
We received a number of team name update requests. This resulted in eliminating duplicates as well as separating different teams with the same name. As a side effect it very slightly changed the adjustment used at each regatta.

One hit wonders
We removed teams that we know were combinations of existing teams and are not likely to appear in future regattas. "Non-merged" teams such as Level 6 remain in the rankings.

Rankings quality
We investigated the Piranhas and Team Chaos ranking (sorry for dragging your names on to this forum). Both are excellent teams but we expect the Piranhas to be ranked higher. These teams met head to head at the GWN Sport regatta and the ranking for that regatta is reasonable with the Piranhas about 4 seconds faster. The problem results due to the Piranha's relatively slow time at Welland, and Chaos's fast time at Milton.

The Piranhas won the entire Welland regatta but the level of competition was not high enough to push the Piranhas and the rankings essentially punished the Piranhas for not winning by a greater margin. We consider this a flaw with the rankings since there was no real need or motivation for the Piranhas to go any faster than they did. To fix this problem every regatta now has a competitive multiplier. A regatta that has very few competitive teams will effect a team's ranking by less than a competitive regatta. The new rankings have the Piranhas ahead of Chaos. There are other rankings order changes due to this change as well but we believe overall the new rankings are closer to being on par with our intuition. Comments on the competitive multiplier are given below.

Competitive Multiplier
The competitive multiplier uses a formula but that formula is one that we subjectively chose. We want to have as little subjectivity as possible but we do think this competitive multiplier is necessary. The competitive multiplier is determined as follows:

The attendance of the any of the top 50 teams award competitive points to a regatta. Team #1 gives 50^2 points. Team #2 gives 49^2 points, Team #3 gives 47^2 points, etc. The sum of competitive points for a regatta is then divided by the competitive multiplier for Center Island. This number is shown next to the regatta name in the rankings. The multiplier is clamped between 0.1 and 1.5.

So if a team attended Center Island which has a competitive multiplier of 1.0 and Welland which has a competitive multiplier of 0.1 then the ranking time will no longer be a straight average. Now it will be ((1/1.1*CenterIsland) + (0.1/1.1*Welland)).

Just curious as to how Shaka Open has a faster time than ADBC Premier Open at the Island when they placed 4th in Open heats all weekend.

Well done Mofos. I'm impressed it's starting to make sense.

Very surprised the Mayfair Warriors are now a mid-top 10 team?

THANKS FOR STEPPING UP TO THE PLATE MOFOS!

The Warriors gave Big Fish a run for their money at GWN3000.

We haven't seen much of them this year but they could steal the show at the GWN Challenge this year.

Sure if they fill it with Preds since they won't be there.

Otherwise Warriors don't make the Final.

Firstly the actual times make no sense. Either the course was short or there were some crazy tailwinds (I'm looking at Dave's posted results here and not the adjusted ones yet). This may have borked the algorithm a bit.

If your formula applied to Woodstock puts the Warriors into 8th spot, then you have some issues to fix there. Not to pick on the Warriors which I think is a fine up and coming team, but I don't think they have ever even finished in the top 10 in a competitive regatta to date.

With a couple exceptions, it looks like your adjusted times for Woodstock are about 1.5-2.5 seconds faster than each teams average time at other events. Of the top 8 teams in the rankings that were at Woodstock and other events, 7 of them posted faster times like this. (7 of 8 is a clear trend)

Maybe everyone did get that much faster, but I somehow doubt it. If you were to add 2-2.5 seconds to the Warriors time that would put them a lot closer to where you would logically put them in the rankings and even then many of the teams below them would have a good argument for them not being that high. Again, sorry to pick on Warriors.

I don't think that head to head they beat any of the teams within 8 or more spots below them where they are currently ranked.

I guess we shall see at GWN.

10:22am poster is hilarious.

You THINK the warriors shouldn't be that high in the ranks? Thinking something isn't enough merit for them not to be ranked that high. And no, I'm not on the Warriors crew. The data looks pretty accurate to me. The rest of the crews in the finals clocked times similar to what was expected and the fact that the Warriors did beat everyone by a good second or 2, makes sense they're 8th.

If you have solid data as too why they shouldn't be, then please do share. But that team was pretty damn fast and isn't the same Warriors team that we've been seeing in B/C divisions.

Congrats to Mayfair for adding another trophy to their case.

I think it's okay to question the results of a "model" when it spits out something that is counter-intuitive.

The problem here is that well known local team with no other results in the table has vaulted into the Top 10 based on a smallish regatta like Woodstock.

Based on this, I think it's perfectly okay to question the rankings model. I don't say throw the results out, but I think you have to ask if this result makes sense, does it "fit".

Maybe the result is perfect just the way it is, but I don't see any harm in questioning the result and not trusting to the numbers implicitly. A model is just a model after all.

To be honest, I think the model is fine. I do believe that the Warriors have gotten significantly better and are no longer the B/C team of yesteryears.

Guys, remember the other "Power Rankings" thread, where everyone gave their two cents about who the top teams in Canada were? Many people had Jetstart and Chaos in their top 12 and some had them just outside the magic dozen.

At Woodstock, the Warriors beat Jetstart head-to-head by a full 1.4s. I don't see why anyone would consider it counter-intuitive for the Warriors to be considered top 10 amongst central and eastern teams.

P.S. I'm not quite sure why Chaos is ranked higher than Jetstart, though. As far as I know, Jetstart has beaten Chaos consistently by about the same margin each time in the finals. My guess is that the qualifiers are giving Chaos the edge.

My memory failed me. Upon review, no one had Jetstart in their Canadian top 12, but many, like I said, had them just outside top 12.

Regardless, still puts Warriors quite high amongst a list of exclusively eastern teams.

"10:22am poster is hilarious.

You THINK the warriors shouldn't be that high in the ranks? Thinking something isn't enough merit for them not to be ranked that high. And no, I'm not on the Warriors crew. The data looks pretty accurate to me. The rest of the crews in the finals clocked times similar to what was expected and the fact that the Warriors did beat everyone by a good second or 2, makes sense they're 8th.

If you have solid data as too why they shouldn't be, then please do share. But that team was pretty damn fast and isn't the same Warriors team that we've been seeing in B/C divisions.

Congrats to Mayfair for adding another trophy to their case."


I provided the information you are looking for in my initial posting. I'm not sure why that is so hard to understand.

Bases on teams that raced at multiple venues on the ranking, 7 of the top 8 teams posted disproportionally faster times at Woodstock than they have at other events. This is even off of the ADJUSTED times which I assure you were adjusted to be slower than the actual 2:01-2:02's etc.

What I am saying is statistically valid and is the basis for "What I think". The times posted in the finals are disproportionally fast from the teams other results even after they are adjusted.

If you really think Warriors would be in the thick of things with the teams around them in the current standings then that is your estimation. I am saying that the number don't make sense and there appears to be a problem. I guess we will see who is right and who is wrong at GWN.

I predict that the Warriors do not make the final unless they are stacked with Preds. This isn't because I don't like the Warriors, wish them any ill will or just feel like attacking them. This is based on the numbers. Until they go head to head with "comparable" teams then I guess it's just my interpretation of the numbers versus yours.

Based on what you would expect given the woodstock numbers the Warriors would have been either in or on the edge of the A Championship division at the GWN Sport Regatta. I just don't buy it. They are currently ranked ahead of far to many teams they have never beaten head to head.

Anyway, I don't fault the MOFOS efforts here. There are just so many variables to take into account. Despite all the efforts and what looks to be on paper a solid methodology the current list has more than a couple teams above other teams that they have been beaten head to head by this year.

Maybe more of a weighted ladder system would be appropriate. With that your highest possible ranking would be one above the highest team you have beaten in a given regatta with the most recent events counting for more than the one before and so on. This type of thing would also eliminate variation in timing which at some events are at best rough estimations remove time penalties and bonus etc.

The Warriors are legit. They placed just behind Big Fish in the short course part of the OHDBC Challenge. Unremarkable at the time because we didnt know how fast Big Fish had become and unnoticed because they got buried by an incident and penalty in the 2k. But the rankings admin got a first-hand look in Woodstock. In adjacent lanes for the A final Warriors beat Mofos by more than 3 seconds. That kind of margin looks pretty strong when you consider how well Mofos did at the sport regatta.

"Guys, remember the other "Power Rankings" thread, where everyone gave their two cents about who the top teams in Canada were? Many people had Jetstart and Chaos in their top 12 and some had them just outside the magic dozen.
At Woodstock, the Warriors beat Jetstart head-to-head by a full 1.4s. I don't see why anyone would consider it counter-intuitive for the Warriors to be considered top 10 amongst central and eastern teams."

But didn't JetStart get beat by a larger margin in Montreal by a number of teams that the Warriors just leaped over in the rankings?

Here I will even give you another little quirk of the Woodstock event. This one would argue that the Warriors should actually be even higher up the ranking than they now appear.

Rounded to 10's of a second 0.1's.

The Warriors actually final time was 2:01.4. and their adjusted time was 2:03.6 or an adjustment of 2.2s

The MOFOS actual time was 2:04.8 and their adjusted time was 2:06.2 or an adjustment of 1.4s

Chaos actual time was 2:05.1 and their adjusted time was 2:06.6 or an adjustment of 1.5s.

I'm not sure why the Warriors would get such a higher adjustment than the other two (especially since their adjustment has to be based of other teams adjustment since this is their only event).

Based on the adjustment that Chaos and the MOFOS got, I would think that the Warriors SHOULD have only gotten about a 1.4-1.5s adjustment. So based on that their "TRUE" adjusted time should have been more like 2:02.8-2:02.9 and not 2:03.6.

That would put them in actually in 6th place in the ranking ahead of Hydros and Verdun as well.

Talk amongst yourselves.

a few questions you have to ask yourselves before you look at the warriors as a lesser crew than they really are:

1. where did all the ex-imperial dragons go? did they just retire or go to csdc? or did they end up on another crew
2. the warriors are not a dbc registered team. is it possible that they can always count on the predators for a few extra bodies - let alone any crew if they are not racing as a dbc registered entity.
3. are the people who were on the team last year improved under the watch of a proven and capable coach?

i think 6-7 yrs ago all of jetstart, chaos, hydrophobic were lesser players in the sport. through training, recruiting and building up experience as a crew these teams got better.

the warriors are doing the same. are they not the evolution of the killer b's from the early 2000's? a crew that's been in existence for 10 years definitely has the potential to succeed and break through when the desire and personnel is there to do so.

Again, please don't get me wrong. I have absolutely nothing against the Warrior and I absolutely DO think that they are an improved and still improving team. I actually feel that they will continue to improve and be a force to be reckoned with in A division. Just maybe not quite yet.

1. I know there are ex Imps on the team
2. I don't think Preds participation is the cause for their improvements. I think they are a good crew that works hard and are getting faster.
3. I would say that this is part of their improvements. Personal, experience, hard work and good coaching is clearly why they are getting faster.

Unfortunately, that does not quite yet have them sitting where they are now in the ranking without having beaten any of the surrounding teams head to head yet. Also, based on the adjustment other A finalists got in Woodstock they should actually be even higher. My comments and observations are "Ranking input related" and not intended to be a cut at the Warriors.

Maybe only teams with two events should be included to help offset some of this stuff. If Warriors win GWN without stacking (which I think they would almost have to do based on their ranking and who will be there) then I will be the first to apologize.

Let's not get hung up on just the Woodstock results. The regatta competitive factor for Wookstock is only 0.32. So when GWN is all done, the woodstock results will only account for about 25% of their actual time (assuming GWN has a factor of 1). But I still believe that they'll be close to the 2:03 average time even after GWN. Just my two cents.

For the people who were at Woodstock, we all saw there were a few Pred's in the boat. But just a few. That does not translate into a big jump. For the rest of the Platinum Div teams, we all know they raced in top gear. What does that mean? That means Warriors were legit. Stop crying about another team at the top. Can't wait to see what will happen at GWN Challenge!

Lots of talented recruits on this year's version of the Warriors, but also some overlap with the Predators. They'll make the A final at GWN. Unlikely to win, but a good bet to medal.

Ahh so Predators were in the boat. I guess this is where they may fit for the remainder of the year if they continue with Predators paddling with them.

I thought they were good enough without them, but whatever.

I guess this is just another of those impossible to factor in scenarios. How many Predators will be paddling with them at GWN will possible tell the real tale as to how they will do then I guess.

Congrats to the Warriors on their victory. The taste of beer is always sweeter when sipped from a Championship trophy! Well deserved, I'm sure. You always seem to be working very hard when we see you on the water.

I'm really enjoying reading many confused posts - brings back a lot of memories:
http://www.network54.com/Forum/135730/thread/1122344365/last-1125852164/True+Rankings

Just wanted to respond about incorporating 200m races. Really easy, just run the regression as you would in your example of Pickering to Toronto island (although in this case you would essentially be doing Lachine vs. Montreal). This will automatically convert 200m's to 500m's. It's the same concept as adjusting a festival that is 490m to a 500m one only the multipler will be much greater since it is 200m to 500m.

You can use this technique in festival if there are a variety of distances, eg. 2000m, 500m, and 200m. Just regress 2000m vs. 500m, and 200m vs. 500m to convert.

PD

Were the Preds paddling with the Warriors at Woodstock wearing PFDs? Does anyone have pictures?

PFD's were manditory for everyone.

A Fish --> "The taste of beer is always sweeter when sipped from a Championship trophy!"

How would you know? LOL...I'm sorry. I'll stop.

Well, truthfully it was a plaque for "Participant" that we spilled our beer on and licked it off, but to us, it was glorious!

In all seriousness we won Stouffville a few years back. Not the biggest regatta, but a win is a win. Not sure there even was a trophy!

I'd take the Woodstock results with some suspicion [slightly].
The 2000m result when compared to GWN Sports Regatta, which has deeper water,
showed a significant improvement from Chaos, Mofos and Jetstart.

Well to be fair, the times for the "500m" were adjusted to be slower to better match the Island times.

I just don't think they were adjusted by enough to be realistic. Based on the average improvement by almost all of the top teams this adjustment should have been more like 3-4 seconds rather then 1.5-2.5 or so.

That said, it appears that the Warriors had more than one or two Predators in the boat, so that may also partly explain the improvement. If this is what the team will look like for the remainder of the year, then they could very well be ranked accurately and will give the field a good run for their money at GWN.

An already very solid and improving Warriors team plus a sizable Predator upgrade should be a strong contender.

Just so you know, the 2000m race in Toronto was more than 2000m, while the Woodstock distance was accurate (470m straits plus turns). You can't compare these times.

If the MOFOs version of the rankings uses the same regression as Power Demon, the distance of the race won't really matter too much once everything is measured in terms of a 500m race.

For better or worse, parachuting in 5 or 6 different bodies in the boat will make a difference in any team's performance but to give credit where credit is due the Warriors had a good base of paddlers to begin with and have been training hard all season.

If the Warriors did have a significant number of non-Warriors in the boat, wouldn't that be cause for a non-inclusion based on the combined team rule? It's unlikely they will race again as the same group again since at GWN both the Warriors and the Predators will be there. I recommend dropping this result.

Agreed.

I think we are just trying figure out how accurate the adjusted times for Woodstock are and secondly to figure out where the Warriors should fit in the rankings.

I think the Woodstock adjusted times are still fast, but maybe not by as much as originally thought.

With 5-6 Predators in their lineup then the Warriors very well could be in the right ballpark in the rankings. Are they racing again anywhere prior to GWN?

It's hard to say that Woodstock times are 'fast' due to the Warrior's results. If you look at the other teams, their times were very similar to their averages throughout the season. Most team have improved their times by a bit but that's understandable since most teams will have a 3-4 months of training under their belt by now.

My money is that the Warriors will have a strong showing at GWN and make a lot of people on this forum eat their words.

That's not true. Read the above posts. Almost all the top teams participating show about a 2 second time improvement over their averages even after they were adjusted slower.

Call me a cynic, but most teams who make the leap up usually don't keep it. The Warriors have had obvious help and I doubt they'll repeat their performance at GWN. The only way they finish high is if no one shows up. That might happen since the Nationals are only a few weeks prior.

This is a slippery slope. As soon as you start removing a teams particular result off the rankings because you believe/know them to be of a different composition, you open the doors for many other complaints about other teams. For example, at Woodstock:
- Jetstart had atleast one hydro on the boat.
- Chaos had atleast one predator on the boat.

This sort of policing has had its opportunity on a different scale of competition too: the 2PP rule for example.

I am sure there are many other teams that needed last-minute-fillers and not just in this regatta. In addition I know of teams that change their name for a regatta if they had enough foresight to know they wouldn't have a full boat and would be using guest paddlers. (Mofos are a good example).

Perhaps for now we should leave it to the discretion of the team if they believe their results should be included or not. At least until there are better ideas out there to deal with this.

Don't be so sure. I heard that the Preds are not doing GWN this year, so the Preds paddling with the Warriors are there to stay. That will make them dangerous at GWN.

Let's take a look at the top 8 teams at Wookstock (excluding Mayfair)

Team / Woodstock Time / Average Time / delta
Choas / 2:06.7 / 2:07.1 / -0.4
MOFOs / 2:06.3 / 2:07.3 / -1.0
Jetstart / 2:05.5 / 2:07.4 / -1.9
Shockwave / 2:10.9 / 2:10.6 / +0.3
Vic Crew / 2:09.5 / 2:11.0 / -1.5
Golden Plate / 2:10.9 / 2:11.3 / -0.4
Ronin / 2:12.0 / 2:12.1 / -0.1
CN Aquatrain / 2:11.2 / 2:13.9 / -2.7

Average difference was about 1 second. But again, you have to account for the fact that teams naturally get faster as the season progresses so a 1 second improvement isn't too far fetched.

Remove the faster Woodstock time from the average time (Woodstock brings the overall average down and reduce the difference) and do the math again.

Then also consider that these times are already "adjusted" slower and you can easily see that Woodstock was abnormally fast.

Also note that not all these teams show a steady time improvement race over race since their previous race may have been slower then the one(s) preceding them.

You guys are right! I looked at the adjustment for Woodstock and it is being 1-2 seconds too generous to the top teams. Here is the adjustment it was using which is clearly not the best fit line:





I found out that I had (basically) a rounding error in my program that got worse and worse the more regattas that we added. (I carefully checked the first couple regatta adjustments which looked good but didn't scrutinize the last few adjustments like I should have). Now with the bug solved the adjustment looks great:






Thanks a lot for noticing this - you're sharp to see a mistake of just 1-2 seconds. Special thanks to the person who noticed all of the top teams were adjusted well below their average which indicates a bad adjustment. The Warriors are still flying but my rounding bug did put them a little too high in the rankings. The new fixed rankings are on-line.

RANKINGS LINK



There are a lot of good suggestions here that gave me an idea that I think people are going to like. I'm a little pressed for time at the moment but I will be revisiting this next week.

Wow. I am truly impressed at your dilligence. Well done!

Nice job. Looks better. Hammers seem a few spots too high. Jetstart seems a bit low. But generally speaking this looks good.

With all due respect to the work that has been put into the rankings, I believe that the Chiros are way too low, irrespective of how many races they may have done. Based on how they have raced in the past, they have been in the top 5 for years!

Did they collapse? When are they racing next?

The Chiro's situation is another argument for being able to filter out one offs. They also raced at Perry Sound (was that it?), but that one isn't on here and I have no idea of the times.

I am sure that with additional events that their times would be coming down and their ranking would be higher. Who knows who went to the away regatta in Ottawa for them. They could have been missing some folks etc.

That said, there are rumors that this is a rather large transition year for them. I would still expect them to sit higher after more races are in.

Re: August 12 2009, 12:26 PM

I think you're drinking some special sauce. Shockwave almost beat those guys at the GWN Sport Regatta. That alone indicates that these guys are ranked too HIGH. From what I hear Chaos pulled out of the Nationals because they didn't believe that had a chance there. These rankings are always based off of the present year, historical performance has no precedence here.

@ 1:03PM

Dude, WTF?! The guy was talking about the CHIROS as in SCC Chiros from Sudbury, not CHAOS. You should tweak those reading goggles of yours, cuz they are failing you, as in uberFAIL.

Above post

HUH? Are we talking about Chaos or Chiros here?!

Epic Failed. I suppose staring at all these spreadsheets has killed my eyes.

Chiros are awesome, but need to see more of them.

Damnit, it figures when I didn't reference what post I was responding too, someone would post in between. But yeah basically what the poster @ 1:23pm said.

I think poster @ 1:03pm needs to lay off that "special sauce"

1:03pm.

That's OK, we forgive you for spilling some juicy news. Didn't know about Chaos pulling out.

Thanks to your feedback we've been able to work out most of the kinks in our implementation of the regression based dragon boat rankings. Now we are pleased to announce a major new feature recently added to the rankings website

(1) Head to Head Rankings
I believe the regression based system is the best consolidated rankings of dragon boat teams but with so many factors involved in dragon boat racing there will always be a small amount of "noise" in the rankings no matter what we do. When two teams are close in the rankings the best comparison is to look at head to head races. If no head to head races took place the next best bet it to look at common opponents. This is exactly what we provide in the new head to head comparison tool

For example, let's take two teams, Mojos, and GSK Firedragons. The tool reports that both teams traded victories over each other but Mojos come out ahead 2-1 in 500m pieces, and 1-0 in 2000m pieces. It then goes on to analyze common opponents. We can see that The Blades and The Saints both hold victories over GSK and both lost to the Mojos thus giving a slight nod to the Mojos. You can perform a similar analysis with any two teams.



(2) Race Archive
Curious about a team you have an upcoming race with that you haven't heard of? Check the on-line listing of all races we have on record for any dragon boat team with the new Race Archive.

Check your own team's history and make sure we are using the proper data!



(3) Volunteers for data entry
I hope some people find these features useful. The more race data we have the more valuable the website becomes. We would love to hear from any volunteers willing to send us formatted race data. We just need the race data to be in an excel file or text file in a certain format. We can provide details.


(4) Stacked Teams
Regarding the discussion on stacked teams earlier in this thread. I will agree with Roger Wilco (Space Quest!!!) that it's too subjective to start removing teams we think may be stacked. Plus even if a team does stack their boat who is to say they won't stack it again in which case the rankings should include the stacked teams result.

I officially love the mofos.

Oh no...the team we love to hate is gulp...doing something positive in the db community!

Everyone run, repent your sins, the apocalypse is here!!

Pretty cool feature! I'll send you and email if you need extra help.

you got the time wrong for the Mayfair Warriors in Woodstock. They got a 2:01 and change for the final.

adjusted?

Oakville is a short course...not quite 500m

Thanks for the Head to Head Comparison. That's a really interesting tool. It does more than I expected it to. Well done.

Great work, love it. Not to be picky but is it possible to just compare two teams side by side. The head to head comparison is great but is there a way to compare two random teams.

> you got the time wrong for the Mayfair Warriors in Woodstock. They got a 2:01 and change for the final.

Mayfair Warriors have a 2:01.4 recorded for the Woodstock final according to the archive. The rankings show the adjusted times not actual times so you won't see 2:01 there.

Anon 10:43, if I understand what you are asking it sounds like this:http://www.mofosdragonboat.com/Rankings/head2head.php

Dang ... that head to head comparison thing is schweet. Great work!

Head to head comparison is cool! I'm impressed Mofos

espeically when im loosing every comparison... SIGH*...

MJC,

what i was thinking, right now you are showing the times of only teams that have raced against each other which of course is the head 2 head results. what i was thinking of is comparing the times of two teams that have never raced each other, for example lets pick Verdun vs Mofos, obviously you two have not raced each other (this year at least) but you can compare the times of the two teams still. hope that makes sense.

If I understand the reasoning, comparing two times from different heats is deceiving because conditions change from race to race. For example, times in Pickering this year changed dramatically from the morning races to the afternoon due the wind picking up and slowing down.

Also, I'm sure everyone that paddled in the Island this year would agree that times on Saturday were all over the place because of the horrible conditions that the morning teams had to endure (strong head winds, torrential rain, etc.) compared to the relatively calmer afternoon weather.

As a result, if two teams didn't race head-to-head, it is pretty much useless to compare their times (even if they appeared in the same regatta).

I like what MJC currently has implemented for the head to head. What you're describing can be easily accomplished by looking for both teams in the rankings and then comparing their times.

What MJC has created is similar to what I, and I'm sure many others, have been doing manually. When you want to find out how you stack against another team, you first check the record of when you two raced each other. But if that record is lacking, or is too old to be reliable, then you want to find out if you can beat teams that can beat your rival.

If anything, my only suggestion is to add a tool that compares two teams by using a third as a baseline. For instance, A hasn't raced B in a long time. But both A and B have recently raced C in different races. Then it would be nice to see how far ahead or behind A is of B by using C. And if this system can generate a list of possible C teams, that would be awesome. (This might be a lot of work or may be in low demand, so feel free to ignore.)

I was referring to 10:48 in the first paragraph.

We added the NDC regatta and the Oakville regatta to the rankings. We also found and fixed many duplicate name entries which should improve the adjustments.

Has "U23 NATIONAL TEAM" raced under a different name previously? They are the highest new entry to the rankings with SRP not far behind. Stonehookers made a fast debut on the women's ranking.

I wonder if SRP should really be a new team or just combined with previously listed Piranhas.

I don't think the Komodo dragons in Oakville were the usual Komodo dragons team. I think it was just a local community team.

"If anything, my only suggestion is to add a tool that compares two teams by using a third as a baseline. For instance, A hasn't raced B in a long time. But both A and B have recently raced C in different races. Then it would be nice to see how far ahead or behind A is of B by using C. And if this system can generate a list of possible C teams, that would be awesome."

Given the regression methodology, the above is essentially what the rankings do in order to rank teams. EXCEPT, it does it in a very advanced way - for many teams across many festivals and not just one example.

Given that teams don't always show up at each regatta with the same roster, this should end up being a better indicator than the above example. Beating someone once head-to-head at one festival does not necessarily mean you'll get them next time, particularly if your team or their team roster is different. With so many teams ranked so close, and a good paddler having the ability to make at least a second difference, team performance will vary.

The science and math is great, but too many people here place too much importance on the rankings.


They had other paddlers on the boat, they didn't have all their people, blah, blah. Every crew has issues. Show up and race and let the chips fall where they may. If rankings are that important then post your roster up here for all toe scrutinize.

"The science and math is great, but too many people here place too much importance on the rankings."

Personally I think the ranking are great. Placing in a division in a festival has a lot more to do with who shows up than your performance. We've gone to a festival as a much weaker team and placed much higher than when we were faster and good competition showed up.

However the rankings give you a good indication if the hard work your crew has put in is paying off. I find them much more interesting than placing at some festival.

The Rankings have been updated to include the nationals.

Next, we need to incorporate other distances besides 500m as Power Demon described earlier.

Its about time....j/k

Keep up the good work, I guess those matlab sessions must've worked lol