Zezo forum:

Post by YourMomSA | 2020-03-19 | 19:32:34

It looks like VR will be running the same set of record attempts in 2020 as 2019. Possibly annually going forward. Since the courses are all known now, would it be possible to open each event in Zezo before they open in VR?

For example, LA-Honolulu opens in 12 days when Spindrift ends. Opening it in Zezo a week from now would allow people to investigate the quality of a "start immediately" approach.

I guess the biggest risks would be if VR changes the course or the boat (or its polars). It will be impossible to notice that kind of thing before the race starts.

I'm not sure how many of the records I'll do this year, though... They get tiresome. They're more about start time than race execution.

38 Comments

Post by zezo | 2020-03-20 | 20:25:43

That's a good idea. Please remind me in few days if I forget to implement it.

Post by Black Hawk | 2020-04-27 | 22:18:17

Post by zezo | 2020-03-25 | 15:53:16

Added back link to the last year's version to the homepage.

Also started the update process to collect some data.
http://zezo.org/honolulu/times.html

Post by YourMomSA | 2020-03-25 | 17:36:41

Thanks! For reference, I believe last year's winning time was 3:13:15:25. (85 hours, 15 minutes). I can't remember who won.

Post by turb | 2020-03-25 | 19:59:20

3:13:15:25 -> correct
I can't remember who won -> is it getting too big for your boots ? =D

Post by zezo | 2020-03-25 | 22:15:08

The player who won reached peak VSR and has only been playing and wining select races since ;-)

Post by SlyStarLeRetour | 2020-04-26 | 20:18:36

You can't remember who won?.. well to give a clue: the same sailor will win this year too... :)

Congrats ! That's awesome two years in a row ! You 're Master !

Post by YourMomSA | 2020-04-23 | 14:41:22

Could probably open Honolulu - Yokohama soon if you want...

But we don't know what boat they'll run it with. I guess you could initially run it for multiple boats and then kill whichever ones wind up being wrong... But maybe that's more hassle than it's worth. This one is less likely to support a "start immediately" strategy anyway, simply because the overall event will last longer. (6 weeks, if I remember correctly)

Post by zezo | 2020-04-23 | 19:09:11

Maybe Ultim and the Maxi. It won't add server load if run consecutively, only delay the results a few minutes.

Post by zezo | 2020-04-24 | 13:16:54

Ultim crew http://zezo.org/yokohama_ultim/chart.pl
Maxi 100 http://zezo.org/yokohama_maxi/chart.pl

The Maxi color palette needs adjustment, but let's collect some data first.

Post by YourMomSA | 2020-04-24 | 16:59:35

Excellent, thanks!

Last year's winning time was 5:20:44:49... (140.75 hours). I don't recall which type of boat.

So unless the forecast changes, there's no need to start immediately. I might go anyway, though, just to collect some cards. This data really helps with those decisions!

Post by zezo | 2020-04-24 | 17:30:18

Should have been Ultim crew according to my records.

The predicted time tables should be ready about 22 minutes after 5/11/17/23 UTC

Post by SlyStarLeRetour | 2020-04-26 | 20:15:06

That's great - it helps - highly appreciated - Thanks Cvetan

Post by Black Hawk | 2020-04-27 | 22:19:23

can someone share a partner code please

Post by YourMomSA | 2020-06-09 | 15:46:02

I guess Transatlantic will open next week if you want to start its routings... I haven't done that one the last couple years... If I remember correctly, I tend to feel they put too high a Category on it, so the price felt like too much for a relatively short sprint.

Post by zezo | 2020-06-09 | 16:12:53

I was thinking about that too. A week in advance should be OK for a 5-6 day race. And the Yokohama start is closed now, so the CPU cycles should be redirected elsewhere ;-)

Post by zezo | 2020-06-09 | 16:39:23

Configured. Running a Maxi routing too, just in case.

Two options added to the home page drop-down list.

Post by BooBill | 2020-06-09 | 18:12:29

Okay, ones a maxi, what's the other one? Last year it was a maxi.

Post by zezo | 2020-06-09 | 18:36:01

Maxi trimaran.

Hmm. I thought it was Ultim Crew, but that was in 2018. Last year was a monohull indeed.

Post by LinusVanpelt | 2020-06-09 | 20:03:28

Thanks from a newbie sharing your informations, so helpfully.
I was asking me for this next record to see in advance the forecast and their stability, especially that for Hon -> Yoko it was necessary to seize one of the first opportunities.

It's not the same boat, as we are in virtual, for all the records of the year, it could be a monotype championship ?

Post by BooBill | 2020-06-09 | 20:41:50

Post by LinusVanpelt | 2020-06-10 | 06:07:00

I understand now why i didn't have the hoped speed the tahiti race ;)
Ok i have some good results but i'm not an experimented sailor with my only 7 races.
I'm seing, reading, trying to understand (not always with success but often) and learning, and i like very much learning ;)
So see you with pleasure in a next race
Thank all !

Post by BooBill | 2020-06-16 | 00:39:46

Switched it. This year it's an Ultim.

Post by zezo | 2020-06-16 | 12:40:49

Ultim Solo. The current predictions are for Ultime Crew (+10% performance)

Post by Fouras_17 | 2020-06-16 | 10:31:00

Hello dashboard is problem ? for my

Post by 7 SEAS | 2020-06-16 | 16:18:39

Thanks Good job Cvetan

Post by LinusVanpelt | 2020-06-16 | 19:30:51

Thanks Cveto,
we had a tool this morning to start, although I would prefer that it give me better news this evening ;)

Post by LinusVanpelt | 2020-06-16 | 20:41:57

Hi Cvetan,
Are the forecast still 10% "opimistic ?
Will you correct them ? when?
or do we stay with forecasts that make you smile, no problem we are knowing the 10% factor now
Thanks for your job

Post by CRISTAU | 2020-06-17 | 03:37:05

Hi Cvetan,
On Atlantic, I notice that my green and red lines are 8% faster than the prevision (black line), is it that you took off the 10% optimistic only for the black line?

Post by zezo | 2020-06-17 | 06:37:31

Ooops. I changed the router but forgot the polars used in the chart.

I'm taking a little break this week in a village house and trying to minimize my screen time.

The 1-week ahead forecast uses the same configuration as the router.

Post by Luis75 | 2020-06-19 | 08:56:21

I believe you deserve it :)

Post by YourMomSA | 2020-06-23 | 22:20:22

A thought... the "predicted times" table gets updated after the final forecast update is posted for each cycle... Which is roughly an hour and 45 minutes after that forecast update cycle begins... and that's fine, but I'm wondering... does it consume a lot of resources? If not, would it be difficult to instead run it every 15 minutes from when the update cycle starts to when it has finished? Add the new column the first time it runs, and then update that column each time. It would stop changing after the final run, which would coincide when the only run is currently executed.

Why? Well... most of the time, this wouldn't have value. But... when your planned start is imminent, it would be really nice to get the latest table before starting instead of having to wait an hour or more to see if Zezo agrees with the prior update or not.

If this is do-able and worth the effort, the next question would be... can the chart somehow show which boxes are fully covered by the new forecast, versus being hybrids of the new forecast and pre-existing? Maybe the boxes show in white if they're hybrids and get their green/red color only if the updated model goes beyond the finish of the box's time? Or use a different font? Or put an orange line under the last box that is covered by the latest update (to be similar to the orange isochrone in the chart)? Or maybe don't post hybrids at all and just post the numbers that are fully covered by the new forecast? Or keep it simple and put an indicator in the header of how much of the GFS has been processed.

No worries if this would be difficult or would consume excessive resources. Just a suggestion.

Post by BooBill | 2020-06-24 | 18:41:49

Making the timeshift functionality available all the time would achieve the same purpose and a few others, wouldn't it?

Thinking of starting a record run in an hour? Timeshift would allow you to plan the same way you do at the start of a fleet race.

Being able to timeshift to a mark rounding in a fleet race and then run a route from that mark to the next would definitely add some pillow time for a lot of people. Groupama was a case in point. Because it was such a sharp turn at the north mark the router couldn't navigate around it to the finish. As it is, anyone in EST had to get up at 1:30am to run a route once they'd cleared the mark. As if the 3am highly technical start wasn't enough late night fun.

Post by YourMomSA | 2020-06-24 | 19:26:15

Good point - The timeshift already achieves the purpose I was describing, if you're up for spending the time to look at each time and check its ETA, although it only gives one estimate per 3 hours rather than per 1 hour. My suggestion just gives the 1-hour granularity all in one place, if it isn't difficult to do. No worries if not.

I was also going to suggest Cvetan could add banner ads to the predicted times page for a bit of harmless added revenue.

Cvetan has answered the question about timeshifts in non-record events a few times before. He has reasons for not doing it. In many cases, you can achieve the desired goal with a little creativity. For the Groupama north mark, if you're scheduled to get there in 3 hours, set a start point north of the north mark that will get to that mark in 3 hours, and then run a routing to the next mark from there. It works fine as long as there's enough open water around the mark in question to be able to set your start point far enough away to match your desired timeshift.

Post by BooBill | 2020-06-24 | 20:31:45

That's a great trick. Hadn't thought of that. Thanks, Mom.

Post by zezo | 2020-07-02 | 12:49:24

Abut the predicted times - you can run individual routing and check the times before they appear in the table. What I need to do is smaller time increments in the beginning (1 hour?)

What you suggest won't be that hard on resources, but the logic is somehow complicated. You don't want to use a forecast that uses partial data, and you don't know the duration and how much data is needed before you make the test run.

Post by YourMomSA | 2020-07-02 | 14:17:34

Thanks for considering it. Here's what I had in mind, but if you don't like the idea, that's ok with me. I need to use my own time zone in the example, so I apologize if it doesn't match up with yours...

-The new wind forecast update starts at 1030. Usually we see the first updates in Zezo sometime between 1033 and 1038. Then of course it builds out further and further, usually finishing its update around 1200. Then the table update runs at 1215 or 1220 or so.

-My proposal is to run the program that updates the table at 1045, 1100, 1115, 1130, 1145, 1200, and 1215.

-You suggest that we wouldn't want to use a partially updated forecast, but my thought is actually exactly the opposite... My idea is that at 1100, it would be helpful to know what the latest partially updated forecast says. But then it would be overwritten with a "more complete partial" 15 minutes later, and so on.

-The important thing with managing the partials would be for the user to be able to see which table entries are fully within the new forecast's horizon, and which are partial. This could be done with color codes or fonts or maybe just with text indicating the end date/time of the latest forecast update. Or, if you prefer to only show those that are fully represented, you could choose to leave the boxes blank white until the finish falls within the latest forecast horizon. I'm not sure which would be better, as you are correct that partial-forecast estimates could be worse than no data at all in some cases.

-To clarify further, I don't mean to request new columns every 15 minutes. The columns would stay the same as we see now. It's just that each time a new column arrives, it would be updated several times before it's considered final.

Anyway... no need to go to a lot of trouble for this idea. Particularly because it is true that one can run those numbers individually anyway. Just an idea that crossed my mind when it was 1135 or so and the table update was still 45 minutes away.

Post by YourMomSA | 2020-07-10 | 16:11:17

Another small request... Clicking on a time from the predicted times table, if it doesn't match up with one of the three-hour-gap times, causes the menu in the routing page to show "NOW" as the starting time when it actually isn't. Could it show "Other" or "Custom" or just a blank, instead of the "NOW" value? This would allow the user to see the difference, and also to change to NOW for real without having to change to something else first.

Or if you prefer, you could add all of the hourly options to that menu, for the first 48 hours, so that this action could show the actual routing time. This would also mostly eliminate my earlier request for the table to be updated sooner, because it would allow people to directly experiment with each of the short-term hourly options directly.