Forums Latest Members

Speedmaster Production Dates ...With Production Query tool !

  1. ryanbk Jul 29, 2017

    Posts
    214
    Likes
    993
    Great initiative. Thanks.
    Have uploaded extract of my 105.012-66 CB. Movement number 25445... Prediction was november 1967 and actual production date 22. November 1967.
     
    eugeneandresson likes this.
  2. Aria Wibisana Jul 29, 2017

    Posts
    27
    Likes
    284
    It is very useful!... from the 4 of my speedy (two of 321 and 2 of 861) all of them are inline reference results.

    Awesome...
     
    eugeneandresson likes this.
  3. eugeneandresson 'I used a hammer, a chisel, and my fingers' Aug 2, 2017

    Posts
    5,000
    Likes
    14,590
    Some updates.

    Your tool has been live for little more than a month, and as of last week it has been used over 1000 times for production queries (not as cool as aTime4aPint podcasts which hit the 10k mark (congrats!), but alas there is no beer, watch-nerdery or lovely British accent involved). It would appear to be a good time for some stats.

    Top 10 most popular references by query (in % of total) :

    20% : ST145.022-69 Pre Moon
    9.5% : ST145.022-74 (Thanks @Foo2rama ;))
    8.9% : Cal. 321 but Not a Moonwatch
    7.8% : 3590.50
    6.9% : ST105.003-65
    6.8% : ST145.022-71
    6.7% : ST145.012-68
    6.5% : ST145.0022 80s
    5.2% : ST145.012-67
    3.7% : ST145.022-76

    So you folk really dig the 145.022-69's...

    8.9% : Cal. 321 but Not a Moonwatch

    According to the MWO table A, cal 321's with serial numbers that don't fall within those blocks are not from Moonwatches.
    However, it is known that there is a hole in the Omega archives. They have missing info. A lot of folk cant get extracts.

    It is hard to believe that in the small sample of serial numbers that have been entered, about 100 Speedmasters (or 9%) with cal 321's are frankens (perhaps a lot of folks wanted to see what reference Moonwatch their Seamaster was...).
    It is more plausible that the archive hole is bigger than anticipated.
    As such, this 'double check' has been removed. Automated nightly stats will change this evening.
    Humble apologies to those of you who may have got this message when querying your gorgeous Speedy, please try again.

    Birthday Query

    Thanks @BenBagbag ...this is now in. You enter your birth year (e.g. 1969) and birth month (e.g. 2 for Feb) and 2 ranges are returned to you: one range for the month of Feb, and one range for the year of 1969.

    Please note : due to the lack of extracts for ST145.022-76/78's that were produced in 1979, (i.e. where the serial numbers and references transition from 39k to 43k) you may get an incorrect range...it happened to me. If anyone has extracts from this time...you know what to do.

    Pictures

    Got some nice high-res pics of your beloved Speedy (or archive images from NASA etc) and would like to see them on the site? By all means please PM them through.

    Finally


    Please shout (via PM) in case any strange behavior is encountered (and thanks to those of you who already did, much appreciated).
     
    Edited Aug 2, 2017
    M'Bob, gminnj, Pahawi and 7 others like this.
  4. eugeneandresson 'I used a hammer, a chisel, and my fingers' Aug 2, 2017

    Posts
    5,000
    Likes
    14,590
    Ps: A picture of a watch

    IMG_5733.JPG
     
    gminnj and Bienne2998 like this.
  5. wkimmd Aug 2, 2017

    Posts
    518
    Likes
    958
    JMH76 and eugeneandresson like this.
  6. BenBagbag Aug 2, 2017

    Posts
    2,820
    Likes
    8,983
    Great work. And glad to have been the catalyst for a new tool!

    (You can now also get a glimpse at the market size for birth year watches... Maybe time to stock up on some 70's 80's and even 90's watches!)

    Edit: You may also want to consider adding a confidence interval for normal queries and birthday queries. Especially as you get closer to the tails of your sample or in areas with larger gaps or wonky numbering. Something simple like "we are 98% sure that this serial number corresponds to 1982, and 70% sure February..."
     
    Edited Aug 2, 2017
  7. BenBagbag Aug 2, 2017

    Posts
    2,820
    Likes
    8,983
    Just tried the birthday tool. You may want to spell out the format. I tried putting in "april" and it didn't parse. I eventually figured out to put in "4" but perhaps stipulating that it needs to be in numeric format beforehand would make it easier to use. :)
     
    eugeneandresson likes this.
  8. Fattony Aug 3, 2017

    Posts
    327
    Likes
    609
    Morning Eugene
    I forgot to thank you when you told me about this resource last week it really is fantastic and the more it gets used the better it will become personally it will be my first stop in future.
    All the best Tony
     
    JMH76 and eugeneandresson like this.
  9. eugeneandresson 'I used a hammer, a chisel, and my fingers' Aug 3, 2017

    Posts
    5,000
    Likes
    14,590
    Noted.
     
    BenBagbag likes this.
  10. kov Trüffelschwein. Aug 3, 2017

    Posts
    4,113
    Likes
    16,084
    @eugeneandresson, simply change your input fields to dropdown lists with predefined values formatted as you need them and your headache is gone ;)
     
    eugeneandresson likes this.
  11. JimJupiter Aug 3, 2017

    Posts
    1,482
    Likes
    6,478
    Thanks for the great tool! Just uploaded one and a half extract :D

    Nico
     
    JMH76 and eugeneandresson like this.
  12. Togri v. 2.0 Wow! Custom title... cool Aug 3, 2017

    Posts
    2,467
    Likes
    10,920
    This should be a sticky! Awesome job
     
    JMH76 and eugeneandresson like this.
  13. Andy K Dreaming about winning an OFfie one day. Sep 18, 2017

    Posts
    1,819
    Likes
    5,884
    Just want to say your Production Query tool was spot on, once again. It predicted November 1967 for my 105.012-66, and the extract arrived last week and confirmed it: 11/29/67. I uploaded a photo of the Extract to add to the data. Well done! :thumbsup:
     
    eugeneandresson likes this.
  14. LeoneFM Sep 24, 2017

    Posts
    109
    Likes
    692
    Really great job!
    I tried this tool on my 60s, 70s, 80s and 90s Speedies of which I have extracts and..it is really great!
    The 60s-70s predictions are always perfect (I tried 10 predictions beetween 1965 and 1978)!
    The 90s prediction is accurate, it may be different of 3-5 months.
    About 80s: is still difficult to predict the 44-45 MLN range (1980-1981), but again accurate enough from 1987 (48MLN - it may differ of only 6-9 months).
    I uploaded as much extracts as I could, especially in the 80s and 90s.

    @eugeneandresson Thank you very much mate!

    Cheers
    :)
     
    eugeneandresson and Nathan1967 like this.
  15. eugeneandresson 'I used a hammer, a chisel, and my fingers' Oct 3, 2017

    Posts
    5,000
    Likes
    14,590
    @Andy K @LeoneFM great that it worked out for you. Thanks should also go to @WurstEver for his brilliant idea.

    Regarding the 'reduced' accuracy from the beginning of the 80's (which is worse the closer you are to it and gets better the later you move away from it, but still affects things in the 90's) : this is a combination of the step of serial numbers from late 70's to early 80's (i.e. from 39mil to 44mil, if memory serves me correctly -> just returned from holidays, so still in a different world) and the lack of extracts from this period. I think the early 70's also has non-linear solution too (i.e. multiple overlapping references at the same time with different ranges = multiple correct answers)...hence the addition of the MWO data which clears that up some.

    Luckily this should easily be remedied over time and people uploading extracts from the 80s'/90's to improve the model, such that there is a similar resolution of extracts over this period as in the 60's and 70's...at present most Speedfreaks are only into 60's and 70's watches, and thus only order extracts for them...there are still very few extracts from the 80's and 90's. All extracts are welcome. Thanks so much for uploading. Will soon add their data to the model for an updated website...

    Edit : one more thing that could result in the 90's accuracy being less : some extracts with lower SN's have later production dates than others -> will be fixed with more extracts.
     
    Edited Oct 12, 2017
    Andy K, JMH76 and LeoneFM like this.
  16. omegaman Oct 5, 2017

    Posts
    255
    Likes
    376
    Wow, great job!

    Checked with a couple of my Speedmasters, difference was between 1 and 3 month.

    Thanks a lot
     
    JMH76 and eugeneandresson like this.
  17. BenBagbag Nov 8, 2017

    Posts
    2,820
    Likes
    8,983
    Can I bump this up and ask OP if there is anything interesting that has been learnt from this process so far? In the few months since people have had some interesting discussions that I think may be informed by your work. :)
     
    JMH76 and eugeneandresson like this.
  18. eugeneandresson 'I used a hammer, a chisel, and my fingers' Nov 9, 2017

    Posts
    5,000
    Likes
    14,590
    Hi @BenBagbag

    Thank you for your question.

    Not much more has been really learnt other than the statistic which are gathered by people using it over time :
    - the trend of what people are looking for (i.e. over all queries the popularity is determined per reference ... etc)
    - the change of price per reference (SM101) over time.
    - it would appear that quite a lot of 321's (when last I checked it was 10%) fall under 'not a moonwatch' according to the MWO block allocation tables -> either there are a LOT of frankens, or something else funky has happened with the Omega archives (i.e. a theory I heard was that back in the early days Speedy production was outsourced to another company and this either forms a hole due to different record keeping OR they will not issue extracts stating 'not made by Omega' -> but its just that, a theory I heard ... then there is the chance that some are damaged etc.)
    - the tool is a 'well lubed machine' that simply needs to be fed more extracts round the time from '76 to '80 ... but generally 'modern' Speedies are lacking extracts...and the more extracts people submit, the better it will get.

    Regards,
    Eugene

    PS: the first two points will at some point be available on the site as graphics for peruse...
     
    BenBagbag and JMH76 like this.
  19. eugeneandresson 'I used a hammer, a chisel, and my fingers' Aug 27, 2018

    Posts
    5,000
    Likes
    14,590
    Hi Folks

    Just a quickie to say : site is currently down. Humble apologies. The provider has made some massive changes to their virtual images as well as their interface. The website couldn't be updated (deployment failed) in the last months. They were able to reproduce this error, but not fix it, and only workaround was to move to a new virtual host. This was successfully done, but it appears there is a problem with SSL certificates ::facepalm1::

    Will post again when its fully back up. Hopefully today.

    Again, humble apologies for the inconvenience...

    Cheers,
    E
     
  20. kov Trüffelschwein. Aug 27, 2018

    Posts
    4,113
    Likes
    16,084
    :whipped: @eugeneandresson , all for sudden Omega will get an increased number of Extract requests :D
     
    eugeneandresson likes this.