Wednesday, 4 April 2012

Samuel Sevian Statistics

To find a comparison for Micheal de la Maza, I searched for the youngest US National Master, and turned up Samuel Sevian:












http://www.bbc.co.uk/news/world-us-canada-12057445

Here is his USCF Ratings History Graph:

http://main.uschess.org/datapage/ratings_graph.php?memid=13493815

Here are his game statistics:

http://main.uschess.org/datapage/gamestats.php?memid=13493815

To provide a direct comparison with de la Maza, I constructed a graph on the same basis as in my previous article, using the data for 2007, 2008 and 2009.  I omitted the last four games, and games against players without pre-event ratings, to leave 10 batches of 33 games.  The resulting graph is shown below:














The curvature of this graph is not so pronounced as it is on the USCF graph over a longer time period, but is still clear in the 1400 to 2100 range of the de la Maza graph, which was a close fit to a straight line.

Sevian’s rate of progress was a little faster than de la Maza’s claimed rate of progress (he could have deliberately depressed his earlier performances).  Nonetheless, Sevian was a very exceptional small child, whereas de la Maza was an adult.  Blackburne learned chess at 18, and became one of the strongest players of his day, but his success has not been repeated in the modern era.

4 comments:

  1. http://www.infoq.com/articles/brain-scrum-maza

    ReplyDelete
  2. It looks like the same guy. I found a picture on Blunderprone using Google images. I have added it to the previous article.

    ReplyDelete
  3. Your method of tacticstraining is now implemented at CT:

    http://chesstempo.com/chess-forum/messages_from_admin/spaced_repetition_changes-t3835.0.html;msg30443#msg30443

    ReplyDelete
  4. It is good to see that I am having an influence. I doubt whether going back to the start on problems that you fail once is efficient. The SuperMemo approach of reducing the ratio of successive repetition intervals while the problem is causing trouble is probably better.

    A rating calculation based on all players taking the same average time on each problem, and all the problems having the same time spent on them would be a good development!

    ReplyDelete