I think the example of 20 ranks going from Jonidan 34 to Jonidan 14 is wrong. Since the count is half-ranks, a 20 rank increase should be to Jonidan 24 (or maybe 23w, if you started at 34e and round 0.4 fractional part to 0.5).
I wonder if excluding the Makushia joi, which has special rules for getting promoted to Juryo, might improve the predictions for Makushita as a whole. If you started the sample from the 81th ranked rikishi (Ms6e) instead of the 71st (Ms1e), the numbers might line up better.
I'm in awe of your ability to digest and interpret all this data. Very interesting stuff!
Thank you very much! And glad to have you on board as a reader.
I think the example of 20 ranks going from Jonidan 34 to Jonidan 14 is wrong. Since the count is half-ranks, a 20 rank increase should be to Jonidan 24 (or maybe 23w, if you started at 34e and round 0.4 fractional part to 0.5).
I wonder if excluding the Makushia joi, which has special rules for getting promoted to Juryo, might improve the predictions for Makushita as a whole. If you started the sample from the 81th ranked rikishi (Ms6e) instead of the 71st (Ms1e), the numbers might line up better.