What approach would you recommend to calculate current tendency in opening position?

You have set of (move+year) values from games.

Amount of total games is unknown but will be somewhere between 1 and 100.

I can just sum some f(year) for each move and compare.

a) Would you do it differently? b) If not, what f(x) would you use?

2 Answers 2


Would recommend choosing a year to serve as a base year, then calculating a weighted average based on that. For example, if we choose 2000 as our base year, and say Ruy Lopez was played once in 2000, twice in 2001, and three times in 2003, we would sum (1 * 0 + 2 * 1 + 3 * 3)/6 = 11/6.

Alternatively, you could also compute the median (which in this case would be 2). Or, you could graph the data entirely, which would probably give you the best picture.

This definitely sounds like an interesting study, good luck! Let me know if you have followups

  • Good point in graphing data! I completely forgot this possibility in this particular case. I wanted to sort moves from most modern, but with graph visible for every move, the sorting method isn't that much important. To the number, I'm not sure it should be done in a linear way. My favorite is sum(q^(gameyear-currentyear)) for q somewhere between 1 and 2. But can be many other approaches, different f(x) etc... In linear case difference between 2005 and 2000 is the same as for 2016 and 2011. Should not be in my opinion, as 2000 almost equal 2005 but 2011 definitely doesn't equal 2016.
    – hoacin
    Dec 28, 2016 at 8:03

You should think what the typical timescale for opening trends would be. Also to make the result meaningful you should take into account changes in move order that lead to the same position/opening.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.