Options

Wenner on Ludescher et al

Steve Wenner, who has been commenting on the Azimuth blog lately, sent me a nice long email in which he studied how well Ludescher et al's method actually works. I asked him I can post it as a blog article. I think I've gotten so caught up in replicating their method that I haven't been sufficiently critical about how well it works.

Our work together with Wenner's might fit together to form a publishable paper.

Comments

  • 1.

    how well Ludescher et al’s method actually works.

    John I cannot find any performance or error analysis on this paper, I checked all over the net. Usually people propose a new forecast algorithm, then they do backtest i.e. start from some time in the past and run their model and see if it could forecast future values correctly moving to present time. Then they do error analysis on the difference between the actual historical values and forecasts.

    Dara

    Comment Source:>how well Ludescher et al’s method actually works. John I cannot find any performance or error analysis on this paper, I checked all over the net. Usually people propose a new forecast algorithm, then they do backtest i.e. start from some time in the past and run their model and see if it could forecast future values correctly moving to present time. Then they do error analysis on the difference between the actual historical values and forecasts. Dara
  • 2.

    The average link strength does seem to go down during El Nino events. This is just based on looking at the graphs, not any formal test. Perhaps Steve would like to test this?

    Comment Source:The average link strength does seem to go down during El Nino events. This is just based on looking at the graphs, not any formal test. Perhaps Steve would like to test this?
Sign In or Register to comment.