Machine Learning with R #ideas

Machine Learning with R Every time I’ve applied Machine Learning, I end up curve fitting but I keep soldiering on expecting that sooner or later something will click. I loved Machine Learning with R which is full of examples and has a nice overview of the field. The snippet on comparing two variables using CrossTable() Read more about Machine Learning with R #ideas[…]

Are Two Edges Better Than One? #idea

Are Two (Uncorrelated) Edges Better Than One?* Our results pre- and post- Market Stress, our second edge: All Market Stress Change Avg 67.61 93.04 38% Std dev 1042.9 299.79 -71% llt (8640.00) (1000.00) -88% w/l ratio 0.83 0.65 -21% win pct 60% 78% 31% f 0.11 0.45 303% f$ 76591 2200 -97% z 2.32 2.22 Read more about Are Two Edges Better Than One? #idea[…]

What’s Missing – the Error Term #idea

What’s missing? [t+1] = how I forecast what comes next What’s missing is the error term – [t+1] = how I forecast what comes next + (error) The error term is what completes the performance triad: Visualize, Execute, Analyze Learning, continuous improvement and perhaps even ideas and serendipity happen faster with the explicit use and Read more about What’s Missing – the Error Term #idea[…]

New Information Rate is 15% #idea

“A good 15% of the search questions it [Google] sees every day are new – queries it has never answered before.”  <http://www.bbc.co.uk/news/technology-23866614>  If 15% new information is coming into search each day, then there should be a similar percentage of new information coming into the markets – an absolutely huge number of underlying change which Read more about New Information Rate is 15% #idea[…]