Daily forecast accuracy sample

I wish to stick to showing you samples of the r.Virgeel activity, in (almost) real time. There is no other way to test r.Virgeel than real time. Also, to avoid the “well chosen sample” effect.

Here, on the left, the forecasted bars evaluated on May 17th, on the right the chart of the actual bars until yesterday close, May 30st.


Left: forecast of May 17th Right: actual bars at May 30th


I show you the forecasted bars because it is by far the most difficult indicator to calculate, bars come in from the future squishly jigsawing and with modulated velocity.  So whenever I want to evaluate r.Virgeel model, I first look at the future bars. Also, the future bars are a totally unattended calculation, meaning that there is no human intervention at all, just the correlation of a huge number of factors is taken in account.

Looking at the forecast, when it was issued, you might argue that a choppy period was to begin, that should not exceed the last high and that might break the last low.  Two down days, followed by a reaction that flattens and returns slightly negative. You ‘d have known all this in advance, various days in advance. Either you are an investor or a trader or even a daytrader, you might take advantage in various ways from the forecast, but you are psychologically prepared to what is coming.

From the comparison ot the two sequences, I point the attention on the fact that the incoming wave is demonstrating to come in faster than expected by the forecast. This is a fact I’m long pondering. Almost all r.Virgeel forecast show this reverse lag, but I’m still detecting. One possible reason is the fact that the databases of r.Virgeel goes back to an era when everything, in trading activity, was slower: no internet, no direct instant execution, non continuous markets, just newspapers and a phone to pass your orders to your broker or bank. The ever accelerating time implicit in the trading activity is an interesting subject of research. Another possibility is that time in itself might be an accelerating media, but here I can’t go further.




error: Content is protected !!