I recently spent some time analysing VantagePoint’s “Predictive Neural Index” indicator and thought I’d share the results with you. As you will see, although the specific claims the company makes appear to be accurate, as far as I can tell it’s still completely worthless. The following text is a copy of an email I sent to the company on 21st May 2012.
I’ve taken a look at the analyses by Dr Meyer and Dr Warnock, and also of those by Dr Harris and Dr Arcuri. Although their tests were technically accurate in terms of their methods and conclusions, I’m afraid I cannot see any evidence that VantagePoint’s Predictive Neural Index (PNI) indicator is capable of providing any statistically-significant advantage, nor in fact that it has any predictive abilities beyond those available with simple, traditional (and free) indicators. Furthermore, I believe I can demonstrate conclusively that it does not –*at least with the sample data you provided.
The PNI indicator is described as a prediction of whether the 3-day simple moving average (SMA) of the typical price will be higher or lower in 2 days time. As I’m sure you know, a 3-day SMA in 2 days time will still use today’s data for one third of its calculation. Also, a 3-day SMA lags the price by 1.5 days, so the prediction is only really for half a day in the future. Furthermore, by using the typical price (ie the average of the high, low and closing prices), the PNI does not even provide a directional forecast that you could reasonably expect to trade, since it only applies to the range of the price, not the closing price.
I compared the results from the PNI with some very simple predictions of my own based on the current slope of a moving average over various periods and using several different averaging methods. In each of these tests, I predicted a rise in the 3-day SMA of the typical price in 2 days time (equivalent to a PNI of 1) where the slope of the moving average was positive, and a fall (equivalent to a PNI of 0) where the slope was negative. The full results are included in the attached spreadsheet.
For the sample data that you supplied, from June 2010 to May 2012, the PNI indicator achieved 78.1% accuracy for both the S&P 500 and the GBPUSD. Over the same period, the slope of a 3-day exponential moving average of the closing price, for example, produced an accuracy of 78.2% for the S&P 500 and 77.6% for the GBPUSD. With a 2-day exponential moving average, the accuracy for the S&P 500 increased to over 80%, while still achieving an accuracy of 76.4% for the GBPUSD. On average, these results have an accuracy of 78.1% – identical to those achieved by VantagePoint’s PNI indicator.
I accept it is possible that the PNI may perform better if I were to test it over a wider range of markets (and I would be happy to do so if you were to provide the additional data). However, given my understanding of the mathematics involved – and given that your own published results state very similar accuracies –*I would be very surprised if this were the case. I’m also surprised that the mathematicians and scientists who have verified your results have apparently failed to spot these facts.
If you believe I have made some error in my calculations, or have some other information you can provide that might change my opinion of VantagePoint, I would be very interested to hear it. As a courtesy, I will wait until Saturday 26th May before sharing these results via my blog, or any other website, in order to give you sufficient time to respond.