2d
6m
ahead
Invesco Nasdaq-100 ETF
Predictive Factors
The models identify what we call the causally predictive factors: a small set of assets judged to be the most important driving forces behind your selected item's forecast. These are what we believe are pushing the search target(s).
iShares 1-3 Year Treasury Bond ETF
Copart, Inc. (DE) Common Stock
Consumer Discretionary Sector SPDR ETF
as of
Our Probabilistic Prediction
With the predictive features above, we run many models repeatedly and aggregate them to account for randomness introduced at the training start.
Goodness of Probabilistic Prediction
When prediction is done well, the realised probability level should not be concentrated particularly around anywhere but uniformly between 0 and 1 (whitepaper). We draw a sample of the realised probability levels and check its empirical probability density. The dotted line for that of the uniform distribution.
Under the hypothesis the realised probability level follow the ideal uniform distribution, a p-value can be calculated for the chance (0 to 1) of obtaining this sample. A smaller p-value for a slimmer chance. Below uses significance level 0.001 as cut-off on the p-value. The user is free to use another e.g. 0.01, 0.05 for the test of goodness.
Invesco Nasdaq-100 ETF
š“
p-value: 0.0
< 0.001
Which direction do the bulk of the models think movement will shift towards? With each model casting a vote on the likely change with respect to a value of 414.65 on the Fri 19 Apr, and following an averaging procedure of the most recent net predictions, expected future positions are
|
% Up |
% Down |
% Undecided |
| coming day |
53% |
4% |
42% |
| the day in 1 months |
55% |
5% |
40% |
| the day in 2 months |
59% |
5% |
35% |
| the day in 3 months |
62% |
6% |
32% |