Money? Markets? Autoregressive heteroskedasticity? Ask me about it.
Just for kicks, I did a similar analysis for a market closer to home: the Nasdaq ETF (QQQQ).
The formula in this case is a similar log-periodic model, but with two instead of one periods.
a + bln((tc - t)/tc)(1+cSin(wln((tc - t)/tc)+phi))(m+dCos(ww*ln((tc - t)/tc)+phim));
And of course the optimization results:
Algorithms: Levenberg-Marquardt
Correlation Coef. (R): 0.979168286789716
R-Square: 0.958770533854707
| Parameters | Best Estimate |
|---|---|
| a | 16.89154148 |
| b | -12.98218423 |
| c | -0.049999806 |
| d | 0.049999986 |
| tc | 8.532222133 |
| phi | 8.874795769 |
| w | 14 |
| ww | 7.3765478 |
| phim | 6.63123054 |
| m | 1.13898033 |
