Skip to main content
Track Fast Fourier Transform
11 / 16
Search the archiveLibrary
Library

1999issue C051-10

Walk-forward endpoint Fourier construction as a same-day mechanical procedure

A full-window Fourier overlay on later-known prices can appear to lead a peak that a same-day window does not signal. Endpoint flattening and a walk-forward indicator that keeps only each day's last noise-filtered Fourier point turn that overlay into a real-time curve a mechanical trading system can test.

  • A full-window Fourier overlay on later-known prices can appear to lead a peak that a same-day window does not signal.
  • The discrete Fourier construction treats a sampled window as periodic, so unremoved trend and mean create spurious frequencies that swamp real ones.
  • Endpoint flattening forces the first and last window values to zero so wraparound jumps do not dominate the spectrum.
  • The walk-forward indicator joins each day's last noise-filtered Fourier point into a real-time curve that lagged major turns by zero to four days rather than leading the price series.
Entries in this reading3 entries
Fourier analysis14 readingsCurrent trackFast Fourier Transform17 readingsMechanical trading system158 readings

The full-window overlay problem

A full-window Fourier overlay on later-known prices can appear to lead a peak that a same-day window does not signal. Fourier analysis of a completed window therefore uses prices that a same-day window does not have, and the same-day window does not produce that leading overlay.

Full-window noise-filtered FFT on S&P 500 futures through the 1998 peak

The red overlay turns down before the July 1998 high, which is the illusion Meyers is warning about: this Fourier curve was fitted to the entire later-known window, so it already contained the crash. Values were read from the published TradeStation daily chart of S&P 500 continuous futures and its noise-filtered FFT.
The red overlay turns down before the July 1998 high, which is the illusion Meyers is warning about: this Fourier curve was fitted to the entire later-known window, so it already contained the crash. Values were read from the published TradeStation daily chart of S&P 500 continuous futures and its noise-filtered FFT.S&P 500 continuous futures (SP) · Daily · 1998-03-25T00:00:00.000Z to 1999-02-12T00:00:00.000Z

Meyers computed the FFT on the 16 January 1998–22 January 1999 window; the screenshot is dated 12 February 1999 with a printed close of 1239.27. Intra-month dates follow the monthly axis labels; 20 July is the closing high named in the text. Digitized levels are to the nearest five index points.

Periodic windows and spurious frequencies

The discrete Fourier construction treats a sampled window as periodic, so unremoved trend and mean create spurious frequencies that swamp real ones. The window has to be prepared before the Fast Fourier Transform, or those wraparound artifacts dominate the spectrum.

Endpoint flattening

Endpoint flattening forces the first and last window values to zero so wraparound jumps do not dominate the spectrum. With those ends pinned, the Fast Fourier Transform is less occupied by the artificial jump created when the sampled window is treated as a loop.

The walk-forward endpoint curve

The walk-forward indicator keeps only each day's last noise-filtered Fourier point and joins those endpoints into a real-time curve. Each new day brings a new window, a new flattened transform, and a single retained endpoint.

That endpoint curve lagged major turns by zero to four days rather than leading the price series. The same-day construction does not reproduce the apparent lead of the full-window overlay.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
11 of 16 in the Fast Fourier Transform track
20021-4 pp.Next on Fast Fourier TransformFrom the power spectrum to indicator windowsFourier analysis reconstructs a complex series as sinusoids of amplitude, frequency, and phase, and spectral analysis uses that same construction to identify dominant frequencies in ordered observations.
All readings on this track · 16 readings
  1. 1982Building FFT spectra to size cycle filters
  2. 1988Fourier cycle models break in major swings
  3. 1988Constructing moving average filters from price Fast Fourier Transforms
  4. 1989Staging Fast Fourier construction under memory limits
  5. 1993Constructing forecast inputs with moving averages, Fourier transforms and intermarket spreads
  6. 1994Preprocessing prices so Fourier peaks set moving-average lengths
  7. 1994Constructing a spreadsheet FFT power spectrum from daily prices
  8. 1994Building dominant-cycle spectra with FFT preprocessing
  9. 1994Constructing labeled cycle lengths from FFT spectra
  10. 1999Fast Fourier Transform reconstruction is not a walk-forward decision tool
  11. 1999Walk-forward endpoint Fourier construction as a same-day mechanical procedure
  12. 2002From the power spectrum to indicator windows
  13. 2003Endpoint Fast Fourier Transform evaluation with walk-forward mechanical rules
  14. 2004Constructing signal and noise from market waveforms
  15. 2012A two-stage case study in market cycle analysis
  16. 2015Whitening pink noise to build a near-zero-lag cycle oscillator
By fieldCycles and spectral analysis77 readingsBy year199959 readings
All 17 readings tagged Fast Fourier Transform
Also on Fast Fourier Transform5 readings
1982Building FFT spectra to size cycle filtersFourier analysis1988Constructing moving average filters from price Fast Fourier TransformsFourier analysis1988Fourier cycle models break in major swingsFourier analysis1989Staging Fast Fourier construction under memory limitsFourier analysis1994Constructing a spreadsheet FFT power spectrum from daily pricesFourier analysis