Comments on: The Inverse Fisher Transform

By: Petra Volkova

Petra Volkova — Fri, 04 Oct 2024 13:59:28 +0000

In reply to Tom. FIR is a very simple filter. For a highpass with a certain cut-off period you need a more complex filter.

By: Tom

Tom — Wed, 02 Oct 2024 20:33:04 +0000

How do you use the FIR filter like that of FIR3 to implement a high pass filter at a certain cut-off period, Tc? Thanks.

By: jcl

jcl — Sat, 23 Dec 2023 15:51:41 +0000

Maybe not enabling the trades?

By: Tom Pham

Tom Pham — Sat, 23 Dec 2023 04:39:52 +0000

I ran the code above (downloaded from the repository) but did not got any trades. What I did wrong? Thanks

By: The Inverse Fisher Transform - QuantInfo - Empowering Algorithmic Trading Insights

Wed, 11 Oct 2023 01:43:04 +0000

[…] This post originally published at https://financial-hacker.com/the-inverse-fisher-transform/. […]

By: Kiran Rama

Kiran Rama — Tue, 04 Jul 2023 14:41:25 +0000

Agree with Rob that it is hard to get positive results in real-word using Ehler’s filters or any of his defaults. They have poor predictability when used as features in ML models too

By: Romuald

Romuald — Sun, 03 Jul 2022 17:10:42 +0000

I am an options trader. Personnally I often use the Ehler’s spectral filters. Not to enter into a trade (I use other methods to do so, based on option strategies) but to estimate a trend, an equilibrium. For example, I will hesitate to enter a bullish strategy if one my Ehler’s spectral filter is in the “up” zone. And it often (not always, as usual in trading) gives me a good advice. Ehler’s spectral filters are one of my “check points” on my check list.

By: Rob M

Rob M — Thu, 26 May 2022 15:02:29 +0000

I’ve used his work from his last book too. I too found it hard to get positive results using his work in my discretionary trading as there were too many signals and trend changes caused his filters to give perfectly wrong timing for entries.

To try to correct this problem I instead first find phases in the market another way besides using his code to give a more orthoganal foundation to his filters. The reason this worked better I think is that, while his work is seminal, his variable lookback periods are derived from within his DSP filters and then reintroduced into other aspects of code in his Bandpass or Decycler filters, for example. Such varying phaselengths/lookback periods are great in principle but his are never free of the filters and can’t be said to be providing them with real world inputs. I think this is why they fail at trend changes.

Instead, I calculate short, mid and long term varying phaselengths from pivots. Short term phaselengths from the overall length of the time between last four consecutive simple high and low short term pivots. (e.g. my short term pivot high has a lower high immediately on either side). And so on with mid and long term pivots being derived from the short term ones where, for example a mid term pivot high is found when a short term pivot high has a short term lower pivot high immediately on either side. (Or as immediately close as they can exist in such cases.) And so on for long term phaselengths.

This took a while but the results are very encouraging and I’m keen to try automating.

By: Petra Volkova

Petra Volkova — Fri, 11 Mar 2022 09:43:04 +0000

I dont know about his workshops, but found his work absolutely useful, at least the algorithms from his books. It may be not so with some of his articles in TASC. But I can tell that we use his spectral filter functions all the time and with good success. You just have to know what they do and when to use them.

By: Jake

Jake — Fri, 11 Mar 2022 09:06:44 +0000

Have you found any of Ehler’s work actually useful in making profitable trading algos? I wasted thousands of dollars on his workshop.