Comments on: Using a Self-Similarity Metric with Intraday Data to Define Market Regimes

By: Jeff Swanson

Jeff Swanson — Wed, 17 Nov 2021 12:47:43 +0000

In reply to Mark. Not sure if that's what the author did. But you're right that you need to test this regime filter with your particular market and strategy.

By: Jeff Swanson

Jeff Swanson — Wed, 17 Nov 2021 12:45:46 +0000

In reply to Michele. Yes, please do.

By: aaron

aaron — Tue, 17 Sep 2019 03:04:04 +0000

Has any one done tested this regime filter on a very basic mean reversion strategy (like 2 period RSI) and a very basic trend following one (like 20 period donchian channel breakout) ??? would be curious to see how the regime indicator adds to those types of strategies.

By: Michele

Michele — Tue, 01 May 2018 08:13:59 +0000

Hi Mr. Jeff Swanson,
first of all many thanks for sharing your code and for your very useful article. Am i allowed to code it in my own R package naming and linking you?
Thanks.
Michele Stasulli

By: Mark

Mark — Fri, 01 May 2015 16:11:46 +0000

Is the graph simply the performance of SPY if you were to go long when the market is “stable” (green) or “chaotic” (red)?

Even if that is the case, isn’t it a leap to say quantitative trading systems in general work better when the market is stable as opposed to chaotic?

By: Gary Hart

Gary Hart — Thu, 16 Apr 2015 23:34:09 +0000

As a follow up…TradeStation’s PercentRank function is dreadfully slow. I found a much faster function (_PercentRankFast) on the internet here: http://dvindicators.cssanalytics.com/community/?vasthtmlaction=viewtopic&t=47.0

Here is the text file for the _PercentRankFast function (you’ll find it is MUCH faster than the TS built-in one):

inputs:
ValueToRank(numericsimple),
DataSet(numericseries),
Length(numericsimple);
variables:
R(0),
X(0);

R = 0;
for X = 1 to Length begin
if DataSet[X-1] <= ValueToRank then
R = R + 1;
end;

_PercentRankFast = R / ( Length – 1 ) ;

By: Michael Harris

Michael Harris — Thu, 16 Apr 2015 16:07:39 +0000

Hi Jeff and David,

I think this is an interesting indicator. I have posted some results in my blog: http://www.priceactionlab.com/Blog/2015/04/performance-in-stable-and-chaotic-markets-based-on-the-cssa-regime-indicator/

By: Bob

Bob — Tue, 14 Apr 2015 11:00:51 +0000

I wish the author could participate in the discussion. There are some questions here. Is this fitted to S&P 500? What about oil of gold? How is this different from a momentum indicator?

By: Jeff Swanson

Jeff Swanson — Fri, 10 Apr 2015 00:57:10 +0000

In reply to Gary. Agreed. I've updated the code. Thanks!

By: Gary

Gary — Thu, 09 Apr 2015 18:00:44 +0000

The PercentRank formula doesn’t look correct to me. The way I read it you should be ranking the AvgMetric against its past 252 values not against RawMetric.

Instead of:
pRank = PercentRank( AvgMetric, RawMetric, 252 );

Use this:
pRank = PercentRank( AvgMetric, AvgMetric, 252 );