Subject: Re: Timing: Zweig Breadth Thrust indicator
Hi Ted .... That is the correct hypothesis! The answer or approach in these things(for a ruleset of such prognosis) should be:

Generally markets are up about 70% of the time 1yr+ out - so for at least a +20% lift for the rule - the accuracy needs to be 84%+. This one is close I think ( at least 80%)

However, the attraction for this rule (at least in folklore) lies more so in the YoY% returns which are distinctly higher over the dataset.
But therein lies the FALLACY. I think when Zweig published it - and no ill will towards him - it must have taken quite a lot of research and without the computing power these days - but he must have fine-tuned it to ensure it ONLY TRIGGERS at end of "then known" major bears.

Things have definitely changed since ( You can see it in the column Days Between) - it was very infrequent all the way to 2009 and then something changed! ( I am guessing the BTD mentality) - it has become a much more frequent one - at least on my threshold version.

I tend to take these things - especially with a whiff of data-fine tuning with a bagful of salt - so in my version the thresholds are <40 and >=60. Why? because in my mind that represents something more intuitively defensible - a +50% turn/reversal in sustained buying pressure over last 10 bars

In the table below I am providing ALL signals tracked by 2 methods - One is based on Original Calculation with Pinnacle data ( ie computed) with my thresholds and the other is published by Worden ( this one I follow) - same thresholds. The Stockcharts version not sure if its based on NYTOTIS ( based on the link you shared it could be) - I couldn't reconcile with either of mine. The original defn is : 10 day SMA ( Adv/Adv+Decis) checked over last 10 bars not TOTIS.


Here's a pic :https://www.dropbox.com/scl/fi...

Even so - source data matters because its almost very difficult to find concordance amongst these methodologies. The SC one has more False Positives on 60% thresholds- but it has a signal on 5/20/2004 ie out of the Tech Bear - the Worden one agrees - if you drop the trigger by 1/2 to 59.5.

Table:
Date Orig_Pinnacle Worden Concordance Days Between
12/23/1987 0 1
5/14/1990 1 0 624
5/15/1990 1 0
8/7/1996 1 0 1627
10/21/1998 0 1 576
10/22/1998 1 1 TRUE
10/5/2001 1 0 772
12/6/2007 1 0 1610
2/1/2008 0 1 42
12/5/2008 1 0 221
12/8/2008 1 0
3/17/2009 0 1 72
3/18/2009 0 1
3/19/2009 1 1 TRUE
11/16/2009 1 0 173
2/18/2010 0 1 69
2/19/2010 0 1
7/20/2010 1 0 108
9/3/2010 0 1 34
3/30/2011 0 1 149
10/12/2011 0 1 141
10/14/2011 1 1 TRUE
10/17/2011 1 0
12/9/2011 1 0 40
11/23/2012 0 1 251
11/29/2012 1 1 TRUE
11/30/2012 1 0
7/9/2013 1 0 158
10/17/2013 0 1 73
8/19/2014 1 0 219
8/20/2014 1 0
8/21/2014 1 0
10/21/2014 0 1 44
10/5/2015 0 1 250
10/7/2015 0 1
10/8/2015 0 1
10/9/2015 0 1
12/29/2015 1 0 58
2/22/2016 0 1 40
2/26/2016 0 1
11/14/2016 0 1 187
11/15/2016 0 1
2/26/2018 1 0 335
11/7/2018 0 1 183
1/7/2019 1 1 TRUE 44
1/11/2019 1 0
1/14/2019 1 0
5/26/2020 0 1 357
5/27/2020 0 1
10/5/2020 0 1 94
11/10/2020 0 1 27
5/27/2022 1 0 404
6/2/2022 1 0
3/30/2023 0 1 216
3/31/2023 1 1 TRUE
4/3/2023 1 0
11/3/2023 0 1 155
8/19/2024 0 1 207
4/24/2025 1 179

NET : I treat this now more as a 2-3 month out bullish signal. :) Maybe I am also "fine-tuning" LOL! -Because that smooths out the 2 falsies in Feb-2008 and Nov-2018

Hope this helps!