When visiting Shrewd'm with a laptop, it can be pleasant to hold Command (or Ctrl with Windows) and '+' a few times. The site scales to allow any font size, and the larger font can be pleasant to read even for Shrewds with perfect sight! For luxury Shrewdness, you can combine that with setting the browser to full screen. You'll then find yourself Shrewding a lot.
- Manlobbi
Investment Strategies / Mechanical Investing
No. of Recommendations: 11
As my next assignment to add to my repository service for the group, I'd like to tackle the minor and major market bottom indictors.
I am reverse engineering the math from a set of posts found here:
https://www.shrewdm.com/MB?pid=809124972If anyone has any great ideas for calculating this, please provide them. I can incorporate them into the timing service.
https://github.com/bunbuntigery/momentum-mechanica...(Also, if anyone has a github account, starring the repository is a nice gesture if you think it's useful)
No. of Recommendations: 11
Code is on main:
https://github.com/bunbuntigery/momentum-mechanica...It explicitly used Jim's clusters to backfit a bottom detector using:
Bottom Detector
===============
Identifies major and minor market bottoms using NDX100 breadth indicators
and VIX. Reverse-engineered from a classic NH-NL capitulation model.
Approach:
1. Compute daily NDX100 new 52-week highs/lows from price cache
2. Compute breadth metrics: % new lows, % below 200d MA, NH-NL spread
3. Load daily VIX from FRED (VIXCLS) with dedicated cache
4. Score each day on acute capitulation (spikes, not levels)
5. Detect bottom signals when score exceeds threshold
Strategy (from original model author):
"Ignore signals multiple days in a row, and not buy till the first
day that you DON'T get a signal."
It's built into the telegram and autoupdating bot so reports to you every single day rain or shine.
Here's today's output:
BOTTOM DETECTOR
-----------------------------------
2026-03-20 strength: 81/100
New lows: 5/101 New highs: 1/101 NHNL: -4
VIX: 24
NL volume: 1.6x avg
⚠️ Elevated — watching
No. of Recommendations: 6
And Claude's comment on the signal as it stands:
From the historical data at threshold 88:
Cluster Date Peak Strength Forward 1yr
Weakest signals
2002-07-22 Jul 2002 96 +23%
2018-10-24 Oct 2018 99 +16%
2015-09-28 Sep 2015 99 +17%
Strongest signals
2008-10-06 Oct 2008 100 +3%
2011-08-08 Aug 2011 100 +28%
2020-03-16 Mar 2020 100 +68%
The strength range for actual signals is 88-100. But once you're above 88, the strength number doesn't tell you much about forward returns — the weakest signal (96) and strongest (100) both worked. What matters is that the signal fires at all.
Today at 81 you'd need roughly:
New lows jumping from 5 to 15-20+ (out of 100)
VIX spiking from 24 to 30+
Volume surging on those new-low stocks
That's a meaningful selloff from here, not a normal down day.
No. of Recommendations: 25
My minor bottom detector last triggered 2025-11-20, an unusually good call. It's usually a good buy signal for maybe a month out, but dodgy for longer time frames.
It also triggered the last 3 trading days 2026-03-18 / 2026-03-19 / 2026-03-20
Since the signals often come in runs, especially during cascading market falls, the conservative interpretation is a buy the first day that it does NOT trigger.
The major bottom detector is a long way from triggering.
Jim
No. of Recommendations: 4
Hey, Jim.
I know you posted this in the old motley fools boards which are inaccessible.
If you'll share the methodology and datasets, I can make this do exactly what you do.
I understand proprietary etc, so not a big deal, but seems like some here might appreciate?
I did have a modestly difficult time sourcing nhnl freely.
No. of Recommendations: 24
If you'll share the methodology and datasets, I can make this do exactly what you do.
I described it in moderate detail before, though not a couple of probably meaningless tweaks that make me feel important and mysterious.
Broad strokes for the major bottom detector:
It's based mostly on new highs and new lows, each expressed as a percentage of total issues, looking for an extreme in both of those at the same time. Low #highs and high #lows. Nothing surprising. Those numbers are smoothed a bit.
Those two numbers are combined with a relative weighting into a single metric, as they aren't equally important. That combo metric is ranked among a lookback history of itself, and the percentile tested against a cutoff. Different levels of cutoff give different certainties of a bottom, and have different average forward returns, which is why I sometimes mention whether the signal was weak or strong on a given day. I do have a single "canonical" cutoff though.
My short term bottom and major bottom models are constructed differently, not just two different cutoffs of the same metric.
The main flaw is that it will give a whole bunch of signals all the way down a "waterfall" market crash like late 2008. Despite being bad signals, one year forward returns are still fine from the earlier signals, which is what it's tuned to accomplish, but the first month or two can be (ahem) arbitrarily low, just as a market bottom can be arbitrarily low. Most of the time you only get a short cluster or two, so it's not a big deal, but it gets confused in a big lasting meltdown and signals too early.
Jim
No. of Recommendations: 16
It also triggered the last 3 trading days 2026-03-18 / 2026-03-19 / 2026-03-20
Since the signals often come in runs, especially during cascading market falls, the conservative interpretation is a buy the first day that it does NOT trigger.
FWIW, no signal from 2026-03-21, so the model would suggest a short term buy today. Future suggest an open flat from yesterday's close, S&P 6581.
Happy Taco Tuesday.
Jim
No. of Recommendations: 18
Just a bit of a follow up...
[the short term bottom signal] also triggered the last 3 trading days 2026-03-18 / 2026-03-19 / 2026-03-20
Since the signals often come in runs, especially during cascading market falls, the conservative interpretation is a buy the first day that it does NOT trigger.
...
FWIW, no signal from 2026-03-23, so the model would suggest a short term buy today. Future suggest an open flat from yesterday's close, S&P 6581.
(I corrected my typo of the date of the post, Monday is 3 days after Friday the 20th, not one day, duh)
It has been 22 calendar days since the suggested short term buy. Though the market didn't bottom till a few days later, the S&P 500 closed yesterday up 5.87% from close on the "buy" day (Mar 23, the first full day after a string of bottom signals), or up 6.33% from the next day's market open (Mar 24). I'd call that a win.
The short term bottom signal doesn't say very much on average after about a month, so this signal's predictive power (if any) is fading away soon. Time to look for other omens : )
I see that the trend of Nasdaq NH-NL is certainly rocketing since its low on March 27, so maybe the short term will be a bit longer yet?
Jim
No. of Recommendations: 7
From 3/24:
FWIW, no signal from 2026-03-21, so the model would suggest a short term buy today. Future suggest an open flat from yesterday's close, S&P 6581.
And today the S&P500 is at 7025 (also resetting the 99-day clock).
DB2
