The new TMF board format wiped out my saved file of pithy, thought provoking posts by mungofitch about his 99-day rule, short term bottom detector and major bottom detector.
I did a search for any remnants on the TMF Mechanical Investing board and found the following that appeared back in May 2022 at the same time when WendyBG posted
“mungofitch 99-day rule” triggers
The “mungofitch 99-day rule” is a historical model developed by the eponymous mungofitch which shows that IF the S&P 500 does not make a new high in 99 trading days it is likely to fall further…sometimes much further. By selling all stocks and going to cash, the nasty bottoms are cut off these drops. This model recommends going back into stocks after 99 trading days of making new highs since that usually indicates a durable bull market. The waiting period cuts out dead cat bounces and head fakes in a bear market.
May 2022
Simple detector ‘Minor Bottom’
In response to the OP author’s comment that the Simple Detector, regarding the NYSE, signaled a ‘Minor Bottom’ on 5/2/2022, mungofitch commented:
My short term bottom detector also triggered yesterday May 2.
It has degrees of certainty…this was not a super strong signal.
For example, not nearly as strong as its signal 2022-02-24.
So at this point it’s more like a whining dog, not a barking dog.
My usual comments on this signal:
It’s usually pretty good for positive average returns 1-4 weeks out.
It often triggers multiple days in a row.
It’s not nearly as reliable as my relatively rare “major bottom” detector signal.
In response to a question, Does your bottom detector trigger independently of the 99-day signal?, mungofitch stated:
Yes, totally independent signals because they look at different time frames.
Remember that the 99 day rule is just a particular tuning of the general observation:
The longer it has been since a fresh recent high, statistically the lower the forward return in the next couple/few months.
It’s a gradual roll-off, not a bright line thing.
An old post with some figures.
https://discussion.fool.com/a-new-sampp-high-so-the-99-day-indic…
That table is worth pondering.
Those are 6-month forward average returns, a little bit longer time frame than is generally used for timing signals.
But still, you can see the gradual roll-off with the time since high.
So there is no specific cutoff that is magical. It’s a “broad sweep of history” view of the markets: short delay good, longer delay bad.
It’s attempting to answer the question: “Is there some evidence that you are still in an ongoing bull market?”
Longer time frames are bad, so you have to pick a useful cutoff. The 99 day level was my chosen compromise.
A person who’s bullish by nature might pick three months. A very conservative investor might pick six months.
This helps explain why you can have a bottom detector when things are bullish in that view, or when they’re not.
The bottom detector is a much more short term “pick a particularly good day to buy” thing.
My short term bottom indicator is a reasonably good omen for 1-4 weeks.
That might happen in a bull or in a bear.
By contrast:
The major bottom indicator, if I’m being braggardly, is much more solid: when that happens, just go buy something, dang it.
Last signal dates were this cluster:
2020-03-17 (borderline)
2020-03-18 (borderline)
2020-03-19 (borderline)
2020-03-23
Average S&P close on those four days: 2426, average one year no-div nominal index return: 34%.
Last signal before that was this cluster:
2018-12-20 (borderline)
2018-12-21
2018-12-24
2018-12-26
Average S&P close on those four days: 2393, average one year no-div nominal index return: 63.5%.
So, the 99 day signal is the climate. Is it still summer?
The short term bottom detector is today’s weather forecast. Is it likely to be sunny in the next week? (up to a month)
The major bottom detector is: is today likely to have been about the coldest day we’ll see in the next 6-18 months?
Jim
——————————-
In response to another question, “How’s your major bottom predictor? You know, the one that I ignored in March of 2020…”, mungofitch posted:
Getting close to a signal.
Being an inveterate overtuner, I have a multiplicity of models.
The traditional major bottom detector hasn’t triggered yet, but it getting quite close.
A couple of spectacularly horrible days would do it.
But, for no reason at all, I think we won’t see that yet.
But my rarest and most strict short term bottom detector has triggered, both yesterday and today.
(the extreme end of the short-term-bottom model’s outputs, so it’s edging towards behaving like the major bottom signal)
It has signalled on only 164 days since 1966, and most of those are clusters, so it’s pretty rare.
Last few signals:
Two this week.
One in March 2020, one year forward 67%.
Three days at Christmas 2018, one year forward averaged 35%.
And oddly, a collection of 10 days in October 2018. One year average forward return only about 9%.
All of the short term bottom detector variants are giving their strongest signals yesterday and today.
They generally portend a higher market level 1-4 weeks out.
On average, returns are slightly higher than average even 3-6-9-12 months out from those signal days.
————————————
I highly recommend reading the entire thread.
Many here already know, Mungofitch no longer posts at TMF … his last TMF post was on 9/29/2022.
Regards,
Ray