RE: One for Agatha Christie & Sir Arthur Conan Doyle, sleuths.12 Mar 2023 21:01
"velo,
can you explain the technique behind "seasonal probabilities"
_ _ _ _ _ _ _ _
Hi Toptiger,
If I don't post now, it'll be late next week at this rate! :)
1) Seasonal (years of data)
2) Probabilities (close to the month - either up or down)
3) % of gain/loss (% movement by month close)
1)
The seasonal part refers to those months that have shown a "statistically significant" repeatability over a given period of years.
2)
The probabilities part refers to the percentage probabillity given to the data that a certain month will close higher than it opened (rather than down).
Both Jan & Feb had 100% assigned to them as virtually a certainty that those months would probably close higher than they opened.  And in this case they both did close up on their month.
3)
However, both failed miserably to close anywhere near the SP expected, as the performance percentage gains/loss assigned to each month were extraordinarily high double figure percentage gains.
On the other hand, those months expected to close down on what they opened, are given v low percentages that they don't stand any chance of closing up on the month.
So a severe 0% probability assigned to a month is the exact opposite of a 100% assignment with the data saying no chance that such and such a month will close higher than it opened. There is only one bearish month that appears every year for SLP with a 0% rating
- and that is June!
It's a single % figure pullback assigned to June but based on Jan & Feb accuracy I'm not looking for that kind of accuracy - only the v strong probability that June closes lower than it opens (down).
0% and 100% probabilities are usually accurate.
Amount of % gains or losses, are a bit hit and miss.
PS. Most stocks/indices vary between lowish and high %.
Specific 0% and 100% probabiliities are v strong but not universally common, but appear often enough to search for.
It's the No: 3) % gain or % pullback that can be very ropey. Usually wide of the mark.
The extreme 0%Â & 100% months usually do comply but as said, are a bit ropey on the specific amounts in either direction.
Data is compilled for month end, to month end.
(End of the preceeding month, to end of the current month in question).
Seasonality is that sphere of investing that searches out statistically significant repeated data over the same periods each year - over 5 years,10 years, 20 years, 80 years.
80 years indices ie., tendencies only - not reliable probabilities - examples:
(Best performance tendency is that generally, it's best to buy stocks in late Oct and sell in mid Jan to replace the outdated sell in May and go away)
( Tendency for the last trading day before a Bank holiday and the first day after the bank holiday to be bullish).
All above not to be used in isolation.
Has to be backed up with research/analysis.
ie., A small side-hustle to add to your primary analysis.
Fun though :)