I don't knowanything at all about stocks, but let us try to talk about prediction in general.
You don't have to know the "Mechanics-Make-Up" of a "Game" to attempt prediction, but your predictions might be more accurate-right if you do know the mechanics-make-up of the particular game, in your case it could be called the "Stock(s) Game".
Let us look at cars on the street, we know that there are many colors of cars, but we don't know how many colors and the names of all those colors, we also know that there are many brands or makes of cars.
So if a white car just came by, we don't expect the very next car to come by to also be white, it can be, but with so many colors, What are the chances?
If a Chevy just came by, we also don't expect another Chevy to be the very next car to come by, What are the chances of that?
Of course a white car can come again next and a Chevy can also be next, but no very likely for either.
Now if the white car was a Chevy, it is for sure not very likley the the next car to come by will also be a white Chevy, What are the chances?
Now if a white Chevy Impala just came by, another white Chevy Impala would be very much unlikely to be the next car to come by, Right?
So the first filter is white cars, the next filter is Chevy cars, the last filter is white chevys.
But the white Chevy Impala that came by was a 1966 year.
So we have another filter, What are the chances of the next car to come by to also be any 1966 year make?
Specially because it is such an old year.
And we don't even have all the needed stats in relation to colors, kinds or brands and years of make, but we can tell due to what little we know about random chance and common sense.
If a yellow 2003 Ford truck came by and then a a red 1999 honda car came by next, we don't expect another yellow 2003 Ford truck to come by followed by another red 1999 honda car, as a matter of fact we don't expect either of those to to come by again next nor any time soon.
We don't even expect any yellow car nor truck to come by next folowed by any red car or truck.
We don't expect any 2003 car or truck to come by next followed by any 1999 car or truck.
That is not quite prediction, but it is a sort of prediction, those are just an example, a beginning on making a prediction filters system for cars in that kind of situation, as I said, we don't even have all the needed stats for proper prediction.
Besides just as one particular number can't often at all be predicted so can not one particular kind of car with all its many characteristics be predicted, we could instead predict a portion of the total amount of cars and hope that the next car to come by is among them, according to the filters' stats.
That is in a way what we do with lottery prediction.
We need to know exactly what is what we want to predict and in which way we want to do it or in which bway it is or might be best.
We need many characteristics related to that which we want to predict.
I can't talk about stocks as I know nothing about them at all.
For possible failure of cars we need to know all the related stats, stats = filters = possible prediction, Which cars have the most chance of particular or general motor failures this year?, out of all the total cars on the street now, or out of a particular chosen pool of cars.
You know, some particular kind of prediction, Which particular group of people of those who are 50 to 70 years of age are the most likely to die of a hearth attack?
White males, white females, black males, black females, other males, other females?
Filter? Maybe: White, Black, Other, Males, Females, their particular past stats are needed.