Ok, I've had time to digest this and ponder a little bit, and i now believe i have a grasp on what exactly has been proposed. Have not set it up in Excel yet to test anything, but am confident i have the idea.
First of all, let's back away from the trees to see the forest.
First a concept:
1.) LWWLWLWWLLWLLL
2.) RBBRRBBBRRBRRB
We all know what these are.
We call #1 a win-loss registry.
We call #2 a permanence of color.
In either case, we can put these STREAMS into an Excel sheet, graph them, and produce a chart. The chart will go up and down and theoretically cross the zero line now and then. When it crosses the zero line, we call it "balanced" because there are equal number of reds on upside and blacks on the downside. Same with wins/losses. But given a random world, the streams will eventually stray away from zero, and we call this "deviation".
Given four streams, we can have several combinations of deviation. Take colors for example. One might be balanced with equal number of reds and blacks, two might be deviating into the black zone, and one might be deviating into the red zone. Could be several configurations, with four going red at one extreme, or all four going black at the other extreme.
Now, we all know what "reversion to the mean" or "mean reversion" means, right? It's the idea that these streams can't stray too far from some average "mean", and you expect them to return, sometime down the road, to a more equal "balance" between reds and blacks, wins and losses, ect. For example, some people have worked on the idea that if the "standard deviation" reaches a number like 3 away from the mean, then it's time to bet that it will now start to return to the mean, after some indicator.
Moglizu's idea is a variation of the above scenario. However, he seems to have brought one or two novel ideas to the table, worth consideration.
Instead of streams built of wins and losses, or reds and blacks, he proposes to build streams of "runs" and "changes". I think we all know what that means. Two reds is a "run". A black followed by a red is a "change". And like any of these kinds of streams, you could bet that a deviation will "revert to the mean", if you were that foolish. But the way Moglizu bets on reversion to the mean does seems to be an original idea.
Take these four streams:
1.) CCRCRRRCCRCRR +1 deviating into the "R" (run) zone.
2.) RRCRCCRRCRCCC +1 deviating into the "C" (change) zone
3.) CCCRCRRCCCRRC +3 deviating into the "C" zone.
4.) RRCCRRRCCRRCR +3 deviating into the "R" zone.
Now, if you were to bet on all four of these to "revert to the mean", then:
1.) Bet for a change, because it's deviating into the run zone (by +1).
2.) Bet for a run, because it's deviating into the change zone (by +1).
3.) Bet for a run, because it's deviating into the change zone (by +3).
4.) Bet for a change, because it's deviating into the run zone (by +3).
In the situation above, we are disregarding how much the deviations are, and simply betting on a reversion to the mean. If we bet one unit on each stream, it's obvious to see that our bets could cancel each other out. But we don't know that yet until we see what colors (or other EC) we are supposed to be betting, for our bet to represent either a run or a change. By examining the color bets that produced this run/change stream, we might find the following:
1.) To bet on change, we need to bet on RED.
2.) To bet on a run, we need to bet on BLACK.
3.) To bet on a run, we need to bet on RED.
4.) To bet on change we need to bet on RED.
In this case, we can see that two of our bets will cancel out, leaving us with a predominantly RED bet over all. How much to bet on RED depends, and Moglizu has his own formula. Moglizu proposes to disregard any deviations less than 2 points. So, we would disregard #1 and #2 above, leaving us still with a predominantly RED bet, by a factor of 2. But as you can see, it's possible to have a situation where your bets would cancel each other out, leaving you with NO BETS at all. Moglizu says this happens about 50% of the time.
Another detail about his betting formula; when Moglizu says bet on HOW MANY of the deviations, not HOW MUCH, he means...well, let's take the above example of four streams again. You can see that the two "qualifying" streams (streams deviating by 2 points or more) give us a total of 3+3=6 points. This is HOW MUCH. But Moglizu says he does not bet on HOW MUCH. Instead, he bets on HOW MANY, which would be, at most TWO (in the above example), because he is betting on TWO streams that are deviating more than 2 points each. So, two chips of RED.
That's one way to do it. You could also bet just one chip on the dominant color, no matter how many 'qualified' streams were deviating by two or more points. A spreadsheet test could tell what is best, but from what we can gather, Moglizu will bet as many as four chips on a color, if indeed there were four streams deviating all in the same direction.
One more detail. In the above example, the 'LOOK BACK' period over all four streams is thirteen (13). This is a detail Moglizu has failed to clarify. Since Moglizu says he leaves the casino after a gain of 5 units, it implies that his LOOK BACK period starts at the earliest whenever at least one qualified stream starts to deviate by at least 2 points. The earliest that could be is after 6 spins. Then, he just lets his look back period "grow" and "grow"...until he reaches FIVE UNITS profit, leaves the casino, and/or starts over. In this scenario, there would be no fixed and/or no 'rolling' look back period. It would always look back to the beginning of a session ending in 5 units.
Now, if these four streams had come from four independent wheels, we would likely have a losing proposition, given everything we know about betting on reversion to the mean. But Moglizu has come up with a way to relate these four streams to ONE WHEEL, and ONE (the last) SPIN. Again, this is the novelty of the idea, making it worth consideration. Whether it works or not remains to be seen.
Only now do we need to talk about how these four streams are related, and i think this issue has been resolved and understood, given Badger's latest spreadsheet example.
The idea is to come up with at least four streams using some unconventional counting methods, relating four spins from the recent past, to the latest (5th) spin. You could come up with more, but Moglizu has settled upon four.
The idea is that, when these four streams are related in this way, a bet on predominantly RED, or predominantly BLACK has MORE WEIGHT than betting on any one stream, or any number of unrelated, independent streams.
Notice, I've been able to explain the whole concept without ever using the term "linearity" even once. The choice of this stupid term is really where Moglizu screwed it up, and dropped the ball. This unintelligent term does not help one bit to understand this concept. I could not find a more confusing term if i tried.
What should we call it when we bet FOR THE REVERSION TO THE MEAN on the AGGREGATE OF SEVERAL RELATED STREAMS?
Anything but #$@ "linearity"!!
I wouldn't even call it 'singularity'.
I might call it 'gang' or 'bank' betting.
That's basically it.
Everybody should know how to develop four related streams, pegging each of the past four colors (or any EC) to the latest color, and marking down whether it represents a 'change' of colors, or a 'run' of colors. This has been covered, and no one needs to ask any more questions about it.
Back up from the trees to see the forest.
This is a bet on reversion to the mean on an aggregate of multiple related streams of information made up of "changes" versus "runs"...in the hopes that the weight of the aggregate will help predict the next color (or EC)...if and when reversion to the mean does prevail. Screw 'linearity'. What kills this bet is, like any such bet, a continuation/trending of deviation away from the mean.
I don't know IF it works, but THAT is HOW it works.
First of all, let's back away from the trees to see the forest.
First a concept:
1.) LWWLWLWWLLWLLL
2.) RBBRRBBBRRBRRB
We all know what these are.
We call #1 a win-loss registry.
We call #2 a permanence of color.
In either case, we can put these STREAMS into an Excel sheet, graph them, and produce a chart. The chart will go up and down and theoretically cross the zero line now and then. When it crosses the zero line, we call it "balanced" because there are equal number of reds on upside and blacks on the downside. Same with wins/losses. But given a random world, the streams will eventually stray away from zero, and we call this "deviation".
Given four streams, we can have several combinations of deviation. Take colors for example. One might be balanced with equal number of reds and blacks, two might be deviating into the black zone, and one might be deviating into the red zone. Could be several configurations, with four going red at one extreme, or all four going black at the other extreme.
Now, we all know what "reversion to the mean" or "mean reversion" means, right? It's the idea that these streams can't stray too far from some average "mean", and you expect them to return, sometime down the road, to a more equal "balance" between reds and blacks, wins and losses, ect. For example, some people have worked on the idea that if the "standard deviation" reaches a number like 3 away from the mean, then it's time to bet that it will now start to return to the mean, after some indicator.
Moglizu's idea is a variation of the above scenario. However, he seems to have brought one or two novel ideas to the table, worth consideration.
Instead of streams built of wins and losses, or reds and blacks, he proposes to build streams of "runs" and "changes". I think we all know what that means. Two reds is a "run". A black followed by a red is a "change". And like any of these kinds of streams, you could bet that a deviation will "revert to the mean", if you were that foolish. But the way Moglizu bets on reversion to the mean does seems to be an original idea.
Take these four streams:
1.) CCRCRRRCCRCRR +1 deviating into the "R" (run) zone.
2.) RRCRCCRRCRCCC +1 deviating into the "C" (change) zone
3.) CCCRCRRCCCRRC +3 deviating into the "C" zone.
4.) RRCCRRRCCRRCR +3 deviating into the "R" zone.
Now, if you were to bet on all four of these to "revert to the mean", then:
1.) Bet for a change, because it's deviating into the run zone (by +1).
2.) Bet for a run, because it's deviating into the change zone (by +1).
3.) Bet for a run, because it's deviating into the change zone (by +3).
4.) Bet for a change, because it's deviating into the run zone (by +3).
In the situation above, we are disregarding how much the deviations are, and simply betting on a reversion to the mean. If we bet one unit on each stream, it's obvious to see that our bets could cancel each other out. But we don't know that yet until we see what colors (or other EC) we are supposed to be betting, for our bet to represent either a run or a change. By examining the color bets that produced this run/change stream, we might find the following:
1.) To bet on change, we need to bet on RED.
2.) To bet on a run, we need to bet on BLACK.
3.) To bet on a run, we need to bet on RED.
4.) To bet on change we need to bet on RED.
In this case, we can see that two of our bets will cancel out, leaving us with a predominantly RED bet over all. How much to bet on RED depends, and Moglizu has his own formula. Moglizu proposes to disregard any deviations less than 2 points. So, we would disregard #1 and #2 above, leaving us still with a predominantly RED bet, by a factor of 2. But as you can see, it's possible to have a situation where your bets would cancel each other out, leaving you with NO BETS at all. Moglizu says this happens about 50% of the time.
Another detail about his betting formula; when Moglizu says bet on HOW MANY of the deviations, not HOW MUCH, he means...well, let's take the above example of four streams again. You can see that the two "qualifying" streams (streams deviating by 2 points or more) give us a total of 3+3=6 points. This is HOW MUCH. But Moglizu says he does not bet on HOW MUCH. Instead, he bets on HOW MANY, which would be, at most TWO (in the above example), because he is betting on TWO streams that are deviating more than 2 points each. So, two chips of RED.
That's one way to do it. You could also bet just one chip on the dominant color, no matter how many 'qualified' streams were deviating by two or more points. A spreadsheet test could tell what is best, but from what we can gather, Moglizu will bet as many as four chips on a color, if indeed there were four streams deviating all in the same direction.
One more detail. In the above example, the 'LOOK BACK' period over all four streams is thirteen (13). This is a detail Moglizu has failed to clarify. Since Moglizu says he leaves the casino after a gain of 5 units, it implies that his LOOK BACK period starts at the earliest whenever at least one qualified stream starts to deviate by at least 2 points. The earliest that could be is after 6 spins. Then, he just lets his look back period "grow" and "grow"...until he reaches FIVE UNITS profit, leaves the casino, and/or starts over. In this scenario, there would be no fixed and/or no 'rolling' look back period. It would always look back to the beginning of a session ending in 5 units.
Now, if these four streams had come from four independent wheels, we would likely have a losing proposition, given everything we know about betting on reversion to the mean. But Moglizu has come up with a way to relate these four streams to ONE WHEEL, and ONE (the last) SPIN. Again, this is the novelty of the idea, making it worth consideration. Whether it works or not remains to be seen.
Only now do we need to talk about how these four streams are related, and i think this issue has been resolved and understood, given Badger's latest spreadsheet example.
The idea is to come up with at least four streams using some unconventional counting methods, relating four spins from the recent past, to the latest (5th) spin. You could come up with more, but Moglizu has settled upon four.
The idea is that, when these four streams are related in this way, a bet on predominantly RED, or predominantly BLACK has MORE WEIGHT than betting on any one stream, or any number of unrelated, independent streams.
Notice, I've been able to explain the whole concept without ever using the term "linearity" even once. The choice of this stupid term is really where Moglizu screwed it up, and dropped the ball. This unintelligent term does not help one bit to understand this concept. I could not find a more confusing term if i tried.
What should we call it when we bet FOR THE REVERSION TO THE MEAN on the AGGREGATE OF SEVERAL RELATED STREAMS?
Anything but #$@ "linearity"!!
I wouldn't even call it 'singularity'.
I might call it 'gang' or 'bank' betting.
That's basically it.
Everybody should know how to develop four related streams, pegging each of the past four colors (or any EC) to the latest color, and marking down whether it represents a 'change' of colors, or a 'run' of colors. This has been covered, and no one needs to ask any more questions about it.
Back up from the trees to see the forest.
This is a bet on reversion to the mean on an aggregate of multiple related streams of information made up of "changes" versus "runs"...in the hopes that the weight of the aggregate will help predict the next color (or EC)...if and when reversion to the mean does prevail. Screw 'linearity'. What kills this bet is, like any such bet, a continuation/trending of deviation away from the mean.
I don't know IF it works, but THAT is HOW it works.