Leaderboard

So many people get very confused with probabilities. Those with no formal education in the subject can get definitely confused. But even those with a formal education in the subject easily can get off track. One area of confusion I see often is people thinking that processes, such as stock movement, follow probabilites, instead of probabilities describing processes. Probabilities are just mathematical descriptions based on past history and other considerations. The processes themselves know nothing about the probabilities. They don't know when you had started counting your sequence, for instance. They have no memory. How about the so-called gambler's paradox. A gambler is watching a roulette wheel and notices that it has landed on black ten times in a row. He knows that the odds are 50/50 (ignoring "0" and "00" for the moment) between landing on a black or a red for each individual spin. He also know that the probability of getting ten blacks in a row is 1/1024 (2 to the tenth power is 1024); that is, virtually zero. So he thinks that FOR SURE the next spin has to be a red since the chances of NOT getting a red in 11 spins is virtually zero. What the gambler has forgotten (or, more likely simply ignorant of) is that the roulette wheel has no memory. For each individual spin the chances are 50/50 between red and black regardless of the recent past. So even though it is true that BEFORE THE 11 SPINS there is an exact probability of 1/2048 that in 11 spins of a roulette wheel you would get 11 blacks in a row, once the first ten have occurred and you notice all blacks, you cannot say anything about the next spin other than that there's a 50/50 chance for either black or red--just like all the other spins. Again, the roulette wheel has no memory of what has occurred prior. The same is true with stock market probabilities. Even if they were exact like the roulette wheel (and stock market probabilites are surely not exact), you can use them only as guides into the future for a particular time range. But once the stock movement has started to occur, you need to at the very least recalculate the probabilities for whatever range of time is still of interest to you. If the stock hasn't moved as much as it was "supposed to" have during the initial part of the time range, you cannot assume it will "make it up" in the latter part. If you think this, you are falling victim to a version of the gambler's paradox. One more thing: Along the lines of what I just said, I've seen traders completely misuse probabilities and standard deviations when attempting to make correct market decisions. I once read a pseudo-scientific article written by someone whom I believe was sincere. But ... oh so how misguided! His system went something like this. Throughout the trading day he constructs a bell-shaped (normal) curve of a stock based on its intraday movements. If a majority of the bell curve seems to be being built on either the left or the right side of the bell, he says there is a great chance that the rest of the bell "has to be" built on the opposite side. Thus, he would use this "knowledge" to day trade the latter half of the trading day. The errors in his thinking are numerous. One of the errors, of course, is that the stock movement doesn't "know" its on someone's left- or right-side of a bell. Indeed, if he had instead constructed a bell in real time with a total time range of only the first half of the day (instead of the full day), he would have ended up with a completely different bell that is is "filled in" on both sides. Everything about his method is arbitrary including the time range. There are an infinite number of time ranges and an infinite number of possible histogram bin possibilities when constructing your histogram. Yet he swore by this "scientific" system. Lots of ignorance abounds.

On your first point, that had also crossed my mind and more so as I've been examining how commissions heavy the trades tend to be. I agree that it is possibly a vehicle for them to generate commissions. I figure that if the service sucks (results in net losses), that they won't be able to have a sustainable business and will lose subscribers. However, if there enough incoming new people to replace them... But yes, the subscription fee per strat is tiny compared to the commissions. I wonder how other newsletters handle auto-trades across multiple brokerages? Maybe I should start another thread for that? To me, there'd be the same issue that we have here, where a trade alert goes out, adjustment, roll, closure, etc. goes out, they could automatically place a limit order but there's no guarantee it'll be filled. And, how far away should the limit be set? With Red Option's use of a house account and divvying up afterward, at least everyone w/autotrade on gets a fill at the price mentioned in the alert.