Let's talk probabilities.

Discussion in 'Trading' started by SHORTY, Dec 21, 2001.



    Tony has obviously sparked quite an interest (myself included) when he posted that a stock that closed within 10% of it's high for the day had a 85% chance of having a higher high the next day.

    I think we all recognize that trading is a game of probabilities with traders looking for the highest probability set-ups. Does it seem strange to anyone else that there isn't published information stating tested conditional probabilities for various set-ups? I've long wanted to perform extensive backtesting on hundreds of set-ups to arrive at the probability, but was basically too lazy.

    I think I'll undertake this challenge and publish my results in a book for $50 or so. I think the probability of me getting rich doing that is greater than me figuring out this trading stuff! :cool:
  2. The "old" Forbes used to say: "You make more money giving advice than following it", so writing a book may not be a bad idea.

    Larry Williams, Jake Bernstein and other figured that out millions of dollars ago :)
  3. Joshua


    Just to throw another log on the fire.....
    I've also heard that a stock that has been consolidating will break out in the direction of its previous trend two thirds of the time. Anybody ever hear anything like that?
  4. jaan


    i don't think that's true in general sense. as i mentioned on tony's thread, i've done extensive analysis of setup probabilities in the last three years, and i've yet to see non-godmode (ie realtime) setups with historical probabilities greatly over their random counterparts.

    to put it another way: i've noticed that the signal-to-noise ratio is less than 10% in stock data. so if the random probability of an outcome would be 50%-50%, then - over statistically significant sample - the historical data would not stretch this beyond 55%-45%. usually much less.

    - jaan


    That's really depressing! :(
  6. jaan


    well, sort of. then again, in short timeframe you don't really need much of an edge to be decently profitable.

    - jaan
  7. jaan,

    actually 55-45 is quite an edge I would think. You can make enormous amounts of money with a smaller edge than that. Systems or pattern-based trading is a statistical exercise. You grind it out over large numbers of trades. The crucial piece of data is how much adverse movement do you have to tolerate to get movement in the desired direction.
  8. Since we are not allowed to play Blackjack in any of the World's casinos (anyone who can read a book get's barred from the bigger tables)...we developed a mathematical system of playing baccarat...since we like to have our travel paid for and take advantage of "comps" and the like.

    Baccarat is nearly a 50/50 game, where you can only bet on the bank or the player, and you have no decisions to make. The bank has a very slight edge in the long run.

    We bet this same system (on the bank) and only need to win just above 35% of the hands to win. We win approximately 18 out of 20 times. If we follow the "system" exactly, and don't get cut off at the table limits (which is really irritating), we have a great time 18 out of the 20 times.

    You know this is coming....those times that we do lose, we can lose everything we made plus a bunch...now since we do this for enjoyment, we factor in the value of the trips and comps, and we overcome the 1.2% negative edge.

    Back to the point...for decades the idea of buying stocks at their high, or near it, for the next days run has been tried...and you see where this is going...the losses far exceed the profits in the long run.

    I think it is extremely valuable that new traders put into practice their ideas and that they keep searching for a better way, but don't try to re-invent the wheel with the old stuff (it can be very time consuming).



    I was being facetious. There has to be favorable probabilities to stock set-ups or stocks wouldn't bounce of of S/R or follow trend lines. Also, the price one day could be multiples away from the price the next based on true randomness. I know they're there, they just need discovered, at least by me. The only truly depressing outcome would be if the buy and hold crowd was right! :p
  10. Originally posted by Don Bright

    [ Back to the point...for decades the idea of buying stocks at their high, or near it, for the next days run has been tried...and you see where this is going...the losses far exceed the profits in the long run.]

    Then why wouldn't the opposite work? Instead of buying at or
    near the high of day, just short at the end of day and hopefully
    cover at a lower price the next day.
    #10     Dec 23, 2001