Theoretical question on Probability

Discussion in 'Risk Management' started by themickey, Jul 11, 2025 at 6:44 AM.

  1. Sekiyo

    Sekiyo

    There's a bunch of mean reversion.
    But 12 months is more meaningful than 1 month.
    A monthly return of 7% is 125% annualized.
    Most stocks up 7% won't return 125%.

    30% a quarter is 107% annualized.
    The safest measure is 12 months @ 100%.

    You could expect 6% over 1month or 19% over 1quarter.

    That's the interpolated mean.
    But that shit is dynamic and mean reverts.
    Because the rolling 12 months return isn't stationary.
     
    Last edited: Jul 11, 2025 at 9:01 AM
  2. 2rosy

    2rosy

    This has nothing to do with probability
     
    Drawdown Addict and Sekiyo like this.
  3. MarkBrown

    MarkBrown

    c or option traders would say all of them will be right where they are.

    going to be such a slow day today let's argue... ;)
     
  4. themickey

    themickey

    I'm attempting to calculate future probabilities by looking at different returns over different periods and was wondering, what is most significant, long term or short term historical returns.
    Maybe there is no answer to this?
     
  5. themickey

    themickey

    From my studies, the more volatile a stock the less the long term returns and vice versa. Perhaps volatility needs adding to the equation.
     
    MarkBrown likes this.
  6. Sekiyo

    Sekiyo

    That's a documented anomaly.
    Higher risk should mean higher return.
    But studies show it is not happening.

    The efficiency ratio is also an interesting measure.
     
    Last edited: Jul 11, 2025 at 9:24 AM
    Drawdown Addict and MarkBrown like this.
  7. MarkBrown

    MarkBrown

    so let's talk about vol, i have found every market has a sweet spot where it will travel but not spike - cause when it gets too excited there is danger and unpredictability. you want to be in at the start and out before the end of vol moving in an above average way.

    also i agree the efficiently ratio is a fantastic indicator - kaufman
     
    Sekiyo likes this.
  8. Sekiyo

    Sekiyo

    Another interesting approach (especially if automated) would be to chose an holding period (such as 3M or 63 days) then get the 63 days roc and run a 84/16 percentiles over the last 63x20 period.

    The higher the ratio, the better.

    Something like that.
     
    Last edited: Jul 11, 2025 at 9:49 AM
  9. 2rosy

    2rosy

    Not sure why it's an anomaly. It's arithmetic.

    Code:
    100*1.1*1.1 = 121.0
    100*1.07*1.13 = 120.90
    100*1.04*1.16 = 120.64
     
    Sekiyo likes this.
  10. Sekiyo

    Sekiyo

    Exactly what I thought.
    Log growth doesn't like volatility.
    Theorists don't think in logarithmic growth.