Profit Factor

Discussion in 'Automated Trading' started by pwrtrdr, Dec 21, 2011.

  1. pwrtrdr


    Defined :

    Profit Factor -This value is the profit generated by profitable trades divided by the losses generated by losing trades. A value of 2 would indicate that twice as much money was made from winning trades than was lost from losing trades. Higher values indicate less risk.

    "Higher values indicate less risk"- not sure I agree. What about 7 losses in a row near the mean loss value ?

    Anyway, I am trying to discover a filter or means to determine why a method I have running with a profit factor of high 2.85 - 2.79 for 4 years in a row, now recently near 1. Obvious a cange in market. But exactly what. On its face it looks like Volatility, however what is best ,measure of volatility in futures ? Average True Range?

    If I look at ATR over the years it has shrunk or "come in" but as a percentage of absolute price, the ATR is still around 3 percent of notional value of the contract?

    Any input appreciated....
  2. well yes and no. No doubt about it the higher the value the less the risk, but not the volatility. Sounds like you are confusing risk with volatility. Two different things. ATR is the only thing I know of to measure volatility if you are trading just one market. I know of no way to measure risk.
  3. ronblack


    You have to first understand how the profit factor is related to other metrics that are directly affected by volatility.

    Read this first. According to the formula:

    pf = w x r /(1-w)

    Now measure your win rate and r before and after. If the win rate is the same, this means you have to increase r to maintain same profit factor, i.e. volatility is decreasing your r.

    If the win rate has decreased but r stayed the same then you are losing your edge probably.

    I hope this helps.
  4. Yep, you alluded to it, but didn't "nail it".
    The likely problem is you are getting stopped out for many small losses.
    That will quickly kill your Profit Factor.
  5. Thorough backtesting would have answered that immediately.
    If the backtesting had no periods of low PF, then your testing sample was inadequate....or as you said, the "market changed".