Probability of Next Trade

Discussion in 'Strategy Development' started by bearmountain, May 19, 2011.

  1. Unfortunatly statistics is not my strong suit. I have a question about probability of win/loss of the next trade after a loss or losing trade.

    A system is profitable 65% of the time.

    When I drill down to the Losing series of trades (consecutive losing trades). 90% of the time, after 2 losing trades there is a winning trade. About 75% of the time there is a winning trade after one losing trade.

    So my question is, what about just taking a trade after waiting for one or two losing trades?

    So trade one profit, trade two profit, trade three loss.
    Why not just wait to take a trade after trade three?

    In black jack when the card count is high, the odds favor a player to bet big. Isn't the same true here?

    I am sure I am not the first person to think of this, what are the issues involved with such a strategy?

    Thans very much.
     
  2. Interesting thoughts, but there is zero correlation.
     
  3. zero correlation? are you thinking of the coin toss example.

    does that apply to the mkts. please explain. thanks.
     
  4. Good question and an important one. There are two answers:

    1. assume that your trades are completely independent with a 65% chance of winning. this means that every trade will have a 65% chance of win and 35% chance of losing. (assuming no b/e.) if this were really true then:

    the chances of you having N losing trades in a row is .35^N. so you have a 12.25% chance of having 2 losers in a row, 4.3% of 3 in a row, etc. however, this is a little counterintuitive, the chance of your next trade being a winner is ALWAYS 65%, even after 2 or 10 losers in a row.

    2. in the real world sometimes trading strategies are "streakier" than you'd expect under theoretical assumptions. it's dangerous to apply simple theory like #1 to the market sometimes.

    i cover this in a book i'm writing... bottom line is the answer to your question is probably no, it probably does not make sense to do what you suggest. the black jack analogy is not valid because the probabilities change as cards come out of the deck... does not apply to the market.

     
  5. BEAR MOUNTAIN: A system is profitable 65% of the time.
    ----------------------------------------------
    STOP RIGHT THERE.

    I trade two systems. One has an 88% win rate and the other is 92%

    Find an edge and then start trading.
     
  6. Ditto. I'm at about a 45% win rate, with a risk reward which makes me profitable.

    Today it was a 75% win rate. Yesterday, more like 20%. The last trade has nothing to do with the next.

    If you wait for a couple of losers before putting on the next trade, you will be giving up opportunity cost by not being involved. If you have a good system, trade the system.

    Losses suck, but they are part of the ebb and flow of winning.
     
  7. If all the figures come form the same backtesting, including the 65% probability for each trade, wouldn't the derivative figures be at least in the ballpark over time, unless the whole thing falls apart altogether?
     
  8. I would ask the OP the simple question which strategy tests out best (assuming they are all profitable) .... 1) take every trade, 2) trade only after 1 "loss", 3) after 2 "losses", ...etc.
     
  9. good post AdamG_SMB and jnbadger. thanks.

    I am not arguing with you, would like to explore this a little further.

    I read, John Henry the famous trend following fund manager, who bought the Boston Red Sox from some pocket change he made in the markets. He said that when his funds are down or have a drawdown of about 40%, he will add his own money to them. so adding fresh money to his fund saying that the performace will rise from being down 40%.

    Along the same line, now this is going back some years, Richard Dennis's Turtle method had a curious rule. After a large profitable trade, he would skip the next trade.

    I believe both John Henry and Richard Dennis are basing this on some sort of 'Trade Series Analysis'

    I believe systems are not designed in a vacum, they are based on capturing some sort of mkt/human behaviour.

    For example my little system is a intra-day trend following system.
    If the trend has turned up and system after a long trade takes a short trade, which most likely will be a loser. So its shorting a pullback within the trend.

    I believe 'trade series analysis' forces one to be aware of what the system interaction with the mkt, rather than just getting lost in the number.

    jnbadger you point of loss of mkt opportunity is also very important. I don't know how to stricke a balance...
     
  10. bear I like your way of thinking. I have several strategies that use the "wait for failure" concept. It's all based on results from backtesting.

    Is there some reason you don't feel comfortable with the backtested results of whatever "wait" strategy you're looking at?
     
    #10     May 20, 2011