What is a Win Rate?

Discussion in 'Trading' started by oldtime, Nov 20, 2011.

  1. It should be relatively easy to compute the win rate of the New York Yankees. So armed with that information I should be able to make money betting on them or against them. All I need to do is overcome the vig, and that could easily be done with a little martingaling.

    What is a win rate anyhow? How do you compute it? If I bet on heads and the first ten flips are tails does that mean my win rate is 0%?

    Win rate over long periods of time may not do me that much good because I may not have enough money to bet over a long period of time, or I may simply not live that long.

    So what good is a win rate?
  2. ronblack


    What good is a pound? or a meter? or the area of a rectangle?

    If your win rate makes money it is a good win rate. If it doesn't, it is a bad win rate.
  3. I'd say 60% is a top level win rate.
  4. Win rate means nothing without looking at the risk to reward ratio with it.

    What good is a 90% win rate for 1 point es gains if your stop is 10 points.
  5. Win rate should be measured because it can be one input into the sizing equation which enables you to avoid ruin. Yes, there's an assumption in there about being able to meet the minimum bet-size requirements, but with today's markets, I'm pretty sure that minimum bet sizes are under $10 for some bucket shop derivative products, so almost anyone can stay "in the game" indefinitely.

    The win rate of 10 trials is relatively useless, though. As is commonly said in statistics, you need at least 30 trials to get something minimally-accurate. If you have hundreds of trials, even better.

    I've tracked hundreds of trades for my strategy and the overall win rate is about 62%. Even when I break the trade set down into groups of 100, I get win rates of about 53% to about 68%, with a relatively small standard deviation. Given that amount of data, I'm fairly confident that the 62% is a relatively accurate representation of my next few hundred trades because while each individual trade has a binary outcome, the overall set of trades is a continuous random variable. Obviously, only time will tell if this is true, but I see no a priori reason to think it will not be. If I really thought that there was a possibility that the win rate for my next 100 trades would be zero, I would simply stop trading. When you have that amount of data, the burden of proof, probability-wise, moves to those who say that the win rate is not 62%. Even saying that the true long-term win rate is less than 45% is to say that I've just happened to draw a sample that is ~8-sigmas away from the true population of trades using this specific strategy.

    Another factor in the going-forward win rate is the nature of the trading strategy which has generated your historical statistics. Let's say you have obtained a certain win rate on long trades on a stock because the company has a very capable CEO, thus favoring long traders. When the CEO retires, you would not be able to depend upon that factor, so your win rate would decline (e.g. GE after Welch retired and was replaced by Immelt). Or you're a long trader of index futures/ETFs in a bull market and the market then turns bearish. Alternatively, if your win rate is based on some deeper factor like [insert deeper factor here], then you'd expect it to remain more stable across market conditions.

    Overall, the longer I watch the market, the less random it appears, so I have relatively little problem with assuming that win rates for the right kind of strategy will remain fairly stable.
  6. Lucias


    See expectancy formula, size of win * prob of win - size of loss * prob of loss = e.

    I like higher win rate. Low win rate strategies are more unpredictable. I consider win rates of 65% to 80% good. Win rates can be high as 80%. Some strategies perform very well with win rates as low as 40% but I'm more suspect of these because they are so close to the edge.

    Size of stop does not determine true risk. That's an old silly idea from TA. The probability and the stop is hit determines the true risk. Taken at extreme, one can see why.. a too tight stop will almost always be hit -- making your loss of 100% very high.

    It is hard to maintain a win rate over 55% for most traders. The statistics of most trading strategies will change dramatically over time due to non stationary of market.
  7. oh, ok, so a tight target will almost always be hit making my profit of 100% very high, right?
  8. Lucias


    Not after account for commissions and slippage (for the losses)... It doesn't work the other way around. Wish it did.. your probability of loss will still be very high.. at about 100% without an edge. You will either have to take huge losses or not take a loss and blow your account out or use so little leverage that it would be not worthwhile.

    The target is not also the opposite of the stop. A lot of people think it is an inverse.. its not. It has different properties. Stops are more dangerous then targets. Targets work more like insurance but stops don't because insurance is suppose to protect you in event of loss. You can be right and still lose with a stop.

  9. This seems like an assumption, not something you could show empirically. Granted that each percentage point deviation from the average win rate over a large sample size is a larger percentage deviation, I don't know that in absolute percentage terms the deviation would necessarily be any larger for a low win rate strategy. There are probably numerous 40% win rate strategies that crank along between 38% and 42% winners year after year vs. 80% winner strategies which vary between 76% and 84%. Same amount of percentage variation in both.

    The usual argument for high win rates is psychological, not statistical. Once you overcome that psychological barrier of losing more times than you win, it's all just about expected returns.
  10. NoDoji


    As Volente said, an average win rate is determined within the framework of a risk:reward ratio, and as you point out above, the account size/time window are also important.

    In a fair coin toss (the coin is balanced and released the same way every time), let's say you've determined through extensive research that every 100 tosses, the win rate for either all heads bet or all tails bet averages 50%.

    To bet profitably, you need an edge. With a 50% win rate, you need a risk:reward edge. If you bet $1 on each toss, but win $2 above your initial bet for every appearance of heads, you have a significant edge and should end up quite profitable over time.

    If it takes 15 seconds for each toss to be completed, you have plenty of time to bet, but you only have $10 in your account, you can quickly lose everything despite the favorable edge.

    If you have $100 to bet but can only play the game for 5 or 10 minutes of each 25-minute/100-toss session, you can also lose everything despite the favorable edge.

    Successful trading involves a combination of factors, all of which need to be researched and incorporated into the business plan:

    Win rate + R:R of a setup or group of setups that produce a favorable edge over each time period X (sustainability of the above through varying market conditions - bull/bear, trend/range, orderly/volatile - may also be a key factor depending on the nature of the strategy)

    Capitalization necessary to continue doing business long enough to benefit from the favorable edge

    Ability to trade every appearance of the setup(s) and manage each trade according to plan so as not to dilute the favorable edge

    It's like any business, requiring market research, development of a viable business plan, proper capitalization, and adherence to the plan.

    Another key ingredient, often overlooked, is ongoing analysis of results and ongoing research to help recognize, early on, changes in the business environment that may require changes to the business plan.

    An increase in raw material costs will affect profit margins of a manufacturing business; the business will need to increase the selling price of its products to ensure profitability.

    An increase in market volatility will likely result in the need to increase the size of a trader's average stop loss. The trader who then takes advantage of this market volatility by increasing the average profit target accordingly has a much greater chance of remaining in business.
    #10     Nov 20, 2011