[Predictor]

I want to share a technique that can help you to choose your desired winning frequency before you enter into a trade. I would guess a larger number of retail traders don't lose because they are wrong but because they use stops improperly and lose money even when they're right.

Assume for these purposes that the market is mostly random, if true then that means that an X point move up or down has little value in determining what the market will do next because, of course, if it did have value then we could use that predict it. Now, it is true that I predict the market but I will agree that the market is mostly random to most people most of the time. This implies we can determine our win frequency before entering a trade simply by choosing a stop and target of the appropriate size. The expected profit must be zero.
The formula is:

WIN AMOUNT * WIN FREQUENCY – LOSS AMOUNT * LOSS FREQUENCY = 0

Examples:
+5 and -5 point target and stop will win 50% by random chance
+5 and -15 point target and stop will win 75% by random chance
+5 and -20 point target and stop will win 80% by random chance

Formula examples:
5*W-5*(1-W) = 0
5W – 5 + 5W = 0
10W = 5, W = ½, W = .5

5*W-20(1-W) =0
5W -20 + 20W =0
25W = 20
W = 20/25 = .8
If these are challenging, you can use the equation solver at
http://www.algebrahelp.com/calculators/equation/

You can choose any winning ratio that you desire using this method. I personally would feel quite comfortable with 65% to 80% win ratio. Of course, your profit expectation is still going to be zero unless you have an edge. My point is that many people may have a slight 2% to 5% edge but ruin it with poor stops. It is very easy to erode such a small edge with a poor stop. The goal is to use the mathematics of randomness to prevent that. While the expectation is the true measure of a system, if one chooses a system with a very low win rate then by random chance they could experience a very long streak of losers. Psychologically this will be difficult for many to work through. I have, also found that when win rates drop below around 55% that the resultant equity curves become rather unstable.

It is worth to consider the assumptions in which choosing a higher win rate would work better then not. If our model is such that we can predict that the market will move in a given direction more times then not then choosing a higher win rate should be beneficial. However, if our ability is more about predicting the larger movements then we may want to choose a lower risk and a higher reward payoff.

This is an updated authorized partial excerpt from My Favorite Edges Bonus. Chapter "Choose Your Win Rate".
----
Curtis

[eip]

Where can I get a copy of your book "My favorite edges bonus" ?
Ed

[SunTrader]

Originally Posted by Predictor »

[url]....... I have, also found that when win rates drop below around 55% that the resultant equity curves become rather unstable,,,,,,,,,.

Don't know why the author says this since I have always heard most large, successful traders win < 50% of the time.

[bobcollett]

Hi predictor
I like your post
And just to make sure I understand it............
In a random market,
To achieve an 80% win rate,
To win $5 , you need a stop of $20.
The risk looks a bit lop sided?
Correct?
regards
bobc

PS
If my understanding is correct, my next question is....... Who worked this out?

[Tams]

if you have to use an extra W-I-D-E stop loss in a day trade to improve your win rate,
you don't know what you are doing.

one of these days, the market will take your stop loss,
and you will give back 100 days of profit.

[mikewinfrey]

have you actually coded this as you describe with random entries? i just did and it doesn't work out as a profitable system or at least i don't see how it can. i used risk:reward ratios of 1:2, 1:1.5, 1:2,1:3 and 2:1, 2.5:1, 3:1. none of those were profitable. in fact, they were horribly unprofitable which is exactly what i figured they would be. Also, i used a renko brick that accounts for the full range that price goes to form the brick. using renko creates a uniformand bounded view of the market and as long i choose stops and target beyond the range of the brick, i don't have to worry about intrabar wiggel as opposed to other bar types. that is why i chose the renko for this exercise.

[neoikon]

In the example of 5pt target and -20pt stop, you're saying an 80% win rate.

So, for every 10 trades, 8 will work, 2 will fail.

This means:

8 wins x 5pts = 40 pts
2 losses x 20 pts = -40 pts

Net = 0pts

Still not making money, right? (losing, if you count commissions) This is why you need to have your reward greater than your risk... even if only slightly. This is how casinos make money, even if the odds are only slightly in the house's favor. You have to be the house.

[TheDude]

Originally Posted by neoikon »

In the example of 5pt target and -20pt stop, you're saying an 80% win rate.

So, for every 10 trades, 8 will work, 2 will fail.

This means:

8 wins x 5pts = 40 pts
2 losses x 20 pts = -40 pts

Net = 0pts

Still not making money, right? (losing, if you count commissions) This is why you need to have your reward greater than your risk... even if only slightly. This is how casinos make money, even if the odds are only slightly in the house's favor. You have to be the house.

True, but as I understand it, the thesis is that although gross expectancy is 0 in a 'random' market, your edge will slant the odds in your favour, or to put it another way, add a skew to the random outcomes.

Although I think Tams was correct in what he said about stop placement, but Predictor is on the right track I think in terms of a starting point - especially for someone going down the algo/mechanical route