A Real Trading Edge, Quantified: Trading and Stop Management

Beginner Questions
1

I took a poll here in the Babypips community for the most popular time frame that people use for trade execution. To my surprise it’s the daily time frame, which is great. But from my time on this board, I hear the neophyte traders screaming 2 things over and over, Can Someone share a profitable strategy? And Why do I keep getting stopped out so much? The broker must be hunting my stops! I am here to bring you a quantified result to help you get an understanding of what your 2 failures are in this area. I believe whole heartedly this will help struggling traders. It’s not an entry, an indicator, a signal provider, whatever Holy Grail seeker wants. It’s purely trade management.

Let’s get the base of this research out of the way here. I have used daily data of the EUR/USD pair from 1/1/1999 the first day the Euro was allowed to be an accounting currency. I am using 1 minute bars as the minimum granularity and combining them together to become a daily bar, so there are 5760 points of data per daily bar (OHLC * 1440). I created a random entry system with a 50/50 bias that will take trades on the daily bar and enter at the market either long or short. I used the 14 day average daily range as a baseline 122 pips, which is what it was when I did this calculation. This is using 1 lot as the baseline position size.

So look at the control, here is a 1:1 Reward Risk system with random entries, using 1 ADR for the Take profit and the stop loss.

You can see here that it is basically 50% profitable trades, but the Expectancy is negative. Over 1065 samples taken which is statistically significant (~3% error estimated by 1/sqrt(#Trades)), So obviously taking random 1:1 trades loses money, we don’t think that anyone would disagree that monkey dart board trading would be profitable, did we?

Let’s go for a 2:1 Reward Risk ratio using 1 ADR for the stop loss and 2 ADR for the profit target.

Ok so our expectancy is worse than before, this is because our win rate goes down. That would be expected as we are pulling our profit targets farther away. If you have looked at my other studies in the Statistics and Technical analysis thread you know that this market is basically neutral, 50/50 chance to go up or down. So this is to be expected,

Continuing on to the 3:1 RR using 1 ADR stop 3 ADR profit target.

Wow ok now we have [B]POSTIVE EXPECTANCY[/B]. Oh boy, look we increased our Risk Reward by 1 R but our win rate only dropped a fraction of what it did from 1:1 to 2:1. Is this just a fluke?

4:1 RR using 1 ADR stop and 4 ADR profit target.

Nope not a fluke, 4:1 is also profitable. Ever wonder why people say you should cut your losers short and let your winners run?

So we have taken a random entry system profitable, this isolates the exits solely for construction of this experiment. We have taken the ability for trade management only using fixed stops, that are executed [B]CONSITENTLY[/B]. Not tampering not adjusting in any way purely set at the beginning of the trade. This is enough to give someone out there who can’t control their own consistency, playing with their trade exits. Also traders out there who are not getting good Reward Risk Ratios. Look, every system can be profitable, it’s all about the win rate to reward risk ratio. I am not saying high win rate low RR systems don’t work; I know they work because I trade them. But they require higher levels of consistency, which are what all novices do not have.

2

Now you’re going to ask me, well I do have 3:1 or 4:1 RRs but I get stopped out a lot. Your results are BS.

Here is measure of the intermittent swings of the daily time frame. Basically measuring the average swings (close to close), so how does this help the novice trader? They are still using 4:1 but still losing. Why? Probably because you’re placing your stop in the low end of this spectrum, right into the noise of the market.

Let’s run the experiment again but place the stop inside the noise range.

Here is a 3:1 Trade using a 50 pip stop and a 150 pip target.

This is an extreme example trading 50 pips on a daily time frame, but you’re forcing yourself to get stopped out. So yes you have a 3:1 RR but you’re not going to be profitable because your well within the underlying noise of the market. It is not a broker hunting your stop, just the random noise swinging on this time frame is large enough to take you out a high percentage of the time. This is before you can even have a chance to reach your target.

Here’s the same 3:1 pip ratio but we expanded the stop to 150 and the target to 450.

Oh wait, we were profitable. This is using the SAME entry system, using the same exit ratio. Nothing on the underlying structure of the system changed. Just by removing our stops and targets OUTSIDE the random noise of the market. Giving our trades a chance to work and enough room to breathe.

[B][I][U] BOTTOM LINE: High Reward Risk Ratios and keeping your stops outside of the noise of the market are key components to a winning trading system. The entry system does NOT matter; random entries can be made profitable. If I can make a monkey dart board system profitable, why can’t you be?[/U][/I][/B]

3

Some solid stats here, and a nice comparison between varying RR’s.

Keep up the good work, it’s not often we get to see the proof behind talk here.

4

Outstanding.
It is gratifying to see quantitfied what I have had to learn the hard way. In years of writing automated strategys, and contrary to convential wisdom, I have not had any that had improved results with tight stop losses set. After optimizing all indicator variables and in various time frames to make the most profitable system, tight stops [U]always[/U] cratered profitability.
Stops have their place to prevent catastrophic loss due to price shocks from adverse news events, but they must be set wide for optimal results. After learning this through trial an error its great to see a research that clarifys my experiences.

Thanks for this, MeiHua.
In fact it has given me a new trade idea to test and impliment. I’ll let you know how it comes out.

5

Valuable info here. Please continue this thread. Thanks for your effort.

6

Nice contribution. This is one of things that I’ve noticed but never had the capabilities to test. It seems that this is a very cool thing is that this applies to all time frames yes? Shooting for high R:Rs like 4:1 or 5:1 also allow you to have a much lower win rate. Of course this takes some good mental strength to be willing to take a trade setup after losing say 7 or 8 trades in a row, knowing that you just need 2 or 3 good trades to be up. You can afford to be much more conservative as well, which can hopefully boost your win rate a bit

7

Hey MeiHua, this is very interesting. How are you defining the swings? Is it close-to-close of each daily bar?

Edit: Also by random entry do you mean every time a SL or TP was hit, a new entry was executed immediately in a random direction?

8

So I can enter the market at random with a 200 pips stop and be ok? And if I have a slight edge, say entering in the direction of the daily trend, money?

9

Yes i was wondering about that too. 200 pips is most probably more than the ATR for the eurusd for the daily.

Alsohow about the hourly, if one were to use a 1 times ATR for stops on the hourly with 3 times TP, would it yield similar results?

10

Swings are defined by a zig zag indicator measuring the close to close swings.

Regarding the entry timing there was a pause between the close of 1 trade and the open of another, 1 daily bar as it enters with a market order after the close of a bar. This is just when the program was running is rangen to come up with the new trade. But yes basically thats how it works.

edit: to be honest, the zig zag using close to close measurements is not the most accurate. You would need something that used the high of the bar on swing highs and the low of the bar on swing lows. But that was indeed too complicated for me and i just wasn’t going to spend the time. I think this estimation proved the point just as well, though is subject to larger error.

11

Theoretically if you maintain high RRs you will have positive expectancy. As long as you were 100% consistent no matter how bad the losing streak or draw down. I think the estimate above is conservative a 15 trade losing streak is not inconceivable with these low win rates. This is not trading advice. You can take this knowledge and do what ever you want with it.

i used ADR (average daily range) not ATR which is totally different. Without doing the specific research I can not be certain. Again theoretically, assuming markets are fractal. Which I have proven to some degree on the statistics in technical analysis thread. Then it should translate. There is a big caveat, the smaller time frame you use the more noise. So that will have to be taken into account. It won’t be 100% transferable without modification but is a good baseline.

12

Okay, thanks for the explanation.

So may i ask wahts the difference between ADR and ATR?

ATR on the daily, isn’t it ADR?

13

Not how I calculate it, ADR is calculated as the average high over X lookback - average low over x lookback. ATR is the highest high of x lookback - lowest low of x lookback.

14

Thanks for clearing that up. Can you recommend any books for a noob to get started with statistical analysis of financial markets? How about what sort of tools are need or can be used?

Thanks again

15

Not really for newbies, but Tsay - Analysis of Financial time series. That’s a good starter for modeling and basic quant. There really isn’t any books out there that do what I am doing currently. I basically use my knowledge of markets then code up something to extract the data I need to give me the answer, think intuitively about what you question you want to ask the markets then find the data that will help reveal it. But don’t torture, the data too much, it will tell you whatever you want to know.

16

Thanks MeiHua

You have added credence to all the comments I have ever made about stop losses.

I see people getting stopped out too many times and i try to suggest that they should widen the stop loss. I’ve found it hard to get anyone to take my comments seriously. When I enter a trade I make sure I know which way the trend is going and I try to enter wisely. Then I trust in the trend without any stop losses. I only use stop losses when I decide to close the order and take profit. I find the stop losses can let the trade carry on and get me more profits than if i just closed instantly. So people should think about using stop losses to their advantage. Thanks MeiHua

17

Thanks, I’ll check it out. It seems like I should also look into some basic college statistics books as I’m kind of rusty on that.

18

I’m bumping this back up.

In amongst all the general silliness and dross here lately, this slipped away very quickly. But it is just about the most interesting and valuable piece of research that I’ve seen posted here.

19

Thanks for the bump and likes. You know I was posting in the statistics in the technical analysis thread my ideas and research. I guess this could just be repository of my own personal quantitative stats. I don’t know how many people are interested in it. The fact of the matter is every time I do something like this it falls away. I always figure its just not worth the time. For me doing this type of homework is what not only addresses the psychological issues of trading ( i know exactly the edge I am working with and its probabilities) but also helps me isolate what works and what doesn’t in a more mechanical fashion ( i know if something is worth pursuing or not based on its expectancy). If there is enough interest I will continue to post updates on my thoughts and research. IMHO the trader with the best homework wins, the trader who seeks unquantified grails loses or the lazy trader loses.

20

Great thread MeiHua :35:. This forum needs more such insights and less of the usual clutter.