Forex Factory (https://www.forexfactory.com/forum.php)
-   Trading Systems (https://www.forexfactory.com/forumdisplay.php?f=71)
-   -   Open Range Break - 100% Mechanical (https://www.forexfactory.com/showthread.php?t=482654)

jmflukeiii May 3, 2014 10:35am | Post# 1

5 Attachment(s)
If I gave you a coin and every time you flipped heads, I would pay you $94.64, but every time you flipped tails, you would pay me $36.10, you would gladly accept the deal and flip that coin until your thumb fell off. What if I told you that I had modified the weight distribution and aerodynamics of the coin such that 65% of the time, it would land on tails. You could quickly rattle off the expectancy in your head as the brilliant statistician you are and come to the realization that, if you could maintain that payout ratio, you would continue to flip away. Perhaps, you would, if you could, build a robot to flip that coin for you? [Hint, hint].

In the index futures markets, I rely heavily on an open range that forms following the open of the RTH session, i.e. 9:30 EST. This is when liquidity pours into the market, volume is its highest, and the most people are trying to use the most money to accomplish something. I hadn't thought about it till yesterday, but why not see if there was any edge to that concept in the FX market. So I set about manually back testing the following strategy.

System

  1. Draw a box around the first 5 minute bar after the NY open, 8am EST (12pm GMT, I believe), clearly defining the highs and lows of the bar. This will be called the Opening Range (OR, for short).
  2. Wait for price to close on one side of the this box, i.e. close above the high of the first 5 minute bar or below the low of the first 5 minute bar.
  3. Between the hours of 8am EST and 11am EST, if price closes above the OR, enter a buy limit at the top of the OR + spread+.5 pip. If price closes below the OR, enter a sell limit at the bottom of the OR - .5 pip.
  4. Wait for one of the following to occur:

    1. Price CLOSES on the opposite side of the OR from your trade. Exit at market and repeat step 3.
    2. Exit for profit at a multiple of the OR.

      1. I tested EUR/USD only, for a 2x, 3x, 4x, 5x, and 10x multiple. As you increase the multiple, you decrease your win rate, but increase your overall net profit and per trade expectancy.
      2. Return for the first 4 months of the year for a 2x multiple was 36.0% risking 2% per trade, with a 51.05% win rate.
      3. Return for the first 4 months of the year for a 3x multiple was 57.3% risking 2% per trade, with a 43.36% win rate.
      4. Return for the first 4 months of the year for a 4x multiple was 79.8% risking 2% per trade, with a 37.06% win rate.
      5. Return for the first 4 months of the year for a 5x multiple was 150.5% risking 2% per trade, with a 35.66% win rate.
      6. Return for the first 4 months of the year for a 10x multiple was 158.1% risking 2% per trade, with a 27.27% win rate.

        1. These assumptions assume more than 50:1 leverage, but NO COMPOUNDING: position sizing was consistent throughout the period. Position sizing is explained below

    3. If price does not close on the opposite side of the OR nor reach your multiple target, you will close the trade at the 23:55 EST.

  5. The only other rules is to enter no more than 5 trades per day.

Position Sizing

  1. Position sizing is based on the size of the opening range. The math is as follows:

    1. ((Account Size)*(Risk %)) / (10 * OR Size)
    2. This creates consistent risk and reward between all trades with the one caveat as mentioned below.

      1. As you read above in the stop loss exit procedures, this method doesn't actually use a stop. It uses a close beyond the opposite side of the OR at market. Often price wicks through the OR taking out stops. We don't want to be one of those. It adds to the number of trades significantly and reduces the edge.
      2. By using a close beyond the OR as the stop, this obviously means that are risk is not limited solely to the size of the opening range as defined above. Our risk is greater. So theoretically, it would be wise to use a risk that is half a % point less than your normal comfort level.
      3. Most of the time, when a trade closes for a loss, it is just barely on the other side of the OR, within a pip or two. However, using a close beyond the opposite side of the OR for a stop, in some instances, forces you to hold the trade to a large loss. Nonetheless, this is how I performed the back test; perhaps there is a better way.

        1. In my testing period, the largest loss experienced was 52 pips. The average loss, however, was 8 pips, and the median loss was only 6.4 pips.

Notes

  1. I tested this only on NY open and only on EUR/USD. Perhaps using London open would be more profitable, particularly for GBP pairs. Perhaps it would be more profitable on larger range instruments. Perhaps it would be more profitable using some random candle - I don't know and don't have the time to back test all these permutations manually. However, the findings I came up with I think warranted further investigation, for a 100% mechanical, profitable method.
  2. I put the logistics of this method together in about 5 minutes and then started testing. There could be:

    1. better ways of managing trades - I will say that in my experience, adding trailing stops, or BE stops has ALWAYS decreased profit, decreased expectancy, but increase win rate. Perhaps closing half at 5x multiple and setting the rest to BE to close at the end of the day or at 10x...
    2. better ways of entering - sometimes price took off from the OR without any pullback and no chance to enter - these all went on to be big wins, but were missed as orders were all pending orders at the OR.
    3. As it sits, it appears to be profitable. I wonder if it is on other pairs? Building a basket of mechanically managed OR break trades each day could produce massive gains as shown above.

If anybody had any interest in pursuing studying this further, particularly with the development of an EA, I'd be very interested to aid in any way I could.
Click to Enlarge

Name: eurusdm5 5.png
Size: 36 KB
Click to Enlarge

Name: eurusdm5 4.png
Size: 47 KB
Click to Enlarge

Name: eurusdm5 3.png
Size: 50 KB
Click to Enlarge

Name: eurusdm5 2.png
Size: 63 KB
Click to Enlarge

Name: eurusdm5.png
Size: 48 KB


dkrock May 3, 2014 10:47am | Post# 2

If you are interested in doing more analysis, check into a technique called Shirley Hudson's London Close Trading Strategy.

jmflukeiii May 3, 2014 10:54am | Post# 3

If you are interested in doing more analysis, check into a technique called Shirley Hudson's London Close Trading Strategy.
I'm not terribly interested in doing more analysis of other methods, to be honest. I have my own methods of trading FX, but was just curious if what I apply to index futures was at all applicable to FX markets. It appears it is, and further, could be automated, which is my real interest.

Thanks for the info though, I have not heard of that strategy and am always curious in the ways people look at the market.

EDIT: The FF thread I found about that strategy says its a scam. And it claims 90% win rate, so I'd probably agree.

mario777 May 3, 2014 7:32pm | Post# 4

If you are interested in doing more analysis, check into a technique called Shirley Hudson's London Close Trading Strategy.
dkrock, do you have a winning strategy to share? It seems you are good at giving advice to thread starters even those with 2 year old thread

jmflukeiii May 3, 2014 8:08pm | Post# 5

{quote} dkrock, do you have a winning strategy to share? It seems you are good at giving advice to thread starters even those with 2 year old thread
To be frank, I am not interested in his or anyone's advice regarding a different strategy. I merely presented an idea here that seemed to demonstrate a positive, mechanical expectancy for further evaluation.

mario777 May 3, 2014 8:13pm | Post# 6

{quote} To be frank, I am not interested in his or anyone's advice regarding a different strategy. I merely presented an idea here that seemed to demonstrate a positive, mechanical expectancy for further evaluation.
Yes I dont like it either. Thats is why I asked him if he has something to share so he should start his own thread and do not invade your thread.

jmflukeiii May 3, 2014 8:22pm | Post# 7

{quote} Yes I dont like it either. Thats is why I asked him if he has something to share so he should start his own thread and do not invade your thread.
Oh, I apologize... sarcasm is easily lost, unfortunately, in writing. Or maybe I'm just dense

insider999 May 3, 2014 11:26pm | Post# 8

Hello,

I would like to know what is so special about the 8:00am EST 5 minute candle? I just don t get it. Why this one and not any other one?

Insider

jmflukeiii May 4, 2014 12:14am | Post# 9

Hello, I would like to know what is so special about the 8:00am EST 5 minute candle? I just don t get it. Why this one and not any other one? Insider
As I mentioned, the idea stemmed from my use of the 5 minute opening range in the US equity markets and the massive influx of liquidity at that time in the futures markets. I decided to look at the opening 5 minute range of the EU for the NY session which is 8:00am EST. Also, this is the time period I already trade, and further it is the overlap of the London and NY sessions. However, perhaps there is a better (i.e. high expectancy or larger gross profit) time such as the London open.

Thinking about it further, perhaps the choice of time is actually irrelevant. When you look at a daily chart, for example, you rarely have a day that is a pure doji: price typically moves one way or the other. One could perhaps make a range out of any 5 minute, or perhaps 15 minute bar, and trade the breakout pullback of either side of it, and just managing to a very high reward per unit of risk. However, using a market open seemed more logical than just a random bar, plus it has proved useful in the equity index futures markets.

Having this idea automated would greatly assist in the testing of:

  1. different management strategies, i.e. perhaps a trail stop of some kind
  2. different currency pairs, i.e. GU, AU, UJ, etc.
  3. different "opening" bars, i.e. London open, Asia open, a random time bar
  4. and even different "opening" time intervals, i.e. 5M, 15M, etc.


abokwaik May 4, 2014 5:58am | Post# 10

So I set about manually back testing the following strategy. System Draw a box around the first 5 minute bar after the NY open, 8am EST (12am GMT, I believe), clearly defining the highs and lows of the bar. This will be called the Opening Range (OR, for short). Wait for price to close on one side of the this box, i.e. close above the high of the first 5 minute bar or below the low of the first 5 minute bar. Between the hours of 8am EST and 11am EST, if price closes above the OR, enter a buy limit at the top of the OR + spread+.5 pip. If price closes...
I like the idea, and probably gonna code an EA for testing it.

But wouldn't be better to use Stop Orders instead of Limit Orders ?

jmflukeiii May 4, 2014 9:53am | Post# 11

{quote} I like the idea, and probably gonna code an EA for testing it. But wouldn't be better to use Stop Orders instead of Limit Orders ?
Thank you for the input - I assume you mean placing, for example, a buy stop a pip or half pip above the top of the opening range so that it will be triggered upon breaking. This would solve the problem of the swift breakouts that never retrace enough to fill your limit back at the opening range.

However, what I noticed in testing is that price can and often will wick out one side of the opening range, but not close in that direction. So then you will have been triggered into a trade that would have otherwise not been taken.

Further, let me provide you a scenario: say the first confirmed break (close outside the OR) is up, and you get long. Price goes up for a bit, but no where near your profit target. It then, in the course of a single 5 minute bar, spikes down through the opening range and then immediately reverses to close back above the OR. Per the rules I backtested, we would still be in the trade as there was no close to first, exit the long, and second, set a pending limit order at the bottom of the OR to go short. Using a stop order we would have both exited and shorted only to find that the bar we entered on would have never even taken us out of the initial long, causing a loss and causing us to re-enter the long.

Of course there are a number of permutations that can occur, but I saw the above scenario occur a lot in backtesting. And I see it even more in the futures markets - often this "opening range" is defended upon retests, but often not until taking out stops on the other side of it. Just my thoughts... perhaps two versions would be worthwhile for testing, one with limits, one with stops. I think the stops will ultimately provide too much whipsaw.

coolsnake May 4, 2014 10:14am | Post# 12

This method is very good and accurate for the testing. I am eager to see the EA for this strategy.

jmflukeiii May 4, 2014 10:17am | Post# 13

Some other stats, not sure if they are relevant or useful:

  1. If the first break of the OR was down, the most likely trade to be the big winner for the day was a short. Opposite for if the first OR break was up.
  2. 85.3% of the opening ranges were between the size of 3 and 6 pips.

    1. There were too few data points for large opening ranges, but I suspect that reducing target size (i.e. target a 2 or 3 multiple instead of 5 multiple) on larger opening ranges could improve expectancy

  3. 50.1% of entries reached a 2X multiple of the opening range. Unfortunately, due to using the close beyond the other side of the opening range as the exit for a loss, the win:loss ratio is barely above 1 (1.09) so that is verging on not being profitable.
  4. Only 16.1% of entries achieved a 10x multiple of the opening range. However, the 10x exit method was the most profitable because even the trade did not reach the 10x, the day would often end with price having drifted off somewhere between 5x and 10x, allowing for a large win nonetheless (remember trades are held to target or stop or closed at the end of the NY day (23:55 EST).

    1. Despite a 16.1% rate of reaching the target, the actual win rate was 27.27% due to the reason mentioned above. I personally think trading a 27% win rate would be rather frustrating, but the 5x multiple win rate wasn't too much better (35%). I guess having it automated would mean that it would just do its thing without having to worry too much about your win rate or psychological baggage that comes with losing so frequently.


jmflukeiii May 4, 2014 10:19am | Post# 14

This method is very good and accurate for the testing. I am eager to see the EA for this strategy.
Thanks for the comment.

Unfortunately, its not very accurate (<50% win rate), but it has mathematics on its side (i.e. R:R). I would hate to trade this manually.

Perhaps using some sort of HTF filter, like SuperTrend or a simple MA on say a 60M or 4HR chart, and only take the range breaks in that direction. Just thinking out loud.

jmflukeiii May 4, 2014 10:50am | Post# 15

1 Attachment(s)
{quote} I like the idea, and probably gonna code an EA for testing it. But wouldn't be better to use Stop Orders instead of Limit Orders ?
I happened to be just doing some review and came across this day in the UJ - first trade was simple, break down, pullback, hit 5x. Second trade took forever - but the point was, look at how many times it wicked through the open range only to finally hit target at the end. That is why I used a close on the opposite side of the OR as the exit, and why I didn't use stop orders to enter.
Click to Enlarge

Name: usdjpym5.png
Size: 58 KB

FerruFx May 4, 2014 11:07am | Post# 16

{quote} I happened to be just doing some review and came across this day in the UJ - first trade was simple, break down, pullback, hit 5x. Second trade took forever - but the point was, look at how many times it wicked through the open range only to finally hit target at the end. That is why I used a close on the opposite side of the OR as the exit, and why I didn't use stop orders to enter. {image}
In this particular case, why don't you stop trading after the 1st profit ?

jmflukeiii May 4, 2014 11:27am | Post# 17

{quote} In this particular case, why don't you stop trading after the 1st profit ?
That is a good question. I will run back through my back test numbers real quick to see if the overall expectancy improves by stopping for the day once achieving "the win", i.e. the big move away from the open. If I recall correctly, once you catch the big fish, that typically is it for the day. My entries were limited to the first three hours following the NY open, and following the big fish, you didn't often get a chance at getting another one in that three hour window.

jmflukeiii May 4, 2014 11:54am | Post# 18

The following outcomes occurred:

  1. 1 occasion of 3 wins out of 5 total trades entered from 8am - 11am EST
  2. 1 occasion of 2 wins out of 3 total trades entered from 8am - 11am EST
  3. 1 occasion of 2 wins out of 2 total trades entered from 8am - 11am EST
  4. 38.8% of wins occurred on the first trade, 34.7% occurred on the second trade, 14.3% on the third trade.
  5. It appears I accidentally picked a bit of a rarity for that above example!

Anecdotally, I think the time window and trade maximum might keep you from getting too chopped up, and the extra juice provided from high range/volatility days boosts overall return.


FerruFx May 4, 2014 12:07pm | Post# 19

The following outcomes occurred: 1 occasion of 3 wins out of 5 total trades entered from 8am - 11am EST 1 occasion of 2 wins out of 3 total trades entered from 8am - 11am EST 1 occasion of 2 wins out of 2 total trades entered from 8am - 11am EST 38.8% of wins occurred on the first trade, 34.7% occurred on the second trade, 14.3% on the third trade. It appears I accidentally picked a bit of a rarity for that above example! Anecdotally, I think the time window and trade maximum might keep you from getting too chopped up, and the extra juice provided...
We might have a greater expectancy by stopping after the 1st profit (not after the 1st trade !).

jmflukeiii May 4, 2014 12:10pm | Post# 20

May very well be true, especially over a larger sample size. Or perhaps you trade till the first profit around the London open, and then again the first profit around the New York open. Of course, you risk giving back a profit made in the London open come New York open.


© Forex Factory