Central Banks & Big Players

Hi rtrs111 et al,
Just for interest, my take on some of this. These conclusions are based on my backtests of various variations on Tony's two methods outlined in this thread. It is not science but is my overall view of results that I have obtained from my limited tests. I do not consider a test valid unless it is over more than six years if data and preferably ten.

The London/NY open method makes use of the volatility and extreme R/R ratio. How much better is the entry than a coin toss exercised at the same time? I do not know, but I would guess very little, if at all. Increasing the stop won't do much at all for this method as it is does not rely on being "right". However this is not a negative criticism it is leveraging some powerful circumstances to good effect. The input that an experienced trader can add is difficult to quantify but let me give an example. Imagine that it is Monday the 13th Sep 2010 and you are looking at your weekly chart. As often is the case we get 3 or 4 weeks of congestion after a good sized move. The market opens almost bang on support as can seen in the attached image. That would be a good risk to take an entry on the open method, don't you think? Well of course I cherry picked that from the chart, but I did so as an example of judgement. Such examples can turn what might look like a dog on a computer backtest into a stellar trading system.

The 8ish pip extreme bar is a completely different animal. This is a pattern trade and may benefit if the stop is increased as it has a tendency to be right that is greater than random. One has to quantify precisely what "follow through" etc means. I tried a period of seconds after the 8 seconds and defined what had to occur within that

Both of these methods can and should use tight breakeven stops, and simple trading rules to improve their potential. As an example of a rule. I want to be out over the weekend, so if I get to a certain time on Friday and I do not have my target pips than I will take 40% of the target and I am done.
It is not my intention to pass judgement on Tony's work but rather to give some insight into what may happen if you start tinkering with it.

note: The chart is from ProRealTime and may not coincide with London open at 08:00 BST.

@Tony

Do you watch a few hours after? I've only watched it for 10 minutes before!

thanks

Hi guys,
few thoughts about history testing. If you want to backtest Tony's method for real then forget about MT4. The spread plays crucial point for this method (2 pip SL including spread !!!) and don't forget that MT4 history tester doesn't support Ask prices. In other words it doesn't have a clue about historical spread. When you start to backtest it takes last known spread from last real-time tick and uses it as fixed spread for every single historical trade. Not even mentioning that if you have data from MT4 Download Center then you have very crappy data. It has a lot of gaps - missing ticks.

So if you want to make a solid backtest you need a very high quality price data including both Bid and Ask prices (to take spread factor which is essential in this method). Besides you need the spread to be smaller than 1.0 pip and possibility to trade on such platform.

Oktar,
absolutely, I forgot about the very small spread. Discretionary manual trade management makes backtesting useless. I still think there is great potential if entering on a higher TF: low volatility BO or as Salient described trendline bounces/BOs, even for example 3 level ZZ semafors, after a 5sec 8p candle. Trades should not be taken more than 1,2 /each setup, as Tony does to avoid losses if price is ranging. I guess the only solution then is to monitor/test the PA around one of these setups, but only for who has time to do this .

This method is based on very tight stops connected with strong moves possibility before they happen.Tony is using this on market opens, unexpected news (5-8 pip candles) and breakevens from s/r levels. I agree that last two need some experience or some time to catch good setups. However the last method - the one listed on 1st post at this time, I mean entering on London/NYC opens, I believe neither require any analysis nor time. The original idea is simple, just enter 1-2 pip after first tiny move after market open. And that's it. No technical analysis or watching PA. If it fails it fails, no reentering. I think it is connected with believes that big players sometimes put massive orders right after opening which results in big moves.

So i would say 5-8 pip candles hunting needs mostly time, breakevens from s/r needs some time and some experience and finally market opens needs discipline. It needs it because you have to sometimes wait 50-100 days for a winner and not missing a single market open. You can't have any feelings even when you have been trading for 60 days in a row without a win. It also needs a good platform with high speed execution and the lowest spread possible - preferably below 1.0 pip. O***a was quite good for that - 0.9p spread and no commission but unfortunately they have widened the spread. Tony: have you found anything new yet? I saw screenshots from some DMA platform, it was aroung 1.0p for gbp/jpy lol ! Imagine what is it for eurusd - 0.2p? This kind of stuff would be great for your method.

Some people suggested D***scopy. I have been thinking to give them a try as well. And here comes something great from this broker. They have history dealing feature on their platform the best I've ever seen (from the free ones). Why? Because they have data feed on demo accounts and on history exactly same as on their real accounts. Because they have good historical tester. Because they provide their own high quality data taken from their live price feed. And finally because their data includes both Bid and Ask prices (wow!). It means that we can backtest and the results will be exactly same as we would trade it live with them. I'm currently coding an Expert Advisor for their platform, I will give you results once I'm finished.

Would you be kind enough to share the jforex code once finished?

I'm trying this method out, but its hard to wait for the 5-8 pip candle... because I have to measure the length every time... anyone got a better idea how to solve that ?

1. On your 5 secs chart, Add Study - ATR (1).

2. Click Draw Trendline icon - Horizontal Trendline - drag the line to the specific pip threshold you want. You can also add more than one line and change their colors or styles -- click on a line to bring up a menu.

3. To save your settings, Tools - Save Current Profile or Tools - Profile Manager.

Currently 0.7 spread, even with a major holiday in the US...

^pure definition of success... no support no resistance nothing,... just pure success epic win

i still cant figure out how you got that performance... i was sitting tight on my chair gazing at the chart for that candle... but still didn't have the chance to jump right in ... where you've entered... :-/
you didn't trade the London open only do you ?

Hi,
Here comes first part of my D***scopy research on Tony's method (London/NYC opens). I decided to make accurate spread analysis which plays crucial role for this method as I mentioned previously because we need as many 0SL setups when it fails as possible. If we always hit -2p SL then even epic disco ball wins not gonna help us

So I've created an EA that looks for setups and measures what was the spread on exact moment of entry. Technically it waits for London/NYC open, then waits for 2 pip initial move and then in the moment of entry it captures the spread. I've checked 2009, 2010 and 2011. Here are the results:

2009:

0.5 42.25%
1.0 35.08%
1.5 16.08%
2.0 5.04%
2.5 0.78%
3.0 0.58%
4.0 0.19%

http://i56.tinypic.com/2nizfxx.jpg

AVERAGE: 0.949 pip

2010:

Total:
0.4 0.58%
0.5 35.47%
0.6 0.97%
0.7 2.91%
0.8 3.1%
0.9 1.55%
1.0 35.67%
1.1 2.71%
1.2 1.74%
1.3 0.39%
1.5 10.86%
1.6 0.19%
1.7 0.19%
1.8 0.39%
2.0 2.52%
2.3 0.19%
3.0 0.39%
6.0 0.19%

http://i55.tinypic.com/2quq07p.jpg

AVERAGE: 0.91pip

Note one important thing. By the end of 2010 they changed the way of spread calculation. Previously the spread value has been increasing/decreasing with 0.5 pip step. Now it seems to be more detailed. Let's separate these two methods to see if something got better.

Here are seperated results for 2010:

old method spread calculation:

0.5 42.83%
1.0 40.43%
1.5 13.4%
2.0 2.87%
3.0 0.48%

AVERAGE: 0.89pip

new method spread calculation:

0.4 3.61%
0.5 4.08%
0.6 5.1%
0.7 15.3%
0.8 16.33%
0.9 8.16%
1.0 15.3%
1.1 11.22%
1.2 9.18%
1.3 2.04%
1.6 1.02%
1.7 1.02%
1.8 2.04%
2.0 1.02%
2.3 1.02%
6.0 1.02%

AVERAGE: 0.997 pip

2011:

0.4 11.1%
0.5 4.2%
0.6 12.5%
0.7 6.94%
0.8 9.72%
0.9 18.05%
1.0 18.05%
1.1 11.1%
1.2 5.56%
1.3 2.78%

http://i51.tinypic.com/jj7l95.jpg

AVERAGE: 0.835 pip

So average spreads during our session is below 1.0pip. We can filter trades when spread is > 1.0 but we don't know how it would exactly influence on results. I'll find that out later.

Remember this is accurate only for D***ascopy and besides spread there is a commission for each order.

To be continued...

Is your EA for MT4?

Looking forward to test!

Hey oktar,

good job mate! I'll be following with interest..

best,
Indiana

Nice one! Too bad i'm only monitoring first 2 hours of London and NY But i'm sure patience will pay off sooner or later

epics.... I'm currently working on determining the most probable times for a 8 pip candle occurrence, with no news involved. I will make a heat graph of the best times for surveillance. It will be done with the help of my hundreds of charts I've gathered during the last years.

Tony,
That would really be appreciated.

Bruce

Hi,

Anyone know a way to find an accurate historical data on E/U for Backtesting on 5sec chart ?

On any plateform..

I will be happy to share results with everybody.

Wal

Hi,
I wrote a strategy for JForex and made a backtest with Dukas data..
Starting from 2010, January 1st till now - EURUSD instrument

The strategy is trying to spot all 8 pips candles on 5 sec timeframe
with spread < 1.5 . No news filter..

Attached a csv file with result. All times are GMT.

Unfortunately you will find duplicate candles... like this:

potential buy at time 8 Jan 2010 13:30:12 GMT
potential buy at time 8 Jan 2010 13:30:14 GMT
potential buy at time 8 Jan 2010 13:30:14 GMT

That does mean that you have only to take in consideration candle started
at 13:30:12 GMT . I'll have to improve the output format of my strategy.


Best,
chriz

Tony on holiday?