Adamus 

Every day almost 99% of the time the EUR/USD turns into the twilight zone when the NY session starts to wind down. It shuffles backwards and forwards making a whole ton of H1s and H2s (or Ls) right through until about 6:00AM London time.

After saying that based on sim trading May to July 2010, I'm now seeing some days where the market is alive in the Asian session as though it was London or New York. Maybe this is just the August holiday season effect.

My year of sim trading May '10 to April '11 has got as far as Sept.

I'm sim trading the EUR/USD using solely the Al Brooks H1 and L1 pattern - so nothing else at all, no MA, nothing. This gives about 3, 4 or 5 signals per entry, and about 5 to 10 entries per day. Many of the signals, perhaps half, come during the quiet, choppy Asian session.

I've got a few points here I want to make so I remember them better and so anyone who wants can add their 2 cents.

I'm not making such good progress through the Al Brooks book. The "Price Action" chapter is so concentrated, it's like eating an elephant. I have picked up a few things, but I'll have to write them up seperately to drill them into my memory. The one thing that is very relevant to using H1s and L1s is the microtrendline breakout failure. I didn't read it in the book so far but I picked it out of one of his articles.

Here's a chart displaying a microtrendline break failure. It's like a corollary of an H1 or an L1, but you have to watch the microtrendline at the other end of the bar - i.e. if you are watching out for L1s in a bear trend, then you're watching the lows for a higher low, after which the next lower low will be your L1. But the microtrendline though is drawn downwards on the highs, so you're watching for a break above it which might fail.

H1s and L1s can appear after only two bars - think of a massive bear trend bar followed by a small inside bar just above its low, followed by the L1 with its lower low.

However drawing a microtrendline generally requires a few more bars, so microtrendline break failures will generally appear less frequently. I haven't practiced spotting them but I assume they will appear a lot less often.

Essentially the microtrendline break failure results in an H1 or an L1 - and using it as a signal will result in an entry on the H1 or L1 bar, i.e. earlier.

I could use this as a positive filter to cut out the majority of H1s and L1s.

More miscellaneous points:

- it took a while to figure out that an H1 or an L1 are quickly invalidated but if not invalidated, they can stand as good signals regardless of the number of bars between. Having said that, I rarely see more than 5 or 10 bars in between and it is basically only a consistent, well formed rally of trend bars that will leave an H1 or L1 hanging. Otherwise they are quickly superceded by the next H1/L1 or they result in an entry.

- I often wondered whether I could take an entry signal that occurs just before the previous trade exits. It was simple logic really which says the signal can be valid still and result in a re-entry, but it's rare.

- when it comes to determining the trend direction, within my time frame at least, time is irrelevant! Trying to base a decision on the last 2 hours or 4 hours or 12 hours never really results in a decent decision compared to the decision I can make just looking at the number of ticks that the market has moved. I'll need to revise my little "which way is the trend" flow chart.

- runners, i.e. a second lot left to run or hit break-even can be a big winner if I simply exit the runner at the end of the bar. Some of these EUR/USD bars are huge and my 15 point target is puny in comparison. One bar I saw was 80 points. In 5 mins. More testing further down the line. But really what is the downside to using a runner like that? Transaction costs, slippage and a bit more effort managing the trades.

- sometimes having 15 points targets and stops seems real Captain Caveman. I don't know which measure it would be best to keep my eye on, but it's worth thinking about adapting the 15 points to something bigger when some measure of ATR indicates so - I don't know which - daily, hourly, 5 mins?

Adamus 

- I often wondered whether I could take an entry signal that occurs just before the previous trade exits. It was simple logic really which says the signal can be valid still and result in a re-entry, but it's rare.

Just to qualify this, it's only entry signals in the opposite direction that can still be valid because an H1 or L1 entry signal when you are in a trade already will have passed while you are in the trade - if that sounds like I'm stating the obvious, I am but the decision logic is slightly different when you are trading real time. At the moment I am sim trading past data so I am not watching the bars form, I see each bar as a finished article bam bam bam so I made the mistake of thinking I would take a signal after exiting, although in real time I wouldn't get the chance. Clear as mud? My pleasure. I don't know how to clarify that more.

Maybe the chart shows it better. The last L1 signal can't be taken while still in the trade, so you have to wait for the next signal. Sim trading shenanigans.

I'm refining my set of rules on trend direction like a primitive flow chart. This is my current best analysis. Go through the questions until you answer "yes", or you reach the end, meaning there is no trend.

(1) is there a series of consecutive trend bars trailblazing the trend up and down the screen?

(2) is there an obvious series of swings, higher highs and higher lows (or lower highs and lower lows)? For now I'll be content to take the swings in the direction of the trend, not the retraces.

(3) from the highest high or lowest low on the chart (I see about 20 hours on the 5 min bars but that's a random choice which I need to justify), has it reversed from there by a number of points that you'd say constituted a new trend direction on your time scale?

(4) with the chart set to auto-scale, if the price in the top half of the screen or the bottom half of the screen, does it look like it's trended to where it is from the other half?

(5) are you sure that there's no horizontal channel, because if there isn't, there must be a trend moving the market beyond the possible horizontal restrictions.

(6) failing all that, it looks like the market really is going nowhere

I finally got too frustrated with my 'rules' for trend definition. It's the point where the trend evaporates and whether another has formed that is just pure quicksand.

I figure I'm not making it easy for myself because I'm demanding some trend definition - either up or down - without allowing for a sideways market.

This probably stems from my lack of patience. If there's no trend then the trend-based setups I'm looking for (H1/L1s) won't appear. It's a subconscious belief that I need as many trades as possible, a belief I'm struggling to replace with the idea that I should sit back and wait for the good ones.

I just tried sim trading a week of EUR/USD using HH/LLs to define the trend. I'd mark up the HHs and HLs in a bull trend, taking H1 signals. If an LL or LH appeared, I would suspend the signals, and if a second then appeared, I'd flip into a bear trend and start taking L1s.

And vice-versa.

The only way to do that without going nuts is to take a HH and then wait for the HL, ignoring the highs unless it put in a LH. And then wait for another HH.

It's not perfectly objective but it could be - there are some situations where you have to make your mind up on the fly or invent a few little extra rules.

The result is that the defined trends are a lot lot shorter than before, and there are a lot less signals - but it doesn't look like the signals are any better than before.

Chart attached. Green up triangles are HH/HLs, red down triangles are LL/LHs. Green dots are H1s, red diamonds are L1s.

So the end result is, yes it solves the frustration with the subjective decisions about trend direction, and I could probably make it into an indicator because of its objective nature, but it doesn't seem right as a tool for price action trading - it feels like a scatter gun approach and it often feels too random. If I was going to use it, I think I'd switch to range bars (see BM's vid on the 'Elusive' thread) - the charts would look a lot prettier.

That post above is all just part of my continuing Year of H1/L1s and the week of using HH/LLs proved no good, but now I'm not moving back to freestyle trend definition as I've done for 2010-05-01 to 2010-10-22 - I'm going to drop in the 20EMA onto my chart and see how that goes - at least for a week, perhaps for the rest of the test - thanks to Mr @ Bacon for the timely nudge in that direction on the 'Elusive' thread.

It would destroy my first six months' decision matrix, but that was a degenerate tool where I was only marking wins or losses - since I was correcting my mistakes as I went, so there was no 'Avoided' sector, and with H1/L1s there's no 'Misshapen' sector.

For the record, the first six months ended 910 wins to 880 losses, i.e. scarcely discernable profit - although that is irrelevant, this is just a baseline for future testing. I expect with the 20EMA this might improve a little, but I'm not going back over the first six months again, I'm just ploughing on.

My hope is that the 20EMA will remove some of the frustrating subjectivity around whether we're trending or not.

The rules around how to use it:

(1) market above the EMA and not touching - bull trend
(2) market below and not trending - bear
(3) market touching - sideways

I suspect I'll start getting frustrated by the lag in the EMA suspecting the trend is changing when the market and the EMA are converging, but hopefully the testing will give me a good handle on it.

Using the 20 EMA as the trend definition is not the piece of cake I thought it would be.

I've only done 2 weeks so far but the win:loss ratio is really bad. I don't know if there's something different about these weeks. There is more volatility, maybe that's the issue. 15 points both risk and reward doesn't look very clever at the moment.

A lot less signals, which is expected since I introduced the "no trend" state for bars over the 20EMA.

I'm still ploughing through 2010/11 sim trading H1s and L1s although it's going quicker now. It's taking about an hour per week and I've just got to Christmas so there's still a way to go.

From September to December the Asian session overnight came alive, often looking just like a London or NY session. This is in stark contrast to May to August when the Asian session was 99% range-bound in a little range until the London session opened.

The Asian session still does go dead though, creating quiet periods with poor signal strength and presumably purely random results - not something I can verify right now.

I picked up somewhere that the Asian session's range can create useful S+R levels for the rest of the day but I haven't looked into it further - something to put on the to-do list. There seems to be a fairly frequent big move at 6 to 7 AM as London opens - most of which are fake it in one direction and turn around after 30 mins and go all the way back and more.

I'm also seeing dead periods in the middle of the London session which surprised me. The signals in the dead periods increase in frequency and I guess (again) are completely non-predictive.

I also need to figure out how to spot range days early, since the signals should be faded on those days. Generally speaking though I only see them after it's mostly past.

Fortunately I think I now know a good or a bad H1 / L1 when I see it. The bad ones are the one bar affairs when you don't even get to anticipate it, it just happens. I'll post some charts hopefully to demonstrate but I have to remember to save the examples when I see them.

OK here's a chart attached - look at the last bar - it's technically an L1. It doesn't look good, there's hardly a retrace to speak of and I'm beginning to think I'll avoid them when they don't have a HH as well as a HL. (It turns out that signal is never triggered anyway).

However on the bull trend several bars beforehand, there's a pretty much perfect H1, with retrace, a LH and a LL on the bar preceeding the H1. I don't think I'd look for anything better - apart from more profit perhaps.

Basically for a good H1, you need to see a bar with a LH and a LL, not just a LH as Brooks defines it. Plus they seem to be better if they're good trend bars in the right direction. This is something for testing though, I'm only assuming here.

These are all things I'm going to use when running over the whole year again after I'm through with this first run. That, and changing the exits to Brooks-style stops 1 tick below the signal bar, and then also using a runner as well as a target.

I've got my work lined up nicely.

Hello,

this looks interesting. I'm simming the 6E market like you. I mostly trade the setups of Lance Beggs on the 3 min chart.
One point I found it is very important to recognize a trade which is going to fail.

My rule ( which is a subjective rule) is that on the 3 min chart, my trade has to move some points in 3 1 min bars ( or ofcourse 1 3 minute bar) , or else I will scratch the trade immediately. Let's say within 4 1minute bars you should be able to move your SL to entry.

This is because when this happens 8/9 out of 10 times the market will not move in my expected direction.

This approach saves a lot little losses, because if the trade will not go my way quikly, I can scratch the trade with -1/-2 pip loss per contract, instead of losing e.g. -7 pips per contract. So 1) my losses are smaller and 2) losing is less demorilizing
The 1/2 times it does run after I scratched I don't care. That was not my trade. It is important not to eat your hat when this happens