Best Thread Live Cable Trading

[Baruch]

Well, I know you know them. We all do. Central operating levels established by Frankfurt and London 'opens' (only talking Cable here). Basic Support & Resistance. Maybe a little Fib thrown in here and there (they work because they work when they work and when they don't they don't). Tick pressure (sorry, can't live without my proxy shot). Managing the trade based on price action as it develops - or doesn't. All basic, unflashy, no glitz, no glitter stuff.

OK, I agree - but you forgot trading the news (before and after release like marmoset also does). PS. I don't know anything about Fib og tick pressure.

 

[Baruch]

Something about the trading hours:

http://ezinearticles.com/?What-Are-The-Best-Hours-For-Forex-Trading?&id=95813

 

[Priceman]

A farewell from a famous cable daytrader:

http://soultraderdiary.blogspot.com/

What did you do to upset this guy then Baruch ?

 

[TheBramble]

OK, I agree - but you forgot trading the news (before and after release like marmoset also does).

That's a point where we differ dramatically. I don't trade news. They never did. They warned against it. That's what they said "The Crowd" thought they did.

PS. I don't know anything about Fib og tick pressure.

Be happy to expand on these unless it was a definite point of disinterest.

 

[Baruch]

What did you do to upset this guy then Baruch ?

I warned newbies against buying a system without a proven track record for 2300 GBP. And I still do.

 

[Baruch]

That's a point where we differ dramatically. I don't trade news. They never did. They warned against it. That's what they said "The Crowd" thought they did.

Be happy to expand on these unless it was a definite point of disinterest.

Any links to info about tick pressure? And how do you use this indicator?

 

[wasp]

Monday?

Hi,

On good Friday past the markets were flat as... Yet Easter Monday saw a 150 pip rise.

I take it if at least one (major) country is open for trading its worth trading but I think this Monday is a bank holiday here (Mayday) and something in America also? So would it be wise to assume as both are closed, its not going to be worth it tomorrow?

 

[Baruch]

Hi,

On good Friday past the markets were flat as... Yet Easter Monday saw a 150 pip rise.

I take it if at least one (major) country is open for trading its worth trading but I think this Monday is a bank holiday here (Mayday) and something in America also? So would it be wise to assume as both are closed, its not going to be worth it tomorrow?

Europe, Japan and rest of the world are open.

 

[wasp]

Well If US are not, it looks like all action should be early till 2ish then take the afternoon off!

Cheers Baruch

 

[Baruch]

Well If US are not, it looks like all action should be early till 2ish then take the afternoon off!

Cheers Baruch

Cheers wasp. I always (almost) trade the BO from the Japan range - I think it's one of best daily bets in this game. I suppose I will do it again tomorrow morning.

 

[marmoset]

Please don't anyone take this the wrong way, but...I checked the link Baruch provided (thanks Baruch) to Bunnygirl's thread and the 401 pages of posts on a simple system...

I think if you look at the subsequent posts to that one on this thread and you'd have a hard time not visualising a troop of monkeys scampering all over a banana tree.

As I say, I really don't mean offence and I'm not thinking of anyone individually, but this is all not nonsense exactly, it all works until it doesn't, but you're building on so many caveats and discretionary bits & bobs to handle those times it doesn't 'work' you've ended up with something which is largely incomprehensible, complex and therefore most likely, untradeable.

Agree, as I tried to backtest it in Excel, but it had so many ifs and whens and other conditions I got disgusted and binned it. BGX isn't for me.

What shakes me up more than a little is that this very topic, or one extremely close, will most likely come up again, either on this site or another one, in another year or 18 months.

Seems to be a common timeframe on a lot of different forums. I also post on a backpacker forum (well, I haven't lately, too busy working on method and trading) and the same old questions and topics pop up in that very regular timeframe.

Bunnygirl and everyone else who ever had a 'hot' system (JonnyT springs to mind) either fell foul of changing market dynamics (and they are always changing) or had to tweak and patch and cut and paste and glue and tie and bind until it just about hung together again, for a while longer.

That's why I prefer to work on learning different tools for different types of market behaviours. Like in fishing or golfing, you don't want to limit yourself to one rod/club, hooksize and bait. You bait the hook to suit the fish and you go where the fish is etc. etc.

The thing is, we all recognise that the simpler the system, the better, the easier the more tradeable the more profitable it is. And yet there seems to be this fascination with making it more complex. Well, actually it's not a fascination with increasingly complexity, it's more an unwillingness to let go of the time and effort already invested in 'finding' or building a system which worked (past tense) in the first place. Ego.

Truer words were never spoken. The KISS principle isn't very gratifying to the ego, but what do I care about my ego when I'm laughing/crying all the way to the bank?
But then again, I may as well plead guilty, as I've invested quite a bit of time in MonkeyBounce (heck, most of this weekend was spent writing a numbercruncher) and am not quite yet willing to let go of it. But then again, those crunched numbers do point to something interesting :cheesy:

The reason the tried & tested (anyone NOT bored yet?) systems that have stood the test of time work, is because there is nothing complex about them. They go to the root of the issue which is not fundamental market dynamics, nor commercial perspectives, but always personal psychological perspectives and how they are manipulated into mass market action.

Too beautiful a statement to comment.

I may well come back and delete this post later when I realise just how pointless you know it is now, it was.

Natch, leave it up. Sometimes we need to be reminded of the basics.

 

[marmoset]

Hi,

On good Friday past the markets were flat as... Yet Easter Monday saw a 150 pip rise.

I take it if at least one (major) country is open for trading its worth trading but I think this Monday is a bank holiday here (Mayday) and something in America also? So would it be wise to assume as both are closed, its not going to be worth it tomorrow?

Expecting a range trading day here. Trouble is, I have dayshift (8.30-18.00CET), so I can only trade the post-London Close zone, as I have to in bed by the time Tokyo opens (though I may try to catch pre-Big Ben). Friday's NFP also falls by the wayside, but I'll be keeping a keen eye on it (unless some suit at the treadmill decides to call a meeting again... )

 

[marmoset]

Europe, Japan and rest of the world are open.

USD/JPY may be the big mover and shaker then...

 

[marmoset]

Crunchy numbers

Well, on to what I came here for originally:

Wrote three apps to crunch the 10sec data posted on Gain's website: RawStripper, NumberCruncher and Evaluator.

Crunched the 10sec data for Week3 of May06, then filtered for 2:1 winners with limit/stoploss settings 1:1 and ranging from 10-49 (why? easier to code, me lazy monkey ) and price range 00-99 (obvious). Then filtered again for at least 10 hits (must be worthy to be counted as a MonkeyBounce level).

MonkeyBounce bit first: all 2:1 and up winners are either below 50 (long), or above 50 (short), nicely coinciding with the tweaked rules on post #6377 (page 638).

Odd bit: there's not a single 2:1 winner with at least 10 hits in the 19-34 range, neither long nor short, real no man's land.

Another MonkeyBounce bit: at 35, all of a sudden the 2:1 short winners start appearing, not a single 2:1 long winner above 18... L15 anyone?
Remember the L35 that I put on the bench pending verdict (post 6775, page 678), well, I'm afraid he's been let go.

I'm still evaluating data, to identify clusters with a 3:1 win/loss ratio... or what the hey, I'm adding a few lines of code to NumberCruncher, let the machine do the work

 

[marmoset]

Cluster news

Evaluator came up with following clusters that are hit 10x or more for a 3:1 win/loss ratio at minimum 10pips move +/- 4 pips spread:
bid 04=L08 SL80 T28bid24: +20pips -24pips, avgWin: 10x, avgLoss: 3x >> (200-72) = 128pips
bid 13=L18 SL79 T47bid43: +30pips -34pips, avgWin: 10x, avgLoss: 3x >> (300-102) = 198pips

S47 SL83bid79 Tbid12=T16: +31pips -30pips, avgWin: 11x, avgLoss: 3x >> (341-90) = 251pips
S66 SL96bid92 Tbid35=T39: +27pips -34pips, avgWin: 10x, avgLoss: 3x >> (270-102) = 168pips

Next step: write Simulator and plug in above levels, to test for overall performance and MaxDrawdown. Let's see how that +745 pips/week holds up

 

[TheBramble]

Any links to info about tick pressure? And how do you use this indicator?

I've only ever discussed it under that phrase on these boards. I don't know if I invented it or not. Certainly other folk have used the concept, if not coined the phrase.

It isn't an indicator. I use a DDE feed into a prop package which I use. Had to get someone to do a bit of jiggery-pokery (with VBA I think, but I'm not sure) to get it to fit. But It basically does no more than accumulate all the ticks during the selected timeframe and plots them exactly as you would see a standard volume bar. So for a 5 min chart, it'll accumulate all the ticks in the five minute period and plot the total on a dynamic basis.

 

[marmoset]

Simulated MonkeyBounce Totals (to be corrected)

Right, wrote Simulator (still needs some I/O stuff and stats tracking) and ran the levels from two posts up:

Looks like I have the spread not calculated correctly on the shorts
Anyway, manual gross P/L (still needs to be coded):
L1: 580-312 = 268
L2: 660-340 = 320
S1: 770-324 = 446 (to be corrected down)
S2: 682-364 = 318 (to be corrected down)
uncorrected total: +1352 :cheesy:
(see output file attached below)

Diff. totals with 2 posts up: the levels were calculated on the mean of the apparent clusters, so there's bound to be trades that got executed that didn't consistently hit the levels 10x or didn't make the 3:1 hurdle.

Now on to correct the spread error on the shorts, add MaxDD code, file i/o and finish Simulator and then simulate the other weeks of '06 with the MonkeyBounce levels.

 

[Baruch]

Right, wrote Simulator (still needs some I/O stuff and stats tracking) and ran the levels from two posts up:

Looks like I have the spread not calculated correctly on the shorts
Anyway, manual P/L (still needs to be coded):
L1: 580-312 = 268
L2: 660-340 = 320
S1: 770-324 = 446 (to be corrected down)
S2: 682-364 = 318 (to be corrected down)
uncorrected total: +1352 :cheesy:
(see output file attached below)

Diff. totals with 2 posts up: the levels were calculated on the mean of the apparent clusters, so there's bound to be trades that got executed that didn't consistently hit the levels 10x or didn't make the 3:1 hurdle.

Now on to correct the spread error on the shorts, add MaxDD code, file i/o and finish Simulator and then simulate the other weeks of '06 with the MonkeyBounce levels.

Hi marmoset,

I don't understand a word. I hope you one day will have the time to explain us stupid people, what all this means. Please tell about the method - the rules and so on.

 

[marmoset]

Hi marmoset,

I don't understand a word. I hope you one day will have the time to explain us stupid people, what all this means. Please tell about the method - the rules and so on.

Hi Baruch,

I've used the salami approach to check if there are levels that are more likely to be hit and give a halfway decent return at an acceptable risk (MaxDrawdown and win/loss ratio) level. If you don't know the salami approach, try stuffing one down your throat That's right, you'll choke!

So I've cut the problem down in slices (just like you do with a salami) and handled them one by one:
1) I have raw 10sec data in comma separated values (.csv) format, but with some useless data and rates going to 6 decimal places instead of 4. Right: wrote RawStripper to get rid of the useless stuff and format the rates to 4 decimal places.
2) Numbercrunching
2a) Why? Original MonkeyBounce rules were based on observation, not on empirical data. Observation is a good start, but it doesn't cut the (Swiss ) cheese. Time to get serious about MonkeyBounce, eh?!
2b) Why code a numbercruncher when I have Excel? Excel is slow, both in calculation and simulation (and does stuff behind your back like format fields to text, no matter how many times you reformat to Number )
2c) Wrote NumberCruncher to:
- run through all prices
- indicate how many times a price is hit
- open a virtual position at price x
- run position x against real price data with variable stoploss and limit, same-sized.
- report how many times stopped out and closed out (limit hit) per price level, stoploss and limit (or target) level
- save crunched data for further analysis
Time needed: 1 hour (Excel: 8-24 hours, very erratic)
3) Evaluation: wrote Evaluator, originally to identify the trades that are closed out twice as often as stopped out (win/loss ratio of 2:1), for a minimum of 14pips (10pips plus 4pips spread).
Works both ways, by the way, a long that gets stopped out twice as often as closed out is a viable candidate for a short :cheesy:
4) As I still got too much data, I narrowed it down with Evaluator to a win/loss ratio of 3:1 (3 times as often closed out as opposed to stopped out). Also, I wanted to see where the most likely profitable price levels were (either long or short, see end remark of 3), so I added the parameter of "at least ten times successful, long OR short". That gave me, out of the whole 80-something thousand price data a grand total of 110 trades.
5) Finally Excel: load the Evaluator data (3:1, 14pips, 10x or more homeruns) into Excel.
5a) sort by time to identify long runs (price going up, up, up or down, down, down) and count them as one trade to get neutral (as opposed to biased/skewed) data, avoiding double counts
5b) sort by price level, to see where, on average, the prices that got hit in 5) were located, either long or short.
5c) calculate averages/means for 5b) prices, same for stoploss and limit.
6a) Using 5c) data, I wrote Simulator (a simplified NumberCruncher), in order to simulate having traded all four entry levels:
- plug in entry long price L1, stoploss SL1, target T1
- plug in entry long price L2, stoploss SL2, target T2
- plug in entry short price S1, stoploss SLS1, target TS1
- plug in entry short price S2, stoploss SLS2, target TS2
6b) Using RawStripped data from 1), simulate above to be real entries (automated trading system) executed with very real data, takes 2 minutes (Excel: up to half an hour).
7) Look over results: or
7a) If : re-crunch, evaluate, simulate with less stringent parameters.
7b) If , test with other weeks: should results deteriorate: see 7a)
7c) If not, try something else
8) Extreme testing: do 6) and 7) with extreme weeks: straight up or straight down
8a) If results contradict existing levels, numbercrunch, evaluate, simulate till levels are found that work on extreme weeks
8b) Goto alternate sources to find a way of predicting extreme weeks

Voilà, that's all there is to it

And now I wanna see what Tokyo is up to

 

[marmoset]

Just looking at Tokyo open: might take a peek at 8500, but also one above pivot (1.8630, exactly 100 pips above Friday's low).

Oh, hey, Baruch,

Forgot you wanted the MonkeyBounce levels. There you go:
L04 SL80 T24: in other words: go long at 04, stoploss at 80, target 24
L13 SL79 T43: i.e. long at 13, stoploss at 79, target 43
S47 SL79 T12: that's short at 47, stoploss at 79, target 12
S66 SL92 T35: and finally: short at 66, stoploss at 92, target 35