Shelley's multi-pair basket trader

[cytek]

Gary, I agree.

+1

[cytek]

Finally + 204 pips today running original Shelley. + 391 pips since yesterday. Will move to last version for Woh filter but just with TP and closing at NY close.

Cheers

Nice. What time are you opening the basket? (in GMT please, if you can).

[cytek]

The best time to enter the market in a trend is during a retracement. Following this logic, entering at a set hour without considering what state the market is in is for me too risky.

100% - I used this method today over 4 seperate entries to bag 225 pips. That excludes the last HUGE move of the day which I missed as I went to have dinner.

You can edit the code to force the EA to basket trade when you want it.
Find BasketSent in the code and change the line to:
extern bool BasketSent=true;//Set to false if a basket send fails, to force a retry.
.02

Thanks for that code jippo - made the change, recompiled, restarted Empty4 (twice), but cant get it to work for me... I'm on v1k. What version are you on?

[garyfritz]

The best time to enter the market in a trend is during a retracement. Following this logic, entering at a set hour without considering what state the market is in is for me too risky.

In general I'd agree, verb, but Shelley isn't doing typical basket trading. It's just "grab everything that appears to be trending, and manage the whole pile of them as one mega-position." There's no attempt to take advantage of correlation. It's not clear to me that you can enter based on retracements &etc because you don't know the components of the basket until you decide to enter.

So far Shelley seems to work well, which is actually fairly surprising to me considering how simple the entry rules are. Apparently TMA / WOH is a good way to find trenders, and apparently entering the basket at the right time of the day works well. (Though it doesn't seem crucial because e.g. cytek entered 4 baskets in a 6-hour period and all of them won. Maybe he just grabbed 4 little pieces out of an overall large move.)

If you chose your basket ahead of time, then it would be smart to look for a retracement or a reversal before entering. Where we're picking "anything that moves," I don't think that makes sense. Shelley may work well but it seems to me the best way to trade this is to run it 24x5, entering/exiting pairs when they meet the TMA/WOH parameters -- like Steve did in S+G -- but track each pair's profit/loss/TP/SL/etc separately. We'd have to test that to find out. I might knock out a TS test that can at least test that one pair at a time.

[spotdespot]

You know, I wonder if the basket approach is the right way to trade the S+G method. Shelley's original idea opened all trades at once, you tracked the whole basket profit, and you closed the whole thing at once when you hit the basket TP or SL. Makes sense.

But with S+G you enter AND exit trades at any time. Steve's code updates the basket TP/SL/BE/etc pips when the # of open trades changes, so that's good. But with independent pair entry/exit, I don't think the basket TP/SL/BE/etc makes much sense any more. E.g. you might have half a dozen pairs enter and exit while you're waiting to hit your basket TP, and those pairs' profits/losses aren't counted in the current PipsUpl. I think each pair's profit & exits have to be tracked separately.

And that probably shouldn't be crammed into Shelley. It's too different from the original behavior and trying to do both would be a mess.

I agree too - I think it is essentially a different EA with a very similar entry.

BTW Gary - with the settings you posted I think, for example with a 5 trade basket, your BE with be at 50 pips and your first jumping stop at 60 pips. (10 for the pips locked at BE + 2 * 25 jumping pips). I think the jumping stop needs to have a "starting point" (adaptive starting point for adaptive baskets) i.e. Jump starts at Number of trades * AdaptiveJumpStartPips (say 20 for example). So with your settings, for a 5 trade basket, then BE would be at 50 pips (+10 locked) and 1st jump would be at 100 pips - locking in 35pips. Obviously just made up numbers and you would probably want to adjust but I hope you see my point?

EDIT: Was trying to get kids to bed - meant to add:

The reason I think the above option is better is because it is very difficult to get an adaptive jumping stop value (in combination with jumping TP) that either gives enough room for the trade to breathe or allows the jumping stop to be effective beyond the first jump. The example I used, with your settings, the first jump occurs at 60 pips and sets the Sl to +35 pips - just 25 pips breathing room for the whole basket, if you set everything else the same and adjusted the adaptive jsl you would get:

for 6 it would jump at 70 pips with sl set to +40 - 30 pip breathing room
for 7 it would jump at 80 pips with sl set to +45 - 35 pip breathing room BUT if your adaptive TP is 25 then for the 5 basket trade your TP target would be 125 pips and the second jump would not occur until 105 pips.

[getting]

setup for 30May, considering of adding :05 to send time, but no for now.
http://i50.tinypic.com/2d5m34.png

[garyfritz]

I can't keep up with this thread!! And I'm supposed to be WORKING...

BTW Gary - with the settings you posted I think, for example with a 5 trade basket, your BE with be at 50 pips and your first jumping stop at 60 pips. ...

With these settings:

AdaptivePipsStopLoss=50
AdaptiveBreakEvenProfitPips=10
AdaptiveBreakEvenProfitPipsLock=2
AdaptiveJumpingStopPips=5
AdaptivePipsTakeProfit=30

Let's pretend we have only 1 open pair to make it easy. I think it does this:
* SL starts at -50 pips
* When you get to +10, set BE stop (basically the initial JSL) to +2
* When you get to +12, set JSL to +7; at +17 set JSL to +12; at +22 set JSL to +17; etc
* Exit with full TP at +30

Multiply all values by 5 if you have 5 open pairs. So you're right, BE at +50, +10 locked in, first jump at +60, setting JSL to +35.

The initial SL is probably too big but I didn't have a feel for the behavior. Probably 30 would be a better starting point.

[spotdespot]

I can't keep up with this thread!! And I'm supposed to be WORKING...

BTW Gary - with the settings you posted I think, for example with a 5 trade basket, your BE with be at 50 pips and your first jumping stop at 60 pips. ...

With these settings:

AdaptivePipsStopLoss=50
AdaptiveBreakEvenProfitPips=10
AdaptiveBreakEvenProfitPipsLock=2
AdaptiveJumpingStopPips=5
AdaptivePipsTakeProfit=30

Let's pretend we have only 1 open pair to make it easy. I think it does this:
* SL starts at -50 pips
* When you get to +10, set BE stop to +2
* When you get to +12, set JSL to +7; at +17 set JSL to +12; at +22 set JSL to +17; etc
* Exit with full TP at +30

Multiply all values by 5 if you have 5 open pairs. So you're right, BE at +50, +10 locked in, first jump at +60, setting JSL to +35.

The initial SL is probably too big but I didn't have a feel for the behavior. Probably 30 would be a better starting point.

Sorry - I edited my post above just to confuse things more I basically think it is very difficult to set an adaptive JSL that will work. <-- that was quicker to type than the edit

[mackaozy]

Another awesome day..If this continues I will be going live with this on my main account in a month.

[garyfritz]

spot, I think the JSL works OK, especially with the changes Steve made recently. It starts out at the AdaptiveBreakEvenProfitPipsLock level (which was why I said that's the initial JSL setting, but you replied before I made that edit), then it stays between 1* and 2*AdaptiveJumpingStopPips below the max PipsUpl. So with the settings I posted and 1 open pair, the JSL is always 5-10 pips behind the max open profit. 5-10 would be pretty tight for a single pair, but the aggregated basket position should in theory be less noisy. If 5-10 is too tight for you, use a larger AdaptiveJumpingStopPips, but as you say that means there are fewer jumps. With AdaptiveJumpingStopPips = 7 it would jump to 9 (for 1 open pair), 16, 23, and then you hit the TP at 30.