Trendy Gary

[fxozgirl]

nice looking results fxozgirl...which settings are you using? thanks

I have tried TG for about 1 month now with about 10 day of break, and I only have had buys as well.
Very strange.

here is my platform with trades open right now...mix of buys & sells

opentrades_buys_sells.gif

pretty sure it's just OOTB

[fxozgirl]

This week, still only buys.
Testing this for more than a month, and still no sells. I am using it on Alpari UK.

A question to the inventor:
@garyfritz: Do have any explanation for this strange behaviour ?

Guys, I don't think it's the EA...my demo has taken 101 trades this week (50 buys & 51 sells)

There must be something in your settings that is preventing the sell trades

[AGT]

Ok, here we go - I posted all I use (incl. setfile) in the zip folder at jan 09 ... hmmm ...

[garyfritz]

Sorry, Steve's the "inventor" of this EA. I just came up with the original idea and I haven't had a chance to even look at the code. But since Steve and others ARE getting sells, it does sound like it's probably crim-related somehow...?

[Jonathan2FI]

This was just an "I wonder if this works" idea, but it seems to have some legs -- at least when the EA cooperates. Might be worth looking at again to see if we can improve it. But I've already got about 6 balls in the air and I don't have time to look at it...

Gary,
Interesting EA profitable on single pairs!
Reducing RDD and increasing NP opportunities
1. Bounce trades off the supertrend line in direction of supertrend
2. SL moves with supertrend or EMA
3. Bounce trades off an EMA say 5 10 or 20 in direction of trend - long low crosses below EMA short high crosses above EMA
4. Do not trade if trend on supertrend is flat or below threshold change on look back bars LBB
5. Wait x LBB after start of new trend to confirm trend confirmation and not wipsaw potential
Jonathan
ps still have not mastered how to BT multipairs

[DocT]

Thanks, Steve! I will be fascinated to see how this works. I really didn't expect a simple SuperTrend strategy to work, but it did seem to.

George, FWIW I tested it on H4. I suspect running it on only one symbol, you will see some nasty drawdowns. I think you need many pairs to smooth out the results. When I get a chance, I'll post the individual-pair results of my hand backtest.

Gary

I too tested a similar strategy (described below) manually for a period of one month (June 2012) for four pairs and the results were excellent - I tried to build EA but my skills really poor [ok were non existent] so it was brilliant to stumble across Trendy Gary EA.

My NoNameStrategy - Range Bar based with SuperTrend entry and exit:

This was tested using 8 Range Bars and SuperTrend -
Entry on close of bar after ST change in direction of change
Three targets based on FIB extensions - 1.618, 2.618, 3.618
S/L moved to B/E when first target reached
Then Trail S/L with SuperTrend - exit on crossover

The results:
GBPUSD - + 1581 pips [Probability 81%]
EURUSD - + 1619 pips [Probability 91%]
AUDUSD - + 654 pips [Probability 69%]
NZDUSD - + 876 pips [Probability 70%]

I attach a word doc with image of one GBPUSD trade which hit T1 and T2 and S/T exit of T3 - a gain of 105 pips.

I dont know if range bars can be used in EAs - I know one can run a script in Empty4 to create Range Bars, but this was a very profitable strategy so the hopes for Trendy Gary are very high.

Cheers
Doc

[Cyrus7]

Hey Guys,
Thanks for the great idea and ea!
I got mixed results after my first 2 days with gary (see attached pic).

settings: setfiles from page 2
broker: Bully Pepperstone, gmt +2

[garyfritz]

Doc/taggie11, so you were just running SuperTrend on a range-bar chart, entering when ST switched direction, and exiting at Fib TPs, trailing SL, or another ST switch? "8 range bar" = 8 pips?

I'm not sure range bars are necessary for this. I can display range bars on Tradestation and I've converted the Empty4 SuperTrend to TS. I applied ST to 30min EURUSD, which had 168 bars from 1/10 to now. ST went long on 1/10, then flipped 3 times starting on 1/14. Then I tried it on EURUSD 12.5 pip range bars, which matched the 30min compression almost exactly -- 170 bars. ST on the range bars switches exactly as often as on the 30min.

This is probably too small a test to really be sure. In particular range bars might do better in really thin market periods. But in this test it looks like regular time-based bars would work at least as well as range bars. You just have to choose the right bar size. With Empty4 that would be tough since you have so few choices, but for the last week of EURUSD, 8-pip range bars are just about equivalent to M10 bars in terms of # of bars -- but the M10 bars whipsaw SuperTrend less. 8pip range bars switch direction 17 times since 1/10, and M10 bars switch 10 times.

Also, I'm not sure it's worth taking profits at multiple TPs. That works well if your strategy tends to get some open profit and then give a lot away. ST does a pretty good job of reversing before you give back too much. I think you'd probably be better off to hold your whole position until ST reverses.

[DocT]

Doc/taggie11, so you were just running SuperTrend on a range-bar chart, entering when ST switched direction, and exiting at Fib TPs, trailing SL, or another ST switch? "8 range bar" = 8 pips?

I'm not sure range bars are necessary for this. I can display range bars on Tradestation and I've converted the Empty4 SuperTrend to TS. I applied ST to 30min EURUSD, which had 168 bars from 1/10 to now. ST went long on 1/10, then flipped 3 times starting on 1/14. Then I tried it on EURUSD 12.5 pip range bars, which matched the 30min compression almost exactly -- 170 bars. ST on the range bars switches exactly as often as on the 30min.

This is probably too small a test to really be sure. In particular range bars might do better in really thin market periods. But in this test it looks like regular time-based bars would work at least as well as range bars. You just have to choose the right bar size. With Empty4 that would be tough since you have so few choices, but for the last week of EURUSD, 8-pip range bars are just about equivalent to M10 bars in terms of # of bars -- but the M10 bars whipsaw SuperTrend less. 8pip range bars switch direction 17 times since 1/10, and M10 bars switch 10 times.

Also, I'm not sure it's worth taking profits at multiple TPs. That works well if your strategy tends to get some open profit and then give a lot away. ST does a pretty good job of reversing before you give back too much. I think you'd probably be better off to hold your whole position until ST reverses.

Gary

Thanks for your considered reply. The main reason I like range bars is that they capture momentum better - in addition - bars are same height, the close is always at the high/low of the bar, the open one tick/pip above or below the previous bar, time is irrelevant, and volume correlates with price better - you can see accumulation of strength.

When doing my initial test of this strategy, the time based charts did not produce such good results, so I focussed on the Range bar charts.

As for the SL trail stop, you are correct - I have the data for that too (the additional pips captured by ST Trail stop are given second:

GBP - 1858 + 512
EUR - 1619 + 187
AUD - 654 + 113
NZD - 876 + 169

So for one month, with only four FX pairs, this strategy produced 5988 pips! Not a bad result I think.

Sadly I could not code this, and as a manual trade it was easier to set fixed targets, and since we were moving country/house/jobs/etc/etc, I could not trade it manually either.

Trendy Gary works on one hour chart - if 8Range equates to 10m, then perhaps we should test TG on 10m charts? Do you have any data for other timeframes?

Cheers
Doc

[garyfritz]

Doc, you said you hand-tested this last June. What bar size, and what ST settings? I can look to see if June was typical or a lucky fluke.

I forgot Empty4 doesn't support M10. But if the strategy has any robustness, it shouldn't be hyper-sensitive to bar size. I can test on any bar size you want on TS. With the default ST parameters (NumPeriods = 10, Multiplier = 3), EURUSD loses steadily over the last year on M5, mostly sideways on M10, did well from May through Oct last year on M15, lost from Feb to Sept but won big from Oct through Dec on M30, lost Mar-Jul and won Aug-Dec on H1, sideways/down with H4, did OK with D1.

That test is a simple "reverse when ST reverses" strategy. I'd be fairly shocked if something like that made consistent profits, just like I'd be shocked if a simple MA-cross strategy won consistently. I think it needs additional intelligence to improve its edge. Your trailing S/L might help but I think it probably needs more. I was toying with the idea of putting a trend filter on it, so you only go long when the larger trend is up. That might avoid a lot of whipsaws. But you'd also want to enter when you cross a trend MA -- otherwise you'll be trending down, you'll pass on a long trade, and it turns into a huge win that turns the trend. ST does that fairly often. But I haven't played with that yet...