Steve Hopwood's Epsilon Megastar fib auto-trading robot | Page 7 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Time dependent auto fib? 1 reply

Auto fib bar count 0 replies

My Daily Fib Trading Journal 2 replies

Daily hi/low fib trading 35 replies

Steve Trading Journal 3 replies

Attachments: Steve Hopwood's Epsilon Megastar fib auto-trading robot
Exit Attachments

Steve Hopwood's Epsilon Megastar fib auto-trading robot

  • #121
  • May 15, 2010 3:13pm May 15, 2010 3:13pm
I am getting backtest error: "You have not named you fib object 'Fibo' on the latest update of the bot.

I was able to backtest on the version prior to the latest one.
  • #122
  • May 15, 2010 3:37pm May 15, 2010 3:37pm
Quoting octomanoak
LOL. You mean yes there is NOW.
shutup shutup shutup shutup shutup shutup shutup shutup shutup
  • #123
  • May 15, 2010 3:45pm May 15, 2010 3:45pm
Quoting octomanoak
I am getting backtest error: "You have not named you fib object 'Fibo' on the latest update of the bot.

I was able to backtest on the version prior to the latest one.
No idea why. Sorry.Mine is backtesting happily. I do not understand the crap platform we sorry retail traders have to endure enough to understand why backtesting is not working for you.

  • #124
  • May 16, 2010 4:36am May 16, 2010 4:36am
I worked out why the early entry function was not sending trades - I had the swing conditionals set to up and down, not low and high, so they could never be true.

I know this works now because it is happily running a back test. Update is in post 1.

  • #125
  • May 16, 2010 4:43am May 16, 2010 4:43am
Hey Steve

I wrote a Java simulator for the QG strategy back in 2006. It wasn't profitable. Perhaps you will be able to go further than I did.... I can honestly say that my faith in Fibs was shattered after I saw the evidence.

Anyhow, no trips to negative town. I was just passing on my observation.

TD
  • #126
  • May 16, 2010 5:17am May 16, 2010 5:17am
Quoting tdion
Hey Steve

I wrote a Java simulator for the QG strategy back in 2006. It wasn't profitable. Perhaps you will be able to go further than I did.... I can honestly say that my faith in Fibs was shattered after I saw the evidence.

Anyhow, no trips to negative town. I was just passing on my observation.

TD
That's ok. Yours is a respected opinion.

We have some colossal trading minds working here. Success is inevitable.

  • #127
  • May 16, 2010 5:36am May 16, 2010 5:36am
Quoting bluemele
One thing I do on the Russell and ES futures is, I use the smaller timeframe fibs and larger time frame fibs to add a discretionary position as they will typically match on a pivot.

For instance I will mark on my smaller timeframe chart the larger timeframe 50% 61% 76% Fib area's. Then, I will then wait for the smaller timeframe fib to 'match' the larger timeframe.

Typically, a 76 on the smaller timeframe will be the 38 or 50 on the larger etc.. So, I doubt you could program this in or make any sense of it mechanically, but that seems to work...
You and DxTrade are the two most experienced fibbers contributing here, so I am not going to ignore your experience.

I just pulled an H1 fib and see what you mean. 'Fib confluences' are what a lot of people call these and it is easy to see why they are considered so powerful.

My next development will be to add fib drawing functionality to the robot. I am not good with mt4 graphics, but have the model in the icwr ea posted here last week. Once I have got this working, I will add the ability to draw a htf fib and we can take development from there.

  • #128
  • May 16, 2010 5:39am May 16, 2010 5:39am
Guys, I have started a partner thread to this one, so that newbies have somewhere to ask for help - http://www.forexfactory.com/showthre...89#post3727389

I would appreciate it if some of you could visit and bookmark the thread, and answer queries that come up.

Cheers

  • #129
  • May 16, 2010 6:58am May 16, 2010 6:58am
Quoting tdion
Hey Steve

I wrote a Java simulator for the QG strategy back in 2006. It wasn't profitable. Perhaps you will be able to go further than I did.... I can honestly say that my faith in Fibs was shattered after I saw the evidence.

Anyhow, no trips to negative town. I was just passing on my observation.

TD
An EA based on technical indicators alone cannot be profitable long term.

It's been tried by the big boys with some very experienced trading knowledge and bucket loads of $ to throw at development.

There are just too many factors at play in the markets for a fully automated system.

Scoobs.
  • #130
  • May 16, 2010 7:59am May 16, 2010 7:59am
I have been doing some screen time in strategy tester (GBP/USD 5m) and found that we miss out on some pips just after the first initial trade. On a lot of occasions it will reach the first fib expansion level (fib expansion tool) from the previous wave and then retrace and hit stop.

If we split the trades up and take profit on say 75% of the trade at the first fib expansion and then leave the rest to hit the normal stops we should get better results. Hopefully the following picture will help explain:

Attached Image (click to enlarge)
Click to Enlarge Name: GBPUSD_Epsilon1.png Size: 36 KB


I have already tried this by amending the code and get significantly better results.

I worked the TP as

Trend Up:
Inserted Code
double take = NormalizeDouble(FibHigh_prev+(FibLow-FibLow_prev),Digits);

Trend Down:
Inserted Code
take = NormalizeDouble(FibLow_prev-(FibHigh_prev-FibHigh),Digits);

This can probably be taken further by having a second fib take profit etc.
  • #131
  • May 16, 2010 10:06am May 16, 2010 10:06am
Quoting fxdrive
I have been doing some screen time in strategy tester (GBP/USD 5m) and found that we miss out on some pips just after the first initial trade. On a lot of occasions it will reach the first fib expansion level (fib expansion tool) from the previous wave and then retrace and hit stop.

If we split the trades up and take profit on say 75% of the trade at the first fib expansion and then leave the rest to hit the normal stops we should get better results. Hopefully the following picture will help explain:

Attachment

I have already tried...
That is an interesting observation. It makes sense to me that if fibs are meaningful that fib extensions would be areas for profit-taking. I think the idea of scaling out of trades also makes sense, especially if it means "never letting a winning trade turn into a loser" and "letting the profits run."

The QG exit strategy initially sounded great, but some have observed that trades often retrace becoming small losers before a trade can move far enough to be a winner and close by a reversal. This could be a function of the minimum wave size being too small to allow room for a trade to develop before a reversal close, particularly if the market is tending more to range than to trend.

I wonder if thinking of the reversal close as a kind of jumping stop loss might make sense, possibly applying the reversal close on the second wave, rather than the starting wave. That way the trade may have a little more time to develop and hopefully not hit the adaptive or hard stop. Then using a different exit strategy to close the trades in profit, including scaling out.

Just some thoughts. I do not yet have enough of a feel for this bot's behavior in this market to have a clue whether these thoughts are worth pursuing.
  • #132
  • May 16, 2010 2:10pm May 16, 2010 2:10pm
Quoting octomanoak
I am getting backtest error: "You have not named you fib object 'Fibo' on the latest update of the bot.

I was able to backtest on the version prior to the latest one.
I got this error also. To fix change

Inserted Code
   if (!ObjectGet("Fibo", OBJPROP_PRICE1))

to

Inserted Code
   if (!ObjectGet("Fibo", OBJPROP_PRICE1) && !IsTesting())
  • #133
  • May 16, 2010 2:21pm May 16, 2010 2:21pm
Quoting fxdrive
I got this error also. To fix change

Inserted Code
   if (!ObjectGet("Fibo", OBJPROP_PRICE1))
to

Inserted Code
   if (!ObjectGet("Fibo", OBJPROP_PRICE1) && !IsTesting())
Are these changes to create a "special" version of the bot only for backtesting?
  • #134
  • May 16, 2010 3:57pm May 16, 2010 3:57pm
Quoting octomanoak
The QG exit strategy initially sounded great, but some have observed that trades often retrace becoming small losers before a trade can move far enough to be a winner and close by a reversal. This could be a function of the minimum wave size being too small to allow room for a trade to develop before a reversal close, particularly if the market is tending more to range than to trend.
I am as sure as I can be that a 40(0) pips swing is insufficient to keep us out of bad trades these days.

On my live account this week, I am doubling this. I managed to just about break even last week, thanks to a Friday trade that was a lot of pips in my favour when I closed it to make sure it would not get hit by the gap come market opening.

My instinct tells me that we need a decent swing before we can trade it these days. Frankly, I am not sure 80 is enough either; we shall see.

  • #135
  • May 16, 2010 4:32pm May 16, 2010 4:32pm
Quoting SteveHopwood
I am as sure as I can be that a 40(0) pips swing is insufficient to keep us out of bad trades these days.

On my live account this week, I am doubling this. I managed to just about break even last week, thanks to a Friday trade that was a lot of pips in my favour when I closed it to make sure it would not get hit by the gap come market opening.

My instinct tells me that we need a decent swing before we can trade it these days. Frankly, I am not sure 80 is enough either; we shall see.

For XXXJPY pairs then you should be looking at around 100(0) pips swing before a nice retrace comes into play.

As a general rule if a pairs daily range has been broken either up or down then its a good indication that a retrace is likely.

Scoobs.
  • #136
  • May 16, 2010 4:47pm May 16, 2010 4:47pm
Quoting scooby-doo
For XXXJPY pairs then you should be looking at around 100(0) pips swing before a nice retrace comes into play.

As a general rule if a pairs daily range has been broken either up or down then its a good indication that a retrace is likely.

Scoobs.
Cheers Scoobs. The UseDynamicSwingCalculation thingy allied with the MinimumFibSize should take care of this. So, when trading the xxxjpy pairs people should set:

 

  1. UseDynamicSwingCalculation to true to take care of >100 swings.
  2. MinimumFibSize= 100(0) to ensure the fibs are drawn over the latest 100+ swing.

  • #137
  • May 16, 2010 4:52pm May 16, 2010 4:52pm
Quoting SteveHopwood
Cheers Scoobs. The UseDynamicSwingCalculation thingy allied with the MinimumFibSize should take care of this. So, when trading the xxxjpy pairs people should set:

 

  1. UseDynamicSwingCalculation to true to take care of >100 swings.
  2. MinimumFibSize= 100(0) to ensure the fibs are drawn over the latest 100+ swing.

Steve,

You could automate it based on the pair the EA is attached to:

MinimumFibSize=ATR(NULL,PERIOD_D1,14,0)*Point;

or something like that.

Scoobs.
  • #138
  • May 16, 2010 4:55pm May 16, 2010 4:55pm
Quoting scooby-doo
Steve,

You could automate it based on the pair the EA is attached to:

MinimumFibSize=ATR(NULL,PERIOD_D1,14,0)*Point;

or something like that.

Scoobs.
Thanks scoobs. It is already automated like this, but people have to set their minimum accepetable fib size in case Atr returns too small a figure.

  • #139
  • May 16, 2010 4:58pm May 16, 2010 4:58pm
Quoting SteveHopwood
Thanks scoobs. It is already automated like this, but people have to set their minimum accepetable fib size in case Atr returns too small a figure.

oky doky

  • #140
  • May 16, 2010 5:11pm May 16, 2010 5:11pm
I have amended the user guide thus:

 

    1. MinimumFibSize: the minimum fib size you will accept. The higher this is, the wider will be your fibs. Overrides the dynamic calculation if this results in a smaller fib size than your acceptable minimum. Markets have changed since the strategy document was first published. For what it is worth, I think this:

       

      1. Non xxxJPY pairs should have a minimum fib size of 60(0).
      2. xxxJPY pairs should have a minimum fib size of 100(0). Actually, this figure is not mine. It is offered by Scooby-doo, a professional trader kind enough to offer his professional expertise to us free of charge. Only an idiot would ignore it.

Cheers Scoobs

0 traders viewing now