Holy Graily Bob 'n Dotty

[SteveHopwood]

I forgot to add earlier, that I am receiving pm's asking me to apply the latest fixes to all the other varieties of HGBnD and Dottybot et al.

I will do, eventually, but people will have to wait. It will be far easier to replace HGBnD's trend spotting/trade trigger with those from the other varieties of existing bots than it will to apply the fixes/amendments/updates to all the other varieties.

For the time being, HGBnD is the only bot that Thomas and I are working on. Everything else will flow from here.

[SteveHopwood]

It has been informative to have just one pair open and been lucky enough to catch HGBnD in behaviour with is accurate according to the code, but not helpful to the trader.

Have a look at this picture and especially the blue rectangle at the bottom right:

Undesirable behaviour.png

This is GPY on the H4 to detect the overall trend, M5 as the initial trade trigger. Trading done on my Beeks VPS and monitored on my PC.

The pound has tanked this morning after our Prime Minister opened her mouth about Brexit. Those 5 buy trades will have been left behind by hedging the original sell trade. The tanking market filled a lot of sell stop but look at what happened next - inside the blue rectangle.

Sell stops filled, the market bounced back up and the latest sell was closed by offsetting. This happened three times.

Low look at the closed trades:

Result.PNG

The losing trade at the bottom of the pile is offset by the winner at the top - massive overkill to start with, then the profitable trade gets smaller as the trades at the top of the pile close.

A good bounce back upwards will correct this as most of the rest of the sells will close; DD will decrease; buy stops created as the market tanked will fill and all will be well with the world.

But suppose the market is only briefly pausing and resumes its downward plunge? Those unprotected buy trades will go deeper and deeper into DD. Here is the open position when I took the screen shots:

Open.PNG

The DD is roughly twice the accrued profit.

This is undesirable behaviour and I need to think about what to do about it. Any ideas folks?

----------------------------------------------------
Here is the position a few minutes later:

Market rise.png

Market rose one more grid level, offsetting kicked in and there are now 1 sell and six buys open. Great if the market bounces some distance. Bit of a bugger if not.

Definitely food for thought here.

[SteveHopwood]

V 2l is in post 1.

This includes Thomas' solution to the pending trades filling against an opposite direction signal and then closing immediately they reached a tiny profit.

Thomas' idea is to only close the trades if the opposite direction signal is a new one. He added the code. I have added some more to save the old signal so the bot can resume where it was following a restart.

Re my previous post, I have gone for the blunderbuss solution until we come up with something more subtle. I have added a new input to the Offsetting section - AllowOffsettingWhenHedged. This is 'false' by default, so HGBnD will not indulge in offsetting when I set her on my account.

I am taking the latest version straight onto my live account but I am trading ridiculously low lot sizes for the size of the account. You might care to wait until the latest code is shown not to have any bugs if your position is at all complicated.

Time will tell whether this is a good approach or whether it will benefit from refinement. We shall see.

[SteveHopwood]

Thomas just sent me a pm with some extra lines that need adding to his code, so I have updated 2l in post 1.

Easy DIY. Go to line 5391 and copy this over the existing block:

Code: Select all

if (CloseOnOppositeSignalOrTrendChange)
   {
      //Close buys following a sell signal/trend change.
      //Changed by tomele
      if (BuyOpen && (NewShortTradeTrigger || NewShortTrendDetected))
      {
         DebugPrint(__LINE__+"  "+__FUNCTION__+" called CloseAllPossibleBuys"); CloseAllPossibleBuys();
         NewShortTradeTrigger=false;
         NewShortTrendDetected=false;
      }//if (BuyOpen && (SellSignal || ShortTrendDetected)))
      
      
      //Close sells following a buy signal/trend change.
      //Changed by tomele
      if (SellOpen && (NewLongTradeTrigger || NewLongTrendDetected))
      {
         DebugPrint(__LINE__+"  "+__FUNCTION__+" called CloseAllPossibleSells"); CloseAllPossibleSells();
         NewLongTradeTrigger=false;
         NewLongTrendDetected=false;
      }//if (SellOpen && (BuySignal || LongTrendDetected))
   
   }//if (CloseOnOppositeSignalOrTrendChange)
           

[Maximilian]

Hi, here are unfortunately my last results for the 5min basket spinoff.
The demo ran from the end of November till now. Completely hands off. Anyway, it was nice to see the profits coming in.

XTIUSDH4.png

Thanks to you Leon for your efforts to find the best combinations of variables, here and with other EAs.
And of course thx to Steve and Thomas and a happy new year to all of you.

Cheers Max

Hi Max,

I was forward testing the setfiles with various combinations ie. Trend timeframe 1hr and trading timeframe 5 minutes and vice versa. I always like to put a bit of a spin on my tests and absurd as it may seem this back to front combination produced much better results than the logical path we would normally follow...

I am currently on hold with my tests till next year.

Compliments of the season.

Cheers
Leon

[Wavegarrick]

Hi Max,

Hi, here are unfortunately my last results for the 5min basket spinoff.
The demo ran from the end of November till now. Completely hands off. Anyway, it was nice to see the profits coming in.

Your results are simply stunning actually, up to a point. To grow an account from 3000 to 20 000 plus in just over a month sounds too good to be true, but I have had similar growth and yes great to see the profits coming in. I have also seen my account smashed by one single event.

I am going to give this another try this year with less pairs and incorporating s/losses, will play with this.

Ps. For the uninformed. This pertains to Pg 11 of this thread and not the latest developments or ea versions.

Cheers
Leon

[pakeha]

Wavegarrick, Max

I have been running the setfiles live on 4 pairs since the beginning of December with good results. I have limited my drawdown to 15% (a hard stop) as a way to circumvent a real nasty loss. There has been one situation where drawdown approached this level and had to be partially managed. Fortunately, the 2 pairs (GBP pairs) involved recovered so it all ended well. But as we know it could have easily been the other way.

I have also experienced some of the quirks that Steve and others have noticed, which also makes things interesting.

However, so far so good, but it does require active monitoring and manual intervention.

Hope this helps

[Wavegarrick]

Hi Pakeha,

Thanks Buddy for that Will follow your good advice.

Cheers
Leon

[Generaal Goudvis]

Hi, here are unfortunately my last results for the 5min basket spinoff.
The demo ran from the end of November till now. Completely hands off. Anyway, it was nice to see the profits coming in.

Thanks to you Leon for your efforts to find the best combinations of variables, here and with other EAs.
And of course thx to Steve and Thomas and a happy new year to all of you.

Cheers Max

Hi Max,

Impressive!.

But how can you have a bigger max floating loss than max eguity?

Cheers
GG

[renexxxx]

But how can you have a bigger max floating loss than max eguity?

Max Equity = USD 19,317.85 on the 16th of December. This is where the blue line (Equity line) reached its maximum.

Max Floating Loss = USD -21. 625.44 on the 9th of January. This is the distance between the yellow line and the blue line. This distance can be larger than the Max Equity, if the balance (yellow line) is higher than Max Equity, otherwise Equity would be below zero.

Hope this helps.