What to do…

The new version of GridBot was released a week ago. I was not impressed. After running backtests the entire weekend trying to find anything good with it I put it on demo accounts and hoped that the backtests we’re wrong. Well, it didn’t end well and I finally went for the refund. The money hit my Paypal yesterday. I’ve lost quite a lot of money and time on this bot. I really think it is possible to make a profitable grid trader (just look at Robominer), but not this one. Very disappointed actually since the developers are very inactive in their own forum also, not answering questions. (I still wonder where the hedge function in the new version is… I never saw a hedge of any sort).

FAPT Evolution is still running on demo, or at least I think it is, I’m not getting it to trade much at all. Reading forums the Dukascopy demo server really seems to give different performance for most of us. It is quite worrisome I think as they’ve stated that it should give very close to 100% matching performance with their live server (difference of course being the liquidity). The ones running it live seems to have a smoother ride although EURGBP apparently hit a S/L this week. I’m looking forward to Donnas update this weekend as I tend to use her settings for both regular FAPT and Evolution. She is very knowledgeable about these products.

As for MD Pro I put it up on a demo Forex.com/UK account and it has traded quite a bit (compared to regular MD), but is actually a few dollars behind for the week. EURGBP is the main culprit. Perhaps this pair isn’t really working out for scalping at all right now (as it is the same with FAPT and FAPT evo). The other pairs have done well. I’m thinking of replacing my regular MD on Forex.com/UK live account with the Pro version for next week. I still experience daily crashes with regular MD. The Pro version at least haven’t crashed the demo account once running it without MT4 restart all week.

After running Robominer on demo since June 1st I must, again, just state how impressive it is. They just released a new version that is updated to work with NFA regulated brokers after the new rules about FIFO is enforced, and they state that is is even more profitable with these new changes. Sounds good. I must buy it now and put it on a account somewhere. The developers are currently recommending Forex.com/UK, I have a account there already of course. I am also considering FXCM/UK bacause of their advantage in lower spreads. However they only allow trade with 0.1 lots, Forex.com/UK trades 0.01 lot size. The smaller lot size of course allows lower risk taking, but I’m so confident in this bot now I’m considering trading it at 0.1 anyway (think they recommend a account size of $35000 (updated see comments) to do this, but as long as you have funds to put in there in case of a sustained DD you can fund a smaller amount to start with).

Received spam email from FAPT Steve yesterday about a new bot they’re impressed with (they’ve promoted MegaDroid apart from FAPT before, and some manual systems) called IvyBot. It’ll be released next week and have a webinar planned for Monday. I’ll check it out. Could be something, but there’s so many new bots being released I just don’t know

Oh, I almost forgot. The Joury EA bot I’ve written about have been found to be a free EA by a guy called TrendChaser. The developer himself doesn’t consider it to be a good strategy (it can work for a few months only to wipe the account he says). So hope no one bought it. It is available free if you want to play with it, I’m thinking about doing so a little this weekend, just for giggles.

I’m going to make charts for my live accounts later tonight or tomorrow for the weekly update. I can just say that it doesn’t look good now, and that the Alpari UK account have been doing much better than the Forex.com/UK account. The markets have really punished MD yesterday and today.

It gives me terrible results in backtests. Wiped account on default settings on GBPJPY running from 2009.06.01. Lets hope that it perform better in forward tests…

They use limit orders now, good I think, makes the grid lines visible on the chart and it is more likely that our orders get executed without slippage. But they put sell orders over and buy orders under the price. Where is the hedge in that? A hedged grid puts buy and sell orders on every grid line.

This would give the same behavior as version 2.73, just now they’re using limit orders and we can set the grid spacing. Also, they mentioned some kind of trend awareness, but I don’t see that in backtests either.

I’ll continue some tests now (running a ~40 hour optimization run to figure out optimal grid spacing and max orders) and put it up on demo account (with default settings), but I must say that I am not very impressed right now, can’t imagine it to perform better than the previous version. And in that case I have to go for a refund.

Took out detailed reports for the week from my live accounts at Alpari UK (sporting FAP Turbo, MegaDroid and GridBot) and Forex.com/UK (only running MegaDroid). Not as nice as I hoped, with losses yesterday and today.

All in all, Alpari UK finished at a loss of €44.87 and Forex.com/UK at a profit of $66.87. That is basically a break even if you add it up… What a waste

Alpari UK live account

Forex.com/UK live account

So, MegaDroid did well on both accounts. FAP Turbo did well on Alpari UK. So what is left? Well GridBot didn’t do well… Again it loses me money. I will really stop trading it this time thou. Next week I expect them to release the new version (2.9) and then I’ll test it again.

First of all, I got some comments after my post about AutoHedge EA last week. All positive I’m glad to say. However I won’t release it to the public, at least not right now. I think it is better for everyone (including me perhaps) to wait for the v2.9 of GridBot. It will surely be very nice.

So what new functions have I implemented? Well other than some efficiency enhancements and ECN broker support I have implemented a function that monitors when GridBot opens a order and sends an inverse one immediately. The inverse order has the grid spacing as T/P (take profit). This means that the inverse order will T/P exactly when the GridBot wants to open another order.  I currently am limiting it to just open this one order, but it could be more general with some simple adjustments. It’s in demo testing right now

It is real fun to write EA:s I’ve found. Probably will try to make some more on my own, using other strategies.

I was a bit bummed after sacking the GridBot from my live account this Thursday. The reason of course is that it once again hit the Safety net, losing me money. But since it really makes money when the market conditions are good I felt that it was time to try to write my own hedging EA to help GridBot out a little. So I did. Of course I don’t know if it will make me money in the long run, but at least it will reduce the risk of huge draw downs.

Basically what I have written is a EA that monitors all open trades in a currency pair and calculates the running profit/loss. If this running profit goes under a defined threshold value my EA opens a inverse hedge order in the pair, the size of the open orders. It then continues to monitor the price and closes the hedge order if the market reverses back into the grid. When to close the hedge is also definable of course.

In the picture the AutoHedge EA (silly name I know…) monitors AUDNZD, GBPJPY, USDJPY, EURJPY and EURUSD orders made by the GridBot. It is currently hedging AUDNZD because the running profit there is worse than -100. With the cost of opening the hedge the loss is now locked in at -136 (AUDNZD has huge 20 pips of spread making it expensive to hedge). If the running profit of the GridBots orders improve to -75 EUR the hedge would be closed with this configuration.

So what does this cost? Well if the market never returns to the exit threshold the hedge order have a cost of the spread. If it does return it costs the spread and the difference between entry and exit. I feel that this value is the one that will be difficult to set. Too small we may be whipsawed into opening several hedge orders during one session (expensive with the spreads of some pairs), too big we just pay to much for the hedge, taking away profit from the GridBots orders.

Pairs with tighter spreads will be much better for this function as the hedge won’t cost as much, so I think I will primarily test it on USDJPY, EURJPY and EURUSD with Alpari UK demo. I will also open a demo with Forex.com UK.

So as I said, I don’t know that this will save money in the long run, only that it will effectively limit the huge draw downs with the GridBot. Instead of hitting that Safety net of $500 one could have been hedged at a $100 loss, with the possibility of the market to reverse and making a small profit (closing the GridBot orders with profit but  paying for the hedge).

I am thinking about hedging every order individually with a inverse order and use the OrderCloseBy() function instead in combination with the corresponding GridBot order (this saves one spread) if the market doesn’t return and the GridBot is about to close the orders (I would then jump the gun and close them say 1 minute before EndHours). But that would close the GridBot order, with the current implementation the GridBot isn’t affected at all by the hedge orders. Also I know that the  GridBot developers are working on a real hedge implementation in their EA (can’t wait for it). This is just a substitute for the Safety net feature, not a real hedged grid trading strategy. But I’ve been meaning to check out writing some MQL4 for some time, so I saw this as a opportunity. Perhaps it will evolve into a stand alone grid trading EA also, I am quite tempted to just write it all myself…

Oh, and yes, I have been tempted to let the hedge order be open after the GridBot closes its orders, with a gliding take-profit. If the trend is good this could make some money, but it really isn’t the whole point here, so I won’t implement it like that right now.

Been evaluating my EAs. The only one that I’m ahead with now is MegaDroid. FAP Turbo and GridBot lose some more yesterday. GridBot hit the safety net on GBPJPY. FAP Turbo did more damage in EURCHF.

It is clear to me now that I have been overlooking the performance and stability MegaDroid have been giving my account (although it didn’t trade yesterday). It only trades once in a while, but it is very profitable. I am forcing it to trade a little more by running it on three charts. I am also considering experimenting more with it in demo, trying it on M15 timeframe to start with. Read on DonnaForex forum about a guy doing this with nice results. The trading logic in MegaDroid seems very solid, so it would be nice to have it trading more, even though one of the strengths of MD is that it doesn’t necessarily trade every day, just under the right conditions. I’ll run it in demo for a while.

I’ve made the decision to increase the risk taking with MegaDroid a little, and at the same time reduce it with all the other EAs on my live account. I will stop trading GridBot entirely until they release the version with the hedging (expected sometime in July). They will release a new version tomorrow, but it is presumably only sporting a new automatic time zone function. Nothing that will help the actual performance of the EA. I am really looking forward to the hedging function, as it really is sad do deactivate this EA. I have enjoyed watching it trade, but the reality right now is that it is to risky. Others may make lots of money, but I am not. Will keep it running in demo of course, so if it does convincingly well I can always put it back, with small lot sizes (0.01).

For FAP Turbo I will follow DonnaForex settings, but I won’t use the high risk setting in EURCHF again. We’ll see in her news letter on Saturday if she will reduce it herself. If she doesn’t I will.

I am currently trading FD2 live with a fixed lot size of 0.02. It hasn’t made any remarkable trades, in either direction, so I will leave it like that for now.

So with this new setup of reduced risk I am hoping to make some of the losses back. But since I need 100% profit to get back I decided to fund the account with another €500. This hopefully gives me a better chance I think. I am still hopeful.

So, when I went to bed yesterday (at about midnight GMT+2) my live FX account where I run all my little bots at had about €810 equity. It had made some nice profits during the evening, most notably six profitable trades in GBPJPY and USDJPY by the GridBot EA. When I woke up this morning and saw my balance as €616 I was kinda surprised. Seems the markets turned at about 1 am GMT+2 and the GridBot opened three positions in both GBPJPY and USDJPY against the trend raking in losses for €187 in the process.

Exactly this behaviour is why the GridBot is dangerous. It keeps adding fuel to the fire if there is a trend. If that trend doesn’t reverse during the grids working hours, we are in for big losses. If it does reverse we are in for small profits. The risk vs reward just isn’t good enough when there’s such a strong move like that last night.

The developers of GridBot have stated long ago that they have a hedging implementation, but they haven’t released it yet. We don’t know the details of this implementation. I would like the bot to open a inverse position when the market moves outside the grid. If the price reverses the hedge position will close breakeven and the grid can take profits. If it doesn’t we just take the loss that have been incurred inside the grid, the hedge will protect the movement outside.

So basically I am tired of this bot screwing up my account. FAP Turbo, MegaDroid and even FD2 is doing much better right now. I’m thinking about placing GridBot in suspension, only to trade demo until there is another release, which include the hedging implementation. But I’ll give it one last chance. I’m changing the working hours to 18 - 1 GMT+2 (Alpari UK account time). This is the time slot from one hour after London close right up to Tokyo opening. I’m also lowering the lot size. GBPJPY will only get 0.03 lots from now.

If they don’t release the update soon, I’ll go for a refund…

I put up another demo account yesterday trading FAP Turbo (all pairs), MegaDroid (three EURUSD charts with 1 hour offsets and small risks), GridBot (GBPJPY, USDJPY and EURJPY) and FD2 (all four pairs). This demo is at Alpari UK and I funded it with €1000. You can follow it here: http://jasonvoorhees.mt4stats.com/

This new demo is really trading like my live account at Alpari UK, so perhaps the trades there actually will be useful in finding optimal settings. FD2 is making some really nice trades, hopefully it continues and proves that the defects from last week is ironed out and I can put it on the live account.

Unfortunately AUDNZD and CHPJPY isn’t offered at the Demo account, I still trade these live with the GridBot thou. CHFJPY is strongly correlated with the other Yen pairs of course. AUDNZD is both “Comdolls” (Commodity trending dollars) and is therefor very nice to trade with the GridBot. They usually trade in a range over the day and GridBot can get multiple small trades in. I keep out during the Sydney session.

I will keep my old demo (at Alpari US) running for a while, but the account there really was trashed by RoboMiner. I would not run that bot on a live account…

Update: This setup has made 40% gains in just two days now. Let’s hope FD2 continues to convince me to run it live next week.

After another day of grid trading I’m once again changing my setup. One of the most successful, unorthodox, pairs in the past have been EURCAD. Well, that changed yesterday when, for some reason unknown to me, the EUR started to rally against the CAD, effectively destroying the GridBots selling trades and hitting the Safety Net. I know this was a risk of course, but I clearly had put a way too high Safety net, it should have been stopped earlier when it was starting the trend. But just this is really the weakness of the GridBot implementation since it’s completely trend unaware. So, after this experience I’m stopping the EURCAD pair for now, will have to collect more data in order to set a safer Safety net.

Other than that pair yesterday was a good GridBot day actually. It closed 4 out of 4 trades in GBPJPY in profit, and 2 out of 4 USDJPY in profit (but still made €6.06 on this pair). It also made money in AUDNZD trading 9 -23 (basically letting it be when Sydney is open). The other pairs made some and lose some, but nothing out of the ordinary. So if it just hadn’t been for EURCAD…

I’m going to keep running GBPJPY/USDJPY/CHFJPY/EURJPY at 18 - 3 GMT. I might adjust the lot sizes downward a few notches, because of the smaller account I have now (see “FD2 crashes and burns…” post earlier today, and EURCAD losses above in this post). I basically lost 1/3 of my account yesterday, and have to reduce risk, even for overperforming pairs.

I have seen a trend the last week where I think I can make alittle more by using a 4 level grid in GBPJPY. It would have made me one extra profitable trade every day trading as of yet, so I think it can be a good idea. Probably will try it in demo account for a few days before activating it in the live account.

I really hope that the developers of GridBot enables their hedging function (they have stated in their forum that they have it developed and that it is doing better than the current implementation, but hasn’t relesed it because of the NFA regulations about hedgning. That regulation only affects brokers that are regulated by the NFA. Alpari UK ofcourse isn’t), I think it would make the bot capable of handling sudden spikes better. I am also tempted by using the Swiss Army EA to modify the GridBots orders, setting S/L per order and in some cases increasing the T/P. As it is now we always get punished when a trend forms “the wrong way”, but never rewarded when it goes the right way (beacuse the T/P is to tight). I would like to see a gliding T/P in these cases when there’s a big move in the right direction.

Enough rambling now. I’m still believing in the Grid trading system, when used at the right market conditions.

Monday and Tuesday haven’t been great with the settings I was trying out. A lot of safety nets have been hit, some prematurely but most have stopped further losses. I’m going to drop some pairs and alter settings for others.

USDJPY:  LotSize 0.1, SafetyNet -300, trading between 18 and 3 GMT

So far this pair has 7 winners out of 7 (100%), so it’s a keeper, even thought about raising the LotSize, but I will not for now. Raised the SafetyNet since I trust it more now.

GBPJPY: LotSize 0.05, SafetyNet -150, trading between 18 and 3 GMT

Not as stable as it’s brother USDJPY in my trades. Has hit stafety net twice (once prematurely). I’ve put it on a smaller lot size and widened it’s safety net.

CHFJPY and EURJPY: LotSize 0.01, SafetyNet -30, trading between 18 and 3 GMT

These two are working out just as good as GBPJPY, might increase their lot size in the future, but I’ll wait for even more good trades.

EURCHF: LotSize 0.01, SafetyNet -20, trading between 9 and 3 GMT

This isn’t a clear winner, but it is ahead about €3 right now, so I’ll keep it. EURCHF is trading in a range very often thru out the day.

EURCAD: LotSize 0.05, SafetyNet -100, trading between 9 and 3 GMT

This is a clear winner in my book with €63 in profits and 32 winners out of 38. The other 6 trades were all closed with losses at 4:00 am. I don’t think I could have set another EndHour to stop those losses earlier, the orders we’re all put in during a bit of a trending market.

AUDNZD: LotSize 0.05, SafetyNet -100, trading between 9 and 23 GMT, 5 GridMaxOrders

This pair is trading in a range very often. I hit the safety net prematurely (the pair bounced back) on Monday, so I’ve widened it.  I’m also trying a grid with maximum 5 orders of 0.05 lots with it now, perhaps it can use the movements better.

USDCHF: LotSize 0.01, SafetyNet -13, trading between 9 and 18 GMT

GBPUSD: LotSize 0.05, SafetyNet -50, trading between 9 and 3 GMT

EURUSD: LotSize 0.01, SafetyNet -14, trading betweeen 9 and 15 GMT

USDCAD: LotSize 0.01, SafetyNet -20, trading between 11 and 19 GMT

EURGBP: LotSize 0.01, SafetyNet -20, trading between 9 and 14 GMT

AUDUSD: LotSize 0.01, SafetyNet -14, trading between 10 and 18 GMT

NZDUSD: LotSize 0.01, SafetyNet -14, trading between 11 and 19 GMT

GBPCHF: LotSize 0.01, SafetyNet -20, trading between 9 and 15 GMT

The pairs that ultimately was dropped was just to hard to set decent safety nets on. Some the grid bot almost always opened three positions in, against the trend, and they just hit the net or lingered as losers til EndHour was hit. I guess some of them can be traded during some periods of the day, when the market is less active, but it is very hard to set the correct parameters for this. I would like the GridBot to actually have some intelligence about trends and market conditions, not just blindly execute orders at the price levels in the grid and hope for the best.

GBPUSD showed the most promise of the dropped pairs, but with it ending two times in the Safety net during trends it just isn’t sustainable. And in both those cases the trend didn’t stop, so a wider net would have lost me even more money.

Anyway, I hope that this setup will make money, since I have lost quite alot with the experimenting right now. Hopefully EURCAD and AUDNZD can make it up to me in the coming days. Those pairs aren’t usually traded with this bot, but they’ve been winners for me.