Close Topic Options

I've been investigating the BATv2 indicator that someone developed as a modification of the ATR Trailing Stop that Team Aphid posted on ForexFactory a while ago. I'm attaching it here. Using it on default settings on GBP/JPY, I came up with some interesting results, the excel file of which I'll attach in one of the posts that will follow.

The crux of the system isn't in the pip gains, but in the money management, which allows us to maximize profit on big trending moves, and minimize losses in choppy markets. I'll explain with the rules and some charts...

I tested it on a 4hr chart because I'm looking for a long-term trend-following system.

Entry is at the break of a BAT line by the CLOSE of a candle, close above the GREEN line for LONGS and below the ORANGE line for SHORTS.

Stop Loss is set at the closest swing low/swing high, depending on if you're going Long or Short respectively.

Position is sized according to your Risk percentage, i.e. 3% of your account divided by number of pips in the stop loss. In testing, rather than compounding the interest with every trade, I compounded it every 2 months, and figured lot size based on the last 2-month balance. That means that if at the end of December my account was $10,000, then all through January and February I'd be risking 3% of $10,000, or $300. This is both easier for me to test and also may help in recovering losses since you're not fighting against your own account size. (What I mean by that is that if you're compounding every trade, and you lose 20%, you have to win 25% to recover the same amount... this way, you only have to recover 20%).

When a new BAT line, aka the modified ATR Trailing Stop line appears behind the price action, you move your stop loss accordingly to the position of the new line, and you do this every time a new bar closes.

When Price closes above/below this new line, you close your position and enter the opposite direction.

In choppy markets you often end up with very large initial stop losses, in the hundreds of pips. Because of the way the BAT line follows price action, by the time price reverses, a new line has formed and your stop has been moved to a much shorter distance. This means that while you sized your position to lose only 3% if price goes against you 300 pips, you end up getting stopped out at 50 or 75 or 100 or 150 pips, meaning you only lose .5% or 1.5% or whatever it is. This is how you minimize your losses.

On the other hand, a lot of the big trending moves end up with relatively small initial stop losses, say 150 pips, or even 200 or 300. They're RELATIVELY small because the trends on GBP/JPY will get 400 or 700 or 1000+ pips, meaning you win 6, 10, 30%, whatever it happens to be.

I backtested by hand from Jan. 07 to Aug. 08, starting with 5k and ending up with a little over 14k, compounding every 2 months, with only 2 negative bi-monthly statements. That's almost 300% in less than 2 years, with a 3% risk. Again, that spreadsheet will be attached after I post some charts to help demonstrate what I'm talking about.

Attached File ( Number of downloads: 123 )
 BAT_ATR_v2.mq4

Pardon the giant image, it's the only way I could get it legible.

Attached Image (Click thumbnail to expand)

Here's my excel spreadsheet. I backtested by hand (it's not exact to the pip) from Jan. 07 to Aug. 08, doing my accounting and compounding every 2 months. I didn't go back any further because in 05 and 06 GBP/JPY was trending like a mofo, so assuming massive profits, I don't need to do all that work.

I know the hardest part about coding this will probably be teaching the EA to pick out the right swing highs and lows, but it can probably be done to a point.

Attached File ( Number of downloads: 120 )
 batstats.xls

Your results here look great, I will take an in depth view of the system tonight to see what I can do.

Matt

This is an interesting system. I, too, would be interested in seeing if you can develop something to trade this.

Barry

I haven't forgotten about you. I will be starting this soon. Again, any information about the latest state of these systems will be helpful. The more investigation I have to do the longer it will take.

Thanks again,
Matt

I've been testing it using a hard rule for picking the swing high/low to set the original stop loss. If it's opening a Long position, then I look at the most recent Low fractal (that candle's low), and note its position. Then I count back 5 bars from the current bar, and note the lowest low within those 5 bars. The stop is positioned as either the lowest low fractal, or the lowest low within 5 bars back, whichever is lower. This will get the right swing low 95% of the time.

This initial stop is placed with the broker as a hard stop, and rarely gets hit. Sometimes, though, it gets hit and price then continues on into a big trend, so there needs to be a re-entry feature. On re-entry, the original BAT line should be used, and the initial stop/position size should be recalculated based on the formula I gave above.

I did more backtesting back as far as I could on the 4hr charts, to mid-may of 05. GBP/JPY actually did worse in 05 and 06 than it did in 07 and 06, but still turned a pretty decent profit. I also tested it on GBP/USD to try to alleviate some of the losing periods by running it on two pairs, and so far so good. I haven't finished all of 2008 with GBPUSD but I've attached the excel file anyway.


Paul

Attached File ( Number of downloads: 59 )
 batstats.xls

It wouldn't let me upload the entire file, so I split it into two worksheets. This one is for GBP/USD...

Note that the GBPJPY worksheet was re-done with the hard stop/re-entry rules applied, for better results than the initial testing provided.

Attached File ( Number of downloads: 59 )
 batstatsgu.xls

Hi guys

I love the MM you are presenting for this system. I have been trying to use similar MM on systems I am testing. I really like how a TrailingStop (TS) will minimize losses and maximize gains. So by simpily setting a hard SL and a TS we will know our maximum loss, in this case 3% and often the TS will lesson that even more! Of course the best part is how the profits can run in a nice trend.


Dan

Matt,

Here is the original system outline for the BAT in case you don't already have it.

Barry


Attached File ( Number of downloads: 110 )
 The_BAT_by_TA.pdf

Thanks Barry,

I don't have it yet. I will have a closer look.

Are you forward trading the sysem manually either demo or live?

Dan

I am currently forward testing on a demo account with the four majors. Just started last week (not the best of weeks!). I am incorporating the additional orders based on fib levels which has added additional pips on two occasions. This is described in the PDF. I'll let everyone know how trading goes this week.

Barry

Paul,

You have made an interesting change to the system. That is to enter on a close that breaks the line rather then using pending orders like the PDF suggests.

How about you, are you trading this system manually?

Dan

I'm keeping tabs on it, but I'm not actually entering trades. There's a few changes to the system, really. I don't use the additional orders placed on fib entries, I use a 4hr timeframe instead of 1hr, and the ATR Period and Factor are 3 and 1.2 rather than 5 and 4, making for a much tighter trailing stop. The tighter TS can be used successfully because price has to close above/below the line before action is taken, whereas if it were only a touch, the wider stops are necessary.

In any EA that's developed, I think it should also have the ability to use only the trailing stop function and apply it to manually entered trades. I've been using it more for this than anything.

The problem I have with the original BAT system is that it has huge drawdowns. You'll have months where you're -1000+ pips, and that's not okay with me under the original MM. And if you wanted to risk only a set percentage of your account, say 3%, then you have to plan your trade to incorporate the extra fib entries, which are not always hit. Therefore on the best trades, the ones that break out into long trends without hitting the extra fib entries, you're only at like .5% or 1% risk instead of your full position size.

The supposed advantage of the extra fib entries is that they help out in ranging markets, which they probably do... but I'm more inclined to have a longer timeframe trendfollowing system like the one I posted, and have a breakout system as well to catch the shorter moves.

I should mention that I'm also trading this on a 4 hour TF and wait for the candle to close. It does work better than the original which called for pending orders. And, as pointed out, the version 2 BAT indicator does provide tighter stops and does it relatively quickly.

I am experimenting with the additional fib orders, and yes, it does make MM more of a problem. Will just have to see how it plays out over the next few weeks.

Barry

I'm glad to see the discussion happening here about the system. Some of the trailing stop ideas are very similar to the AshFX, PipNailer, and RollerCoaster EAs and the PSAR levels where the previous PSAR dot can be used to set the trailing stop closing the gap. As a matter of fact, the impression I'm getting is that the biggest difference between this idea and the RollerCoaster is the indicator being used and waiting for the closed candle. Of course there are other subtle differences but I think their very similar which means it may not take a great amount of time to develop. If it's ok with everyone then I will incorporate most of my standard features that are included in the RollerCoaster or at least the ones that apply. Things like percentages as apposed to hard coded take profits, and others would be included along with all the flexibility of giving the EA more strict settings.

Anyways, I like the team work here.

One question... This is a freely shared system on a public site, right? In other words, no one paid for anything to provide the indicator, pdf, or anything else here, correct?

Matt

Yes, the original system was free and the v2 indicator is also free.

Yes, the original system was free and the v2 indicator is also free.

Then I don't see a problem. I just didn't want to cross any lines.

To Paul and anyone else interested in this system. I am making good progress with my other EAs and bringing them all up to speed with the newest suggestions, etc. I believe the system here is close enough to the RollerCaster idea to consider them close but I also believe there is enough that makes them different enough to justify a whole new EA. My initial look here would be to modify the RollerCoaster to accomplish either of the 2 systems but they differ in ways that would make things complicated. In short, I think it looks like a promising system and I intend to make this one my next new EA project. Are the rules in the first post still valid? I have just read the whole thread again just now and I think I have a clear picture of what needs to be done.

Many apologies for anyone who has suggested a new system and hasn't seen it completed as of yet. I take great pride in writing my EAs and this can have a negative affect on time since they may take me a little while to complete. I also believe this is good as well because I don't just throw them together and wait to see how they perform or what problems may come up. Writing a new EA requires patience, focus and a clear understanding of the logic behind a system and also troubleshooting potential problems because of the nature of the logic being used. Some of the simplest EAs can be the hardest to write when you consider that a program has to do it all for you exactly the way you want it to. I guess all I'm trying to say is please be patient and know that I'm working hard to provide everyone with the best support I can give.

Thanks again to everyone for your support,
Matt

I've decided to move this thread out of the request section since it's in development. Since it was Paul's suggestion, I would invite him to come up with a name for the AutoTrader or is "BAT AutoTrader" sufficient?