Manually trading SAFT to see if its profitable

[slipshod]

Hi everyone, I find myself between programming contracts, so its back to some intensive trading.

I was extremely interested in Evaluator's SAFT idea, especially using breaks of trendlines on OBV to verify trendline breaks on the price as a means of trading reversals. Hence I decided to give it a go.

Edit: Two weeks were enough to show me that while the SAFT method does have some spectacularly good signals, it also has a multitude of bad ones, enough to drain your account if you take them all. I don't see any way of filtering them out either through use of secondary indicators or my own discretion, so regretfully I ended up having to abandon the experiment.

That's not to say if won't work for other people. If it does, I wish them every success!

[slipshod]

So the basics of SAFT, as outlined in the thread at http://www.forexfactory.com/showthread.php?t=83944 are that you draw trendlines connecting the highs for a downtrend, or connecting the lows for an uptrend. Then you do the same with the OBV indicator, and when there's a break of both trendlines you have an entry signal for a reversal trade.

Below is an example of the basic idea for how this works (as I understand it - if I've got this all wrong please somebody stop me)...

[dakuki]

The big question to me is what OBV indicator to use. On the FF thread you have one by LHDT (either smoothed or not) and another by GG53, apart from the standard one in Empty4.

[slipshod]

If you've stumbled across my older manual trading thread (http://www.stevehopwoodforex.com/phpBB3 ... f=16&t=553) you'll know that I also look for more complex chart patterns than mere trendlines, especially falling & rising wedges which are very reliable reversal indicators. I won't repeat all that here, nor go into Chuvashov forks and how they work (all in that other thread), but I will be using them here so if you're not sure what they're about please spend a moment and glance over the other thread, its not long & hopefully everything's pretty clear.

Now for the twist I'm hoping to bring to this & make it even more reliable & profitable - using Open Order and Open Position data from kind criminals such as Oanda and Saxo, along with Myfxbook, all of whom provide that kind of information free of charge.

So what are these stats & how can we use them to trade?

Firstly, Open Positions shows the ratio of traders who've already entered the market & which way they traded. Its a guide to sentiment, however we must remember that once money's committed its incapable of then driving the market further, so these open positions do not drive future market direction, instead they're at the mercy of it. You can find Oanda's data at http://www.oanda.sg/analysis/open-position-ratios, MyFxbook's at http://www.myfxbook.com/community/outlook, and Saxo's at http://www.tradingfloor.com/topics/forex-open-positions (you'll need to register a free account).

The other side of things is Open Orders. These will include 3 types of order - Pending entries, Take Profits, and Stop Losses. As humans are creatures of habit & prone to groupthink its not surprising that these can end up clustered together rather than evenly distributed. If you know where a cluster of orders are, then potentially you know where the market might choose to change direction. If you know what type of orders they are, you'll also know what big money will likely do - for instance if there's a huge concentration of stop-loss sells about 50 pips under the current price with a large percentage of traders long, is it any surprise to see a smackdown pushing the price down into those stops? Its the kind of thing the big boys love doing, and you can be sure they know exactly where those stops are & where they intend to push the market after they're hit.

Saxo's open orders are here: http://www.tradingfloor.com/topics/forex-open-orders, and Oanda's are here: http://www.oanda.sg/analysis/forex-order-book and a really handy historical view here: http://www.oanda.sg/analysis/historical-open-orders. Myfxbook, not being a criminal, knows nothing about open orders.

For now I'll focus on Oanda, particularly their historical view as you can see how price behaves in relation to order clusters. As an example, below is the AU monthly historical open order graph...

aumn-oanda.png

There's notes on the image you'll probably want to read (click to zoom etc).

So if we have this additional intelligence about the market, how can we use it in our trading? How does it help us decide when to enter & exit when the system just calls for a simple break of 2 trendlines? Doesn't it just muddy the water and make things confusing rather than clearer?

I'll look at that in my next post...

[slipshod]

The big question to me is what OBV indicator to use. On the FF thread you have one by LHDT (either smoothed or not) and another by GG53, apart from the standard one in Empty4.

Hi dakuki, up till now I've just used the standard Empty4 one as it seems to work, though from now on I'll be trying the "enhanced" OBV detailed here: http://www.stevehopwoodforex.com/phpBB3 ... 642#p21642. I've always shied away from using anything that relies on volume, as Empty4's tick volume will be broker-specific and basically unreliable. However I can't deny the correlations I see on the chart so I've decided to just give it a go & see what happens

The more difficult part is how to decide where to draw trendlines. I've been doing it for years so its very simple for me, but I've seen other traders draw them in the weirdest places. At one point I frequented a forex chatroom (the one @ forex4noobs.com) where some traders insisted on linking any old sequence of points up and then calling them Wolfe Waves - no disrespect to my old friend Harald if he ever reads this, your "H Waves" work . I think for anyone who wants to trade trendlines, both on price and on the indicator, there'll be quite a learning curve until they consistently get it right as a chart unfolds.

[dakuki]

Yes I had same reservations about volume on Empty4. Ninjatrader much better for that. Drawing trendlines to me is kind of obvious. But will double check whenever you post charts here.

Please compare different OBV indis when you get a chance. I like LHDTs one.

[slipshod]

Ok definitely will do, thanks

Right, onto the next part - using Open Order data, specifically Oanda's as they provide such pretty pictures. Basically I add it to my Empty4 chart by drawing appropriately coloured rectangles. Then I can see how the price wants to behave near them without having to constantly go back & forth to various webpages...

audm30-ex2.png

What this now gives us is a narrative. If we see clear evidence of strong orders entering to stop a retracement against the prevailing trend, in this case to stop the upward surges in the face of downward momentum, then it gives us more confidence to take that short signal & conversely to be extra careful about jumping into longs when there's a sideways pause.

I should point out that not all pairs have such nice-looking open order charts. EURAUD, for instance, has basically nothing on it - probably because its kind of a derivative pair that doesn't have big money pushing it around directly like the USD pairs do. Where there's no strong open order patterns, you could choose either not to trade at all, or just take the plain old SAFT entries

[slipshod]

Having had a look at the alternate OBV indicators dakuki mentioned, I initially leaned in the direction of the Gadi one as it seemed the most accurate of the 3, however its useless due to a zero divide error. Without volume smoothing the LHDT is identical to the standard Empty4 indicator so its only with smoothing on that its worth looking at, and it is an improvement on the standard Empty4 indicator.

However there's a 3rd alternative, "enhanced" OBV, which I discuss and analyse in relation to the others in this post at the bottom of this page: http://www.stevehopwoodforex.com/phpBB3 ... 642#p21642. I'll be using the "enhanced" one until I find reason not to.

A couple of things I should mention - I do not at all like the idea of smoothing the OBV indicator itself, whichever one is used, as firstly it makes the peaks & troughs off which trendlines are drawn difficult to distinguish, and secondly it'll make the indicator lag. Neither do I like the idea of adding an MA to the indicator & trading crosses of it, as OBV trendline breaks seem much more precise and reliable.

[slipshod]

So, what's left in the way of intro? Yes - timeframes. Anything from H4 to M5, depending on which seems to show the best trendlines on both the price & the OBV at the time. I'll consider trading Gold as well, but only on H4 to avoid the dreaded noisy spikes. A $20 whipsaw over the course of half an hour, which Gold can do easily, may not seem like much until you realise that its the equivalent of 200 pips!

I use 2 entries, one with a TP near the next concentrated buy/sell open order area or near a zone of previous support or resistance, and one to trail in case it turns into a larger move.

Entry sizes - I started with each order being 0.2 lots, may increase that a bit next week if trading's going well. No recovery, as I don't believe in blowing up my account if I'm wrong. Stop loss about the same distance away as the TP of my non-trailing entry.

As I live in Australia I'll be trading the Asian and Euro/London sessions, and the early New York when a bout of insomnia hits.

[slipshod]

I hit a few divide by zero errors with the Gadi OBV, and as its sourcecode isn't available there's nothing I can do to correct it. I suspected that it was based on the "enhanced" OBV documented at http://www.tradejuice.com/Swing-trading ... rap-mk.htm, and not finding an Empty4 implementation I created one myself (attached at the bottom of this post).

Then it was a matter of scientifically comparing the standard OBV, LHDT's volume-smoothed OBV, and the enhanced OBV ...

obvcompared.png

I suggest viewing the above image in a separate window, as there's 14 different price swings in there I want to analyse in terms of how well, if at all, the OBVs guided entries and exits.

Trend A: All three perform well and get you out when the price looks like it wants to spike upwards. Unfortunately they'd all get you into a failed long as well.

B: Trend A turned into a longer downtrend. For some reason both the standard and LHDT's OBV went up while the price dropped, being of no help whatsoever and keeping you out of a lucrative short. The Enhanced OBV was flat which at least wouldn't prohibit an entry. Towards the end of the trend all 3 got their act together and went down, then took you out at the same time just before the price broke its trendline.

C: LHDT did best, signalling a potential end to the uptrend before the others. The Enhanced OBV came in 2nd, while the standard OBV only got you out after the price trend broke.

D: Standard OBV fails entirely, going up again while the price drops. LHDT is sideways and confused, again no help. Enhanced is flawless, getting you in a nice short.

E: Standard continues to fail but finally gets you in towards the end of the trend. LHDT handles this ok. Enhanced is still in the downtrend from D, and signals a potential end to it before the others.

F: All work well on this one.

G: Confusing price action as we have a whiplash consolidation - notice I've drawn both green and orange trendlines on the price to reflect this. The initial short is handled well by all 3, probably better by enhanced, but only the enhanced would work for the ensuring jitter keeping you looking for shorts only. There's a lovely gap there illustrating why its best not to hold over weekends!

H: All handle this well.

I: Again all handle this well - enhanced gets you out a bit quicker than the others.

J: All handle well

K: All handle well, again enhanced gets the exit right earlier.

L: Standard fails, with a flat movement that doesn't help. LHDT is better, although unlike the others it has this a continuation from the J downtrend - this is good except the J trend then breaks necessitating drawing of the new L trendline. Enhanced works well.

M: All do well, though LHDT is a bit laggy getting you out.

In summary, it seems the standard OBV has flaws where-in it diverges from price action for no apparent reason, and this can confuse a trader & cost them pips. LHDT out-performs it, and Enhanced beats them both imho. Bear in mind though that this is just one chart sample - others could tell an entirely different story. For now I'm going to use the Enhanced OBV until I see reason not to, but LHDT with smoothing on is entirely acceptable as well.