Great technique for beginners...

[rollingstone]

Thanks Akash,

Software is Xtick, about $50 per month, I would use MT4 but I am probably one of the only people who doesn't get on with it, the bottom indicator is from a feature they call xpaint where you can paint the candles or the indicators with lots of different parameters, new crosses of EMA's, trendiing emas, indicator through zero etc, etc.

I find it very useful as a visual aid and when used with alerts.

I have started looking at the 5 minute chart when a 30 min KS signal is generated for extra clues, but no conclusions yet although the CCI helps especially if you have had a divergence on the 5 minute before a 30 min KS, but the one thing we don't want to do is complicate what is a very good simple system.

[rollingstone]

Heres an example of some added info we can use just to give more confidence, as I say I do not want to complicate the system and make us debate if to take a cross or not, but any little extra signs we can have the better.

Obviously we don't want to say we have to have this and that before taking the trade, the 20 EMA does a good job if we can get round the fact that it will give a signal sometimes, then the market will retrace and the signal will be gone.

So, when the final crossing of the KS line happened on Friday on the GBP we had a couple of extra little clues, 5 min chart below shows nice CCI divergence with a small double top coming back in from outside the bollinger band, and when the 60 minute cross happens the 5 min price is already well below the KS line.

As I say these are just extra clues not something to rely on.

The same can be seen on the 30 minute chart as well.

5 Minute chart

[rollingstone]

Heres the 30 minute chart showing the same things

[fxquest]

Akash,

This sounds like a great technique - thanks for sharing.

I was curious enough to program it into MT4 and backtest but the results were dissapointing ( the strategy lost money ). 30M with 20pip SL and TP.

What parameters do you give the Kinjen-Sen? (I was using the Kinjen-Sen from the Ichimuko with the default parameters (9,26,52) ). Also, since you are using MT4 why dont you use an expert to backtest and "forward" test it?

I am not an automatic system trader but I think that a good strategy should also backtest well.

[rollingstone]

imho this and almost every other system that is based on indicators cannot be backtested with accurate results, most back-testing is rubbish.

Questions you need to ask:

- Did you backtest the weekly forex period from the Sunday night opening till the friday night closing, if so this will give bad results, as stated the strategy is only for use 5.00 GMT to 17.00 GMT (i.e.when the market is active, all the little crosses that occur during the Asian quiet session will slew the results.

- Does the backtesting know when each trade was entered and exited, does it account for the trailing stops, does it account for letting the price pull back from the KS line to cover the spread.

- How does the backtesting filter out KS line crosses where the 20 ema is very close, these crosses are not to be taken as stated by Akash, the best signals are when the MA is furthest away from the KS line.

- How does the backtesting decide on the trend of the EMA. (it has to be in the correct direction for the trade to be taken).

- Did you program in moving the stop to +10 when the trade is at +15.

I'm not criticizing you, but back-testing in general, you can make it say anything you like good or bad, just look at some of the back-tested signal services that are around.

The only way to test this properly is by forward testing, and manually candle by candle back-testing trying to be as honest as possible.

The most important this to remember is the times that a system is trading, a lot of simple systems work well when traded at the correct times, but include the period after about 18.00 GMT to about 05.00 GMT when the market is generally very quiet and rangebound and the results are rubbish.

[fxquest]

Hi rollingstone,

Those are all good points. Please help me understand why you think manual back-testing better than auto back-testing ( I tend to think that auto-back-testing is as honest as it gets ).

As for the details - apart from the time of day and letting the price pull back from the KS line to cover the spread, the backtest includes all the details of the strategy (as I understood it from the thread). I ran the auto-optimizer inorder to find the best parameter combinations (changing the trailing stop etc. distance from EMA etc). but so far 1777 combinations all produced negative results.

I will add the time of day and KS pullback and re-test.

28/8/2005:
These are the results of the updated backtest:

Profit factor of 1.69, drawdown of about 1%, profit of about 3.5 % after 81 trades.

Tested best when the SL is 40 updated to +10 when hitting 15 and TS of 7 north of +17 pips.

One modification I made to the entry rule is to only take signals if the "Tenkan-Sen" is on the right side of the "cloud" (above for long, below for short). Without this modification my backtests barely broke even.

[fxquest]

Originally Posted by perching

Hello Akash,

The software that I currently am using is supplied by GFT, of which the Ishimoku indicator is in their bundle of indicators. Unfortunitly I can not overlay onto the main chart to observe the crossovers like on your chart.

Had you modified the orginal program to be able to overlay?

Thanks
Ed

Hello Perching,

It is possible to view the Ichimoku indicators on top of the charts in the GFT software:

Instead of selecting the Ichimuko, you can select the individual indicators and they will appear on the chart (e.g. Add studies - "Kijun-sen", "Tenkan-sen" and if you want to see the 'cloud' you can also add the "Senkou Span" indicator).

HTH,
fxquest

[Akash]

Originally Posted by fxquest

Hi rollingstone,

Those are all good points. Please help me understand why you think manual back-testing better than auto back-testing ( I tend to think that auto-back-testing is as honest as it gets ).

As for the details - apart from the time of day and letting the price pull back from the KS line to cover the spread, the backtest includes all the details of the strategy (as I understood it from the thread). I ran the auto-optimizer inorder to find the best parameter combinations (changing the trailing stop etc. distance from EMA etc). but so far 1777 combinations all produced negative results.

I will add the time of day and KS pullback and re-test.

28/8/2005:
These are the results of the updated backtest:

Profit factor of 1.69, drawdown of about 1%, profit of about 3.5 % after 81 trades.

Tested best when the SL is 40 updated to +10 when hitting 15 and TS of 7 north of +17 pips.

One modification I made to the entry rule is to only take signals if the "Tenkan-Sen" is on the right side of the "cloud" (above for long, below for short). Without this modification my backtests barely broke even.

Hi Fxquest,
Thanks for going through the trouble to program an expert for this system. The results are pretty baddd...I think that back testing is as honest as it gets, but there are a lot of events that a a trader can spot before entering a trade, which an automated system can not. I think that I want to forward test this strategy manually, because I do see a lot of potential in it. I would love it if you could find ways to tweak up this expert and let me know. Also, an expert won't be able to recognise a ranging market. Actually this system does work in ranging markets too, but the range has to be big, I don't know what you call a wide ranging market. What do they call it? Also, one other thing I forgot to mention here was that I don't take profit as soon as I have 20 pips, I wait for it to go higher and then slowly increase the stops. Once profit is beyond 20 I set stops to 15 pips and then just keep moving them. Did you take that in account? I think I would really love it more than anyone else, if you could develop an expert for this system, because I have soooo much to do, which requires for me to stay away from my computer.
Rockin' Tradin'
-Akash