Automated Trading Theory Discussion

[apfx]

In my opinion:

1. If in your system you can lock the ranges in 1 hour chart
2. instruct your system not to look more than 1: 1.3 Risk/ Reward
3. use volatility parameters.
4. automate the whole thing.
5. use good money management.
6. You'll be good.

apfx

[permanentjaun]

apfx,

What do you mean by "lock in the ranges in a 1 hour chart?"

[astro]

Here's a tip:
MA cross systems don't work

Save your time and spend it on something else.
The MA systems always look good on the last 1,2 or 3 months you test them, but overall they will generate more losses than winnings.

Try to design something that is based on patterns recognition around key levels. Better results are guaranteed!

Astro

[permanentjaun]

You so sure about that astro?

Try setting up 4 hour chart with 2 MA's. Make both of them 40 WMA, but one off the close and one off the open. 40 may not be the best period, but it's just an example. That's a simple MA cross. Proper money and risk management could make the profitable. Do you agree? Perhaps entering with multiple lots and exiting at progressive profits, such as 20, 50, 70, and finally when a SL or new signal is generated.

I would appreciate if you could go into more detail about pattern recognition around key levels though. I still need convincing that a computer could be programmed to correctly recognize a pattern of all variations. The computer can easily tell when one MA is greater than another. Patterns are more than just 1 > 0. Matt

[astro]

Do a little research on the Technical Trading forum.
You will find dozens of MA cross systems threads.
21emas, 34smas, 100wmas etc etc.
They all work sometimes. The problem is that when they fail, they will eat up all your profits.
Think about it for a moment. If trading was that easy why is everybody still looking for a system? Why wasn't it found yet? How come nobody discovered your 40 wma system yet?
Entering multiple lots can make things only worse! You are talking about risk management? Tripling the amount of lots is tripling the risk!!! If trading one lot is risking 2% of the account for example, then entering 3 lots is risking 6%!!!

As to pattern recognition, I think there are some charting software out there that can identify 1-2-3 patterns which are one of my favourite patterns, especially when they appear around key levels (round numbers, fib level, etc).
It's very hard to program such a thing, but your brain can do it quite easily.

By the way, here's a thought for you:
We all know that MA is lagging, which means that by the time you see the cross, price has already moved quite a distance. Why not use chart patterns in order to identify the trend at the beginning? Many trends start after a reversal (1-2-3 for example). Enter when you see one! So now, when price crosses your MA then you know that you're on the right track and you are already in!
Check my attached chart from yesterday: during asian session, price has formed a small 1-2-3 top (which was, to be honest, not a typical one) and reversed. There was an entry opportunity right after the UK market opened (on the 15 min chart), so I shorted @ 8790. Look where you would go short using the MA cross: 8730! that's 60 pips away!!!

You're right. It's not as easy as "if price > ma then...". Recognising the patterns is very discretionary (in this example many would not classify it as a 1-2-3 top) but it gives you the advantage of getting in earlier and catching a larger portion of the move with stops at a logically better place!

Hope this helps
Astro

[Trevman]

one more final MA system ive just found http://www.adamforex.com/

in reference to astros comments, i think price action may have some indication as to direction but i don't think it is as efficient as it is in the stock market (could be wrong)

[permanentjaun]

Originally Posted by astro

Do a little research on the Technical Trading forum.
You will find dozens of MA cross systems threads.
21emas, 34smas, 100wmas etc etc.
They all work sometimes. The problem is that when they fail, they will eat up all your profits.
Think about it for a moment. If trading was that easy why is everybody still looking for a system? Why wasn't it found yet? How come nobody discovered your 40 wma system yet?
Entering multiple lots can make things only worse! You are talking about risk management? Tripling the amount of lots is tripling the risk!!! If trading one lot is risking 2% of the account for example, then entering 3 lots is risking 6%!!!

All because everyone hasn't found it, doesn't mean it doesn't exist. In fact on another forex forum there is a thread which essentially is exactly what I just described. They claimed very good results, but they were manually trading. I am not sure if anyone got it running automated. For that reason, I won't say it would work. I believe it very well could.

3 lots can be as big or small as you want them to be. I know GFT allows for mini 10K lot sizes even in standard accounts. You have to ask to do this and fax them signed papers, which I've done. I wanted the advanced charting options while still controlling a mini account. Only required a larger deposit. So 3 lots do not have to be 6% of your account.

Even so, is it really 3 times the risk? It's about money management. The thread on the other forum describes it in much better detail, but I'll try to sum it up quickly here. Open a position with 3 lots. Have a hard stop loss say 20 pips from your position. All 3 lots are closed if it hits 20 pips, therefore -60 pips total. Say your play runs however and you take profits at 30 and 50 pips. Then you let the last lot run till there is a signal to reverse position, i.e. a MA cross, or it hits a stop at +30 pips which would the currency moves to your +50 position. The sequence of possible scenerios begins with -60 pips if the play fails. Then -10 if it gets to the first lot profit of 30 and drops to your hard loss. Then +110 if it hits the 30 and 50 profit taking areas and drops to the new stop loss of +30 pips. Last situation is 80 + X where X is an amount where a new MA cross signals the end of the run at the top of the run. It could be 30 pips, it could be 500 pips. It depends on the chart time frame and length of run.

Disregard the best case scenerio and focus on the other 3. If the currency does what you say just once and you make 110 pips it already makes up for one trade that hit -60 and another that banks only -10. 110 vs. -70. You're still in the profit of 40 pips. If you hit the worst case scenerio twice then yes it will lose you 10 pips if you only hit the possible positive scenerio just once. -120 vs 110.

What you'd be playing off of though is how often do you think your system would predict good situations? The cross has to be significant to mean the trend is changing. What if your system is correct in picking the +110 scenerio 65% of the time, -60 20% of the time, -10 10% of the time, and a best case scenerio where you make 100 pips off the last lot for a total of a +180 on the play only 5% of the time. Your expected pip P/L out of 100 trades would be +6750 pips.

Are those percentage numbers unreasonable? 70% win, 30% loss. Lets change the percentage numbers. Let's say +110 happens 40%, +180 is still just 5%, -60 is 40% and -10 is 15. The expected pip P/L out of 100 total trades is still +2750. In this case you're expecting to be right only 45% of the time and wrong 55% of the time. Would you agree those are terrible numbers to be right? This also assumes that the best case scenerio is capped at 180. It could also find plays for 280, 380, if you play a 4 hour chart and it trends well. Entering 3 lots, using good money management, is just another tactic at maximizing profits. 100 trades would take a long time to develop on a 4 hour chart however. Even using all the majors it could take weeks. 2750 pips may not be enough for most people and the system would be dropped. This is why I started the thread to try and find the efficient MA and chart time. Perhaps looking at a 2 hour chart instead of a 4 hour would present the same pip potential, but more signals to make it more efficient.

Originally Posted by astro

By the way, here's a thought for you:
We all know that MA is lagging, which means that by the time you see the cross, price has already moved quite a distance. Why not use chart patterns in order to identify the trend at the beginning? Many trends start after a reversal (1-2-3 for example). Enter when you see one! So now, when price crosses your MA then you know that you're on the right track and you are already in!
Check my attached chart from yesterday: during asian session, price has formed a small 1-2-3 top (which was, to be honest, not a typical one) and reversed. There was an entry opportunity right after the UK market opened (on the 15 min chart), so I shorted @ 8790. Look where you would go short using the MA cross: 8730! that's 60 pips away!!!

MA's do lag. This is why I like the idea of forming MA's off of the same period but using open for one, and close for the other. In doing so the MA's are uniform and do not diverge very much. It only takes a candle or two for them to cross. This is unlike using a 20/50 MA cross where they diverge tremendously and may not correct signal a reverse for as many as 20-30 candles. So lag can be overcome using the correct ones in the correct chart timeframe.

[craig66]

Well put Astro, I have been trying to tell permanentjaun the exact same thing on the Stratergy Builder forums.