Algo trading Fx

Even when you trade by semi-automated method, strategies will lost profitability in duration by time, no matter which method are used ( discretionary, semi-automated, automated). When you trade by semi-automated method, did you know which transaction will be profitable? Regards Greg

1 Like

Great thread. What pairs and time of day are you limiting yourself to? Or is it 24/7, as long as your trades are identified by your algo?

Hey @dudebro, thank you for showing your interest in this thread! I trade EUR/USD and AUD/NZD on H4 time frame between 9pm EST and 9am EST. Initially it was 24/7 with one pair EUR/USD, however after reviewing 100 trades, realized that strategy works good between the time I have mentioned earlier. Demo and back test results showed that strategy works better on AUD/NZD than EUR/USD, hence added that recently. Trade time and pairs may change in future based on next 100 trades analysis.

Cool. Did you say what exactly your strategy is, or is that a secret?

Week#16

Wow! 4 months passed just like that. There was a glitch on the cloud server as a result my partial stop didn’t hit and I ended up losing more. It is now almost a month trading AUD/NZD. It seems to be performing well.

This week:

Till Date:

Journey so far:

Win rate of AUD/NZD vs EUR/USD in one month:

From what I can gather, you’re using a continuation pattern on a single timeframe. I suspect that you might not see much improvement over what you’ve already observed.

Have you considered factoring in multiple timeframes? At least one higher timeframe to ascertain trade direction might help.

@Quadpip

It’s kind of a continuation or I should say it works well in continuation. My strategy is to analyze 5 H4 candles. Let’s say the price is moving up, in this case my Algo looks for the last 2,3 or 4 candles to check the body size and direction. If bodies are positive with small wicks, then it takes a short position assuming the next candle will go down. The assumption is based on the fact that prices don’t move in up/down direction for long and take breaks before proceeding further.

I’m back testing my Algo to determine trends based on lower timeframe. However, it has not yet successfully identified directions. It is still work in progress. Actually, now your question has made me work on it more diligently and also explore a higher timeframe. BTW, I have always considered H4 as a higher timeframe and not thought about daily or weekly.

@dudebro , nothing is secret here, I have kind of explained my strategy in this post. There are other candle patterns that I consider however they all have the same base rule that prices don’t just go up up or down down. It has to take a break.

In addition, I really like that people are taking interest in this thread. I appreciate your questions & feedback.

Looking forward to more questions and feedback.

Week#17

Weekly circuit breaker hit on Monday and no trade was taken for this week. In retrospect, I realized that when I started Forex trading, I took the approach of a 5 year kid who wants to experience all that market has to offer. However, in between I got deviated and started to put lots of restrictions and forgot about my exploration. I guess now I’m back to the kid mentality. I know the market will not treat me as a kid. But we will see.So, the changes are…

  1. Now my trades will be on 30 min candles.
  2. Trend will be considered after setup is identified. Trend identification logic may need some refinement.
  3. No fixed stop. Yes, this one is hard. However, close trade the moment setup in which trade was taken no longer exists.
  4. Time based close. A trade can’t go more than 5 days.
  5. Without fixed stops my strategy has the potential of big losses. Hence only trade small lots. Lot size will reduce significantly if drawdown is more than 10%.
  6. Of course daily and weekly circuit breakers are there to protect my capital unless I manually intervene.

Algo took one trade based on the above changes today and that was the only winner this week.

This week:

Till Date:

Journey so far:

Week#18

This week Algo only took one trade that too ended in a loss. Oanda has updated their Trading Performance tool and it is not showing this week’s trade. So not posting weekly update as it is same as week#17.

One physiological update though, at one point I was around 200 pips down in the trade, however I controlled my emotions and did not close trade manually. To me, that’s a big win. I let Algo do the job which I have coded for.

2 Likes

congrats @s_kushwah :grinning: :+1: :clap:

1 Like

Week#19

Algo took three trades this week. All were winning trade. Not that significant gain as compared to the loss I had. However, it’s a gain.

This week:

Till Date:

Journey so far:

Thanks for the update. Hoping your success continues into this week!

Looking good!

Week#20

This was a mixed week for me. 3 wins and 2 losses.

One thing to observe is that Algo took one trade in demo but not in live. There was no error whatsoever in the logs. This makes me believe the price is not the same in live vs demo. Since calculations consider fractional pips, even small differences in price will affect execution.

Oanda performance tool is not showing today’s trade under performance however the graph is. So posting graphs this time.

This week:

Journey so far:

Week#21

Lots of trades this week. This is because I have added GBPJPY. So now I’m trading EURUSD, AUDNZD and GBPJPY. Oh boy! GBPJPY is very volatile.

AUDNZD is trading on H4 time frame (TF) with different lot size and other two on M30 TF with same lot size.

This week:

This Week by products:

Journey so far:

Week#22

Is this an effect of Friday the 13th? I have more winners than losers this week.

This week:

Journey so far:

Week#23

Wow! Is this my luck or my strategy is working? 2nd consecutive green week!

When I started this week my P&L was around $100. At the end of the week it is $95. Lets see how much time it will take to recover these losses, if it recovers. Now I’m trading 6 instruments out of which AUD_NZD is on a different strategy than others. Instruments: EUR_USD,GBP_JPY,GBP_AUD,NZD_JPY,USD_CHF,AUD_NZD

I’m excited…

Account Summary:

This week:

Journey so far:

1 Like

Just wow! I haven’t enjoyed reading a journey so much before. I study Computer Science at A-Level and for my project next year I too want to make a trading robot. You’re diligence and faith in the algo is inspiring, I don’t credit myself to be a great trader but I do have my own system that I use manually and it has proved profitable.

It sounds so much easier then it will be, but do you think it would be possible for me to copy over my current system into an algo?

Keep up the work man, it is truly a pleasure to follow this thread

Hey Kian McGrath,

Glad that you are enjoying this thread! To answer your question, everything becomes easy once we start doing things. You can for sure transpose your manual trading to an automated trading. There will be hurdles and technical glitches as I have pointed out below but they can be overcome by proper error handling and planning.

  1. Connection problem with API endpoint.
  2. Random errors like API not returning the number of items that are expected.
  3. One time because of a technical glitch I had a position open over the weekend.
  4. Today because of a glitch a trade wasn’t executed.
  5. Due to the disconnect of the stream my trade monitor failed and my losses surpassed the threshold.
  6. Now I have my Algo setup on cloud otherwise every now and then internet connection issues.
1 Like

@s_kushwah be patience :slight_smile: if your strategies show you profitability in back test and live results are similar with current back test, it is only time to recover loss :slight_smile:

@KianMcGrath if you want to trade by algo, you should consider rent a VPS.

2 Likes