Re: ProFx 7 - Server based bot based on machine learning

#21
andrew1 wrote:
Sat Apr 17, 2021 2:26 am
I am unsure that either of those conditions apply here and this deserves more investigation:
Capsmart is also with IC Markets (mine and his are both live accounts) - his took the trade mine did not.

1) What is the slippage (10) stated in points or pips? (I assume points ?)
2) Is there any debugging output in the case of slippage or other decision not to trade ? (if not there should be - other "server based" EA's I have do this so you can see what you missed and why).
Hello Andrew1.
Probably it will be a good idea to track all incoming ticks from your brokers with others, normalize them and then make your own conclusions.

1) It is in points.

2) No, because slippage is not a bug but it might be a good idea to add some info to the journal. I agree. Also, keep in mind that the max slippage input does not mean there can be no higher slippage. It simple means that the trade will be not executed when the expected slippage is higher than the input value. Entries are based on market execution and brokers do not guarantee execution at the actual tick.

I hope this helps.
Alexander

Re: ProFx 7 - Server based bot based on machine learning

#23
capsmart wrote:
Thu Apr 15, 2021 6:09 pm
And this is the first trade of the week on April 12th. This is similar to April 15th's trade. I agree with the general direction (long) but again the trade was taken several bars after the signal on the 15th but following the signal on the 30min and was closed early, when all indicators were pointing up.
15m.PNG30m.PNG
This looks like a momentum trade to me. Opened shortly after London open and closed when the momentum slowed down. Momentum trades are usually closed when one or more of the following conditions are met:

1) Price slows down.
2) Volume drops.
3) Prices comes close to support/resistance level.
4) Price speeds up massively (parabolic price action).

The actual trade is a good example:
GBPUSD long @ 1.39386 [mom]
Close partial [1]
SL [be]
Close partial [2]

momentum_19042021.png
momentum_19042021.png (12.55 KiB) Viewed 1864 times

Re: ProFx 7 - Server based bot based on machine learning

#24
Hi Alexander
Thank you for your reply.
Apparently this is today's trade (Apr 19th) and I agree with you that it looks fine on the 1min chart.
I am attaching the 1m chart with all the indicators so that you can explain me the specific reason the bot closed 50% of the trade. From what I see none of the above mentioned reasons applies (at least not according to the indicators).
I can understand that by closing a 50% of the trade it assures a winning trade overall. This is great. It also moved the stop loss of the remaining position to break even which is also great. I really like this approach. I am now waiting to see when the remaining position will be closed.
Capture.PNG
Capture.PNG (89.2 KiB) Viewed 1861 times

Re: ProFx 7 - Server based bot based on machine learning

#25
capsmart wrote:
Mon Apr 19, 2021 2:22 pm
Hi Alexander
Thank you for your reply.
Apparently this is today's trade (Apr 19th) and I agree with you that it looks fine on the 1min chart.
I am attaching the 1m chart with all the indicators so that you can explain me the specific reason the bot closed 50% of the trade. From what I see none of the above mentioned reasons applies (at least not according to the indicators).
I can understand that by closing a 50% of the trade it assures a winning trade overall. This is great. It also moved the stop loss of the remaining position to break even which is also great. I really like this approach. I am now waiting to see when the remaining position will be closed.
Capture.PNG
Hello capsmart.

Indicator are useful on higher time frames but they are not helpful for managing existing momentum trades. Price, Volume, S/R and price speed provide information in real time while indicators give us one snapshot every minute. That's more than enough for regular trades but for momentum trades it is often not.

I hope this helps.
Alexander

Re: ProFx 7 - Server based bot based on machine learning

#26
Thank you Alexander,
Capture.PNG
Capture.PNG (66.16 KiB) Viewed 1773 times
So the last trade was a momentum trade. Unfortunately the bot did not move the break even stop loss, following the 60 pip up move and waited for price to move 70 pips down and hit the stop loss at break even. But a profit is always a profit.
Since you are working on an bot update what do you think about notifications when trades are opened and closed?
I would personally need only the email but I think pop-up, mobile notification and email go together.
cron