Forex Factory (https://www.forexfactory.com/forum.php)
-   Trading Systems (https://www.forexfactory.com/forumdisplay.php?f=71)
-   -   My "Ultron" EA for GBPUSD H1 timeframe (https://www.forexfactory.com/showthread.php?t=840339)

Telac Apr 13, 2019 11:12am | Post# 1921

I think another thing we all forget sometimes it that market conditions over the past several month with all the Brexit chaos are surely not normal. I don't expect a normal market to return until this is all sorted out and then even a bit beyond. We may actually be curve fitting to filter and optimize under this circumstances. Eventually the big picture will again emerge.

SigmaEnigma Apr 13, 2019 1:38pm | Post# 1922

Despite the conditions and this most recent move, the PDV_106 EA is still quite profitable. If you run a 99% accurate back test from the beginning of the year to today, it still performs well. So while things may be rough locally and losses still occur and the EA sometimes buys right into a resistance zone and vice versa, overall it still makes more than that it loses. I'm confident it will continue to do well this year, even tho I agree right now things are a bit abnormal for the good ole' cable.

groper Apr 14, 2019 5:58am | Post# 1923

Despite the conditions and this most recent move, the PDV_106 EA is still quite profitable. If you run a 99% accurate back test from the beginning of the year to today, it still performs well. So while things may be rough locally and losses still occur and the EA sometimes buys right into a resistance zone and vice versa, overall it still makes more than that it loses. I'm confident it will continue to do well this year, even tho I agree right now things are a bit abnormal for the good ole' cable.
i posted the backtest results earlier and agree its profitable - however that does not mean it cannot be made better. Its already improved significantly from where it all began - are we to stop here OR do we keep improving it? I for one, would be more than happy to put real money on the table for the development to continue...

Paul - are you personally coding logic into this EA or is someone else doing it for you?

majamivice Apr 14, 2019 7:09am | Post# 1924

6 Attachment(s)
3 backtests this year so far. Much better with fixed SL then with trailing. Two are with with Tickstory Lite backtest and one with original data from IC Markets without downloading 3th party data. I think original data is better accurate when I compare to real trades. Not many trades to compare but I've found one there on Tickstory is not right at all but IC Markets accurate.
Tickstory with trailing:
Click to Enlarge

Name: Trail.png
Size: 130 KB

Tickstory with fixed SL 35 pips:
Click to Enlarge

Name: Fixed TP.png
Size: 129 KB

IC Markets original data:
Click to Enlarge

Name: Ic Markets data BT.png
Size: 48 KB

Tickstory backtest trade:
Click to Enlarge

Name: TS Trade.png
Size: 7 KB

IC markets original data backtest trade:
Click to Enlarge

Name: Ic original data trade.png
Size: 4 KB

Real trade:
Click to Enlarge

Name: Real trade.png
Size: 4 KB

It would be interesting to see the same baktest as mine on TDS and if you can compare with your real trades.

groper Apr 14, 2019 7:43am | Post# 1925

1 Attachment(s)
3 backtests this year so far. Much better with fixed SL then with trailing. Two are with with Tickstory Lite backtest and one with original data from IC Markets without downloading 3th party data. I think original data is better accurate when I compare to real trades. Not many trades to compare but I've found one there on Tickstory is not right at all but IC Markets accurate. Tickstory with trailing: {image} Tickstory with fixed SL 35 pips: {image} IC Markets original data: {image} Tickstory backtest trade: {image} IC markets original data backtest...
I backtest with TDS at 99.9% - the last live trade i could model to compare live result with back test i have to do with different broker data due to no data available from ICmarkets. Instead i used alpari pro ECN data to back test. On the 4th april we had a sell using PDV105 - which i was using at the time before Paul released 106. The live result was 2 points better on the entry, and 7 points better on the exit. The live result was actually better in both entry and exit - most likely due to ICmarkets having lower spreads than Alpari. Either way - the modelling is less than 1 pip difference to the live trade, the live trade with ICmarkets more profitable. Thats all the data i have until more is available as i havnt been testing this EA for very long and the last 10 days historical data isnt available yet to compare the most recent trades.

Back test trade above chart, real trade in history below chart;
Click to Enlarge

Name: Image1.png
Size: 291 KB

groper Apr 14, 2019 7:59am | Post# 1926

2 Attachment(s)
default PDV106 set file last 12 months @ 99.9% using TDS;
Click to Enlarge

Name: Image1.png
Size: 249 KB
Click to Enlarge

Name: Image2.png
Size: 278 KB

majamivice Apr 14, 2019 9:24am | Post# 1927

{quote} I backtest with TDS at 99.9% - the last live trade i could model to compare live result with back test i have to do with different broker data due to no data available from ICmarkets. Instead i used alpari pro ECN data to back test. On the 4th april we had a sell using PDV105 - which i was using at the time before Paul released 106. The live result was 2 points better on the entry, and 7 points better on the exit. The live result was actually better in both entry and exit - most likely due to ICmarkets having lower spreads than Alpari. Either...
My point is most people use dukascopy historical data as most reliably source when backtesting and that is clearly not the case when you compare with real trades on your broker. So backtesting in MT4 strategy tester with original data of your broker do just fine without the need to use TDS. That is just my opinion.

PaulDV Apr 15, 2019 2:55am | Post# 1928

2 Attachment(s)
3 backtests this year so far. Much better with fixed SL then with trailing. Two are with with Tickstory Lite backtest and one with original data from IC Markets without downloading 3th party data. I think original data is better accurate when I compare to real trades. Not many trades to compare but I've found one there on Tickstory is not right at all but IC Markets accurate. Tickstory with trailing: {image} Tickstory with fixed SL 35 pips: {image} IC Markets original data: {image} Tickstory backtest trade: {image} IC markets original data backtest...
Hmmm...not sure about using broker data myself, especially with the poor quality and mismatched chart data - too many possible causes for errors for my liking.

However, the results using a smaller TP and no trailing stop are pretty impressive...

TP=35, TSL=23/2:
Click to Enlarge

Name: Screenshot1.png
Size: 49 KB

TP=35, TSL=0:
Click to Enlarge

Name: Screenshot2.png
Size: 49 KB

Invermac Apr 15, 2019 3:08am | Post# 1929

Im using 35 tp and 70SL, no TS, in real account, the backtest with this parameters since Jan2018 to today with a 100usd account is very impressive, more than 40k of profit.

majamivice Apr 15, 2019 3:47am | Post# 1930

{quote} Hmmm...not sure about using broker data myself, especially with the poor quality and mismatched chart data - too many possible causes for errors for my liking. However, the results using a smaller TP and no trailing stop are pretty impressive... TP=35, TSL=23/2: {image} TP=35, TSL=0: {image}
Can you compare backtest's trades with real trades on your account if they match?

PaulDV Apr 15, 2019 4:04am | Post# 1931

{quote} i posted the backtest results earlier and agree its profitable - however that does not mean it cannot be made better. Its already improved significantly from where it all began - are we to stop here OR do we keep improving it? I for one, would be more than happy to put real money on the table for the development to continue... Paul - are you personally coding logic into this EA or is someone else doing it for you?
I am making all of the coding changes myself.

PaulDV Apr 15, 2019 4:05am | Post# 1932

{quote} Can you compare backtest's trades with real trades on your account if they match?
Unfortunately not as I haven't been running that long. Maybe somebody else who has been running since the beginning of the year could compare and let us know?

majamivice Apr 15, 2019 4:11am | Post# 1933

{quote} Unfortunately not as I haven't been running that long. Maybe somebody else who has been running since the beginning of the year could compare and let us know?
It's enough for you that one doesn't match (of those you have) that backtest with ducascopy data if you're running on IC Markets is not valid. Check last couple of trades if they match.

jdva Apr 15, 2019 5:15am | Post# 1934

{quote} Hmmm...not sure about using broker data myself, especially with the poor quality and mismatched chart data - too many possible causes for errors for my liking. However, the results using a smaller TP and no trailing stop are pretty impressive... TP=35, TSL=23/2: {image} TP=35, TSL=0: {image}
Many thanks Paul for your valuable BTs...! They do confirm my reservations about trailing stops in general.

Please may you tell me more about the logic/idea behind the Max Volume=25 trigger parameter at the beginning of a new candle.

Thx, jdva

PaulDV Apr 15, 2019 5:57am | Post# 1935

{quote} It's enough for you that one doesn't match (of those you have) that backtest with ducascopy data if you're running on IC Markets is not valid. Check last couple of trades if they match.
You're right: in the backtest with TDS, there are two trades on the 10th April, but in reality there was only one.

majamivice Apr 15, 2019 6:10am | Post# 1936

{quote} You're right: in the backtest with TDS, there are two trades on the 10th April, but in reality there was only one.
Is it with same setting backtest as you run on real account? Try now with same setting on other MT4 platform with just original data and compare if you get same trades.

PaulDV Apr 15, 2019 8:52am | Post# 1937

{quote} Is it with same setting backtest as you run on real account? Try now with same setting on other MT4 platform with just original data and compare if you get same trades.
I don't get any trades on the 10th, even though there was one...

Either way, that set of parameters looks pretty stellar!

SigmaEnigma Apr 15, 2019 1:04pm | Post# 1938

Just closed out the move that opened last week for -6. Not bad considering the swings it went through the past few days. I might wait for the price to move out of this current range before firing up the EA again.

Feex Apr 15, 2019 1:39pm | Post# 1939

TP = 35, TSL = 0 - this option is really interesting! Which is rather unexpected)

mrdfx Apr 15, 2019 2:27pm | Post# 1940

Very interesting and positive results using SL 70, TP 35 and no TS!

I never thought of trying an inverted TP & SL with this EA and it actually gives better results based on previous data.

Great work guys!


© Forex Factory