DOSI: Daily Open Strategy Improved | Page 5 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Let's Improved this Strategy, Grid based System 1 reply

Improved Rsi Needs Code 5 replies

Creating a Simple Daily Open Price Strategy EA 1 reply

Improved Rsi Needs Code 1 reply

Attachments: DOSI: Daily Open Strategy Improved
Exit Attachments

DOSI: Daily Open Strategy Improved

  • #81
  • Oct 20, 2021 5:09pm Oct 20, 2021 5:09pm
I am truly looking forward to hear your results... My first attempts are well not that encouraging

Just a question... wouldn't an adjustable lot exponent value allow us to tweak the DD a bit more? At least that has been a way for me to control things in both directions on other EA's.

/T
  • #82
  • Oct 22, 2021 2:07pm Oct 22, 2021 2:07pm
Especially when using a low MaxOpenTrades, there can be periods where the EA has scaled-in maximally and is just waiting for the market to return. All that time we are in drawdown and not generating profit.

In version 1.70 the ConcurrentBuyAndSell trade direction option was added. In that mode, two instances of Dosi are started. One instance will only trade long, the other only short. Both instances trade concurrently, fully independent of eachother. They each have their own take profits, stop loss, and maximum drawdown.
  • #83
  • Oct 23, 2021 5:41am Oct 23, 2021 5:41am
Quoting amargam
{quote} I agree. There are some pairs that makes good DD, about 3-7% with minimum 85% winner trades in 10 years at least. Of course the most important parametres to “play” and find good dd and good balance are: Pips Body Size and TP. There are some pairs that if increases lotsize, good DD goes to 35-40% DD. So the key is search good pairs with its own parametres. im only testing it on daily tf, but Im sure that in lower tfs there are good opportunities
Hi Amargam,

Would you mind sharing what pair and settings you see such good DD against? I am struggling right now with pretty high DD's. Would be awesome to get some good of decent baseline to work with for futher tweaking.

Thanks in advance!

/T
  • #84
  • Oct 31, 2021 2:51am Oct 31, 2021 2:51am
Quoting yoriz
{quote} Thank you for the suggestion. The 217-page read, and even post #1 is rather intimidating. Quickly skimming over that thread, it looked like it needed a lot of discretion and manual judgement, or am I wrong? Obviously, only a strategy with clear black/white rules can be automated in an EA. Can you point me to a post in that thread that compactly summarizes the strategy?
I can't praise the 217-page strategy.
Please excuse the bad English via Google Translate.
  • #85
  • Nov 2, 2021 4:32pm Nov 2, 2021 4:32pm
I justed uploaded version 1.75 (see post #1) that adds the input parameters CloseAllTradesOnFriday, FridayCloseHour and FridayCloseMinute that force closing all trades before the weekend at the specified time. Note that the time is specified in the broker time zone (in other words: ignores the BrokerGMTOffset input parameter).
  • #86
  • Nov 2, 2021 8:45pm Nov 2, 2021 8:45pm
Quoting yoriz
I justed uploaded version 1.75 (see post #1) that adds the input parameters CloseAllTradesOnFriday, FridayCloseHour and FridayCloseMinute that force closing all trades before the weekend at the specified time. Note that the time is specified in the broker time zone (in other words: ignores the BrokerGMTOffset input parameter).
Mate, as usual, many thanks for your hard work and your valuable time. Your contribution to the community is priceless.

With Respect,
Bill
  • #87
  • Nov 3, 2021 4:12pm Nov 3, 2021 4:12pm
In version 1.80, I added the MaxOpenBuyTrades and MaxOpenSellTrades input parameters. These can overrule MaxOpenTrades to create an asymmetric configuration if you suspect the market to be bullish or bearish. If both MaxOpenTrades and MaxOpenBuyTrades/MaxOpenSellTrades are non-zero, the latter will take precedence.
  • #88
  • Nov 3, 2021 8:29pm Nov 3, 2021 8:29pm
Quoting yoriz
In version 1.80, I added the MaxOpenBuyTrades and MaxOpenSellTrades input parameters. These can overrule MaxOpenTrades to create an asymmetric configuration if you suspect the market to be bullish or bearish. If both MaxOpenTrades and MaxOpenBuyTrades/MaxOpenSellTrades are non-zero, the latter will take precedence.

Hi Mate:

If I want to turn off the MaxOpenTrades, and only use MaxOpenBuyTrades/MaxOpenSellTrades exclusively, what value should I set on MaxOpenTrades? Zero?

Regards,
Bill
  • #89
  • Nov 4, 2021 3:47am Nov 4, 2021 3:47am
Quoting bdunn
Hi Mate: If I want to turn off the MaxOpenTrades, and only use MaxOpenBuyTrades/MaxOpenSellTrades exclusively, what value should I set on MaxOpenTrades? Zero? Regards, Bill
The Buy and Sell variants take precendence over the MaxOpenTrades parameters. In other words, if you set the Buy and Sell parameters to a non-zero value, the value of MaxOpenTrades is ignored and not used, but for clarity for yourself you might want to set it to zero. If you do not want specialized Buy and Sell settings, you can leave the Buy and Sell parameters at zero, and only use MaxOpenTrades.

Example symmetric usage:

Inserted Code
MaxOpenTrades     = 7
MaxOpenBuyTrades  = 0
MaxOpenSellTrades = 0

Example asymmetric usage:

Inserted Code
MaxOpenTrades     = 0
MaxOpenBuyTrades  = 5
MaxOpenSellTrades = 8
  • #90
  • Nov 4, 2021 5:29am Nov 4, 2021 5:29am
Quoting yoriz
{quote} The Buy and Sell variants take precendence over the MaxOpenTrades parameters. In other words, if you set the Buy and Sell parameters to a non-zero value, the value of MaxOpenTrades is ignored and not used, but for clarity for yourself you might want to set it to zero. If you do not want specialized Buy and Sell settings, you can leave the Buy and Sell parameters at zero, and only use MaxOpenTrades. Example symmetric usage: MaxOpenTrades = 7 MaxOpenBuyTrades = 0 MaxOpenSellTrades = 0 Example asymmetric usage: MaxOpenTrades = 0 MaxOpenBuyTrades...
Well noted, mate.

Thanks for the clear explanation and your great work.
  • #91
  • Edited 11:09am Nov 5, 2021 10:41am | Edited 11:09am
I received PMs from 3 different users that all seem very confident in the performance of the yzDosi EA. Nice to see that you guys like it, but I like to issue a warning. At least in my own experiments, it appears the current version of yzDosi is not profitable in forward testing. Indeed, you can curve fit the most fantastic equity curves but they all break in forward testing.

I suggest you try the following:

1. Pretend you discovered this EA one year earlier! 2021 has not happened yet. Do not use a single tick from the 2021 data. Only work with 2020 data! Ideally, you would throw away the 2021 tickdata from your MT5! You can actually do that using offline charts and importing CSV files. Myself, I have an MT5 instance for backtesting which lags 6 months on purpose.


2. Optimize your settings on the data prior to November 2020. Up to you to use the previous 6 months, 2 years, 10 years, whatever. But do not use any data after November 1st, 2020!


3. Do not "check" or "verify" the settings you found against 2021 data! That will give selection bias, and affects you psychologically. Really pretend it is the year 2020!


4. Do this for all the symbols and timeframes you intend to run yzDosi on. Again, DO NOT LOOK AT THE 2021 DATA. Not even once!


5. Based on the portfolio of EAs found, pretend you forward test the EAs in demo from November 1st, 2020 to December 31th, 2020. In reality you would open a demo account at a broker and forward test for 2 months. In this case we can simply backtest the date range November to December 2020. My guess is, that you will probably already see some EA configurations not performing as well as they did during optimization!


6. Now compile the definite portfolio of EAs based on the results of the "demo" period. If it was Jan 1st, 2021, which EAs would you have used going live? Remember, your money is on the line and you would really have gone live with these EAs last year! Don't say after step 7 below, "mwah, come to think of it, I probably would not have included this and that EA configuration in my portfolio". No, you choose NOW before running step 7. Be honest to yourself. Really, what EA configs are you going to use to run on your live account during the year 2021?


7. Ok, time to invest your precious money and "go live"! Only now may you look at the 2021 data! Run a backtest on the date range January 2021 until now with the actual starting balance that you would have invested. Can you now buy a new Tesla, or do you need to sell your house? ;-)


If you follow the approach above, you will see that unfortunately the current yzDosi EA is not profitable in a forward test.
Please prove me wrong!
  • #92
  • Nov 5, 2021 9:06pm Nov 5, 2021 9:06pm
Quoting yoriz
Please prove me wrong!
With no offence, I tend to believe it is your EA which is not well coded and as a result gives bad forward test results, and not the strategy itself, because in the original thread (DOS) the founder shows there from time to time his ongoing updated success with his strategy/expert-advisor in his live account on several pairs.

Why don't you ask him for help?
  • #93
  • Edited 6:01am Nov 7, 2021 5:27am | Edited 6:01am
Quoting MT4-EA
With no offence, I tend to believe it is your EA which is not well coded and as a result gives bad forward test results, and not the strategy itself,
No offence taken ;-) We confirmed the EA behaves exactly like the original strategy if you configure it like that (see post #45 and #46), so I don't think the problem is in the code.

Quoting MT4-EA
because in the original thread (DOS) the founder shows there from time to time his ongoing updated success with his strategy/expert-advisor in his live account on several pairs.
Unfortunately, exactly this is the psychological risk with averaging down, grid or marti strategies which have a very asymmetric skewed risk profile. The equity curve will look like a sawtooth with a constant nice steady increase, but then occasionally it will have a very deep drawdown. If you are lucky the strategy will recover in time and you only see a deep downward "spike" in your equity curve but it continues its path, but if you are unluckly it will hit the MaxDrawdownPercent too often and you lose all gains of the last months. I am not saying a sawtooth equity curve can not generate profits (see below), but you need to consider a long time range to know the overall performance, not just those few months you are on the rising part of the sawtooth.

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot 2021-11-07 at 13.01.04.png Size: 43 KB

What you see on ForexFactory is often people forward testing for a few months and proudly showing their success. Check back with them after two years! Excuses like "black swan" events, manipulation by the broker, etc. are often used to justify why the equity curve took a hit after so many months of success.

Forward testing is unnessarily slow. To see whether a strategy is viable, you can follow the approach outlined in post #91 and know after one weekend of hard work. For strategies with a more balanced risk profile (e.g. one trade at a time, fixed risk per trade, no "recovery trades", etc.) forward tests can be more representative of performance, but also here it makes more sense to use the outlined approach to get insights in the performance quicker and span a larger date range with many more trades.

But please prove me wrong! Please post here your settings that are profitable during 2021, that you found by optimizing only using the year 2020 and earlier. If so, I like to learn your approach to choose the most robust setting from the many Strategy Tester's results.
  • #94
  • Nov 8, 2021 6:29pm Nov 8, 2021 6:29pm
But this is exactly what happens.
The founder of DOS shows his live results, using set of parameters of his own.
His set did not come out of nowhere, but he got them probably from optimizations and walk-forward which he must have executed prior to his account's initiation.
Why do you think he did not do exactly what you well explained earlier about in-sample and out-of-sample?
  • #95
  • Nov 9, 2021 3:03am Nov 9, 2021 3:03am
Quoting MT4-EA
Why do you think he did not do exactly what you well explained earlier about in-sample and out-of-sample?
Fair enough. Indeed, I don't know. He seems to trade 11 symbols: "All major pairs except CHF and NZD pairs.", so that suggests he trusts he found stable settings for many pairs. His last update was Juli 17th, so not sure what happened since. I hope he is still successful and did not hit a deep drawdown.
  • #96
  • Nov 9, 2021 5:22am Nov 9, 2021 5:22am
Hi,

one question referring to the main DOS thread... I still don't get what Smilee meant by "If Open price of current daily bar is not behind last entry - don't open a consecutive trade" does he refer to the gap?

Thx
  • #97
  • Nov 9, 2021 5:59am Nov 9, 2021 5:59am
Quoting yoriz
{quote} His last update was Juli 17th, so not sure what happened since.
September I believe.
https://www.forexfactory.com/thread/...4#post13706754
  • #98
  • Nov 9, 2021 6:44am Nov 9, 2021 6:44am
I just want to say a few words:

Yoriz did a very good kind job by converting Smilee's strategy into MT5 EA. There is nothing wrong or fault on Yoriz's part or his conversion of Smilee's idea into MT5 codes.

If Smilee has anything to say on Yoriz's EA for not following his original trading ideas, Smilee is more than welcome to voice out here.

In case you don't know, there are two MT4 EAs coded on original thread by other kind members here. If you want something different, pls try those versions and use whichever you feel comfortable with. Or, just trade manually or hire the coders that you trust.

If anyone has questions on this DOS original strategy, pls direct the questions to Smilee for the best answer.

This thread is open for the purpose of improvement, evolution and changes for better results and endings.
  • #99
  • Nov 9, 2021 8:59am Nov 9, 2021 8:59am
Quoting MT4-EA
September I believe. https://www.forexfactory.com/thread/...4#post13706754
Ah, indeed. Still going strong in September!
  • #100
  • Nov 9, 2021 9:04am Nov 9, 2021 9:04am
Quoting lorecavoli
Hi, one question referring to the main DOS thread... I still don't get what Smilee meant by "If Open price of current daily bar is not behind last entry - don't open a consecutive trade" does he refer to the gap?
When in drawdown, only open an additional trade when the market moved even further away from the open price of the first trade. The goal is to pull the average price (of all open trades combined) closer to the actual price of the market. That way, the market needs to revert less strong to reach breakeven than when only sticking to the single opening trade. The minimal gap between the last opened trade and the market must be at least MinScaleInPips.
0 traders viewing now