(binned per thread starter’s request) MM (Money Maker) Detective | Page 320 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Money management: Capital needed per 0.07 USD per microlot 0 replies

The Volume Detective 411 replies

Attachments: (binned per thread starter’s request) MM (Money Maker) Detective
Exit Attachments

(binned per thread starter’s request) MM (Money Maker) Detective

  • #6,381
  • Aug 31, 2019 5:26am Aug 31, 2019 5:26am
Dear all,
A comment earlier in the thread warned against using this on lower timeframes. Why? Because this is a range based system,. and if you scroll down enough, there is always trend.

And trends with a certain slope are very deceiving when using LWMA centered channels and viewing historical data. However, the same principles will apply to all timeframes, high or low. The market is fractal. Here are two pictures from strategy tester. loaded with BDX, 4HR, 100%. One picture is real_time, triggering many entries, in strategy tester, and as it appears live, and one is after-the fact rendering of the channel on historical data. Somewhat shocking if you are not used to working with centered MAs and extrapolated data.

Perhaps this could be fixed with a slope filter on the LWMA? Essentially turn of the system and avoid trades when there is a certain slope over x number of bars. Thats the classical manual technique, and with this wonderful automation, should be possible to include.

That would also enable to have the system running on lower time frame also, and shutting itself down when the slope was too steep. (and possibly activate a trend following system, but thats another story)
Attached Image(s) (click to enlarge)
Click to Enlarge Name: post bdx render.png Size: 34 KB
Click to Enlarge Name: real time bdx render.png Size: 51 KB
  • #6,382
  • Aug 31, 2019 5:45am Aug 31, 2019 5:45am
and here is on a radically different timeframe

real-time, BDX would trigger sell-entries on the pink dots.
After the fact, on historical data, it looks very different
Attached Image (click to enlarge)
Click to Enlarge Name: post-fact BDX.png Size: 32 KB
Attached Image
  • #6,383
  • Aug 31, 2019 6:18am Aug 31, 2019 6:18am
@Tetra,its been mentioned many times,can you post a full chart,we all want to be on the same page and see what we are looking at,what pair is it??? cheers...
  • #6,384
  • Aug 31, 2019 7:12am Aug 31, 2019 7:12am
Quoting Tetra144
Dear all, A comment earlier in the thread warned against using this on lower timeframes. Why? Because this is a range based system,. and if you scroll down enough, there is always trend. And trends with a certain slope are very deceiving when using LWMA centered channels and viewing historical data. However, the same principles will apply to all timeframes, high or low. The market is fractal. Here are two pictures from strategy tester. loaded with BDX, 4HR, 100%. One picture is real_time, triggering many entries, in strategy tester, and as it appears...
Hi Tetra144.

Just some thoughts: I dont think its a centered LWMA it should be a centered TMA. Take the indicator FAST TMA line and lay it over with period of 66. It matches nearly exactly. Do the same with a nrp centered TMA (LWMA). It is a range based system yes. But it tries to give a probability for a reversal. The signal is outside the adr-bands and therefore most of the time part of a strong move (trend). Every slope indi lags price, so there would be very little (quality) signals. But why not firstly running some test with it and send the statistics to the creators of this strategy? When the edge increases.
I think theres no need to add anything. Its perfect kiss.

Cheers and fun,
Klausi
Stick to the plan!
  • #6,385
  • Aug 31, 2019 7:41am Aug 31, 2019 7:41am
Problem with continuous Open / Close of trades...

I think / hope the problem has been solved. The following post will announce a new release. Thanks for your feedback and help in reporting the problem and providing details.

Using @Blackeagle's chart as an example, the problem arises when there is a price spike such that the same bar includes BOTH an Entry and Exit signal. It's an easy thing to fix if you know to look for it in advance -- which I didn't. Below is an example of what I mean:

Attached Image (click to enlarge)
Click to Enlarge Name: Same-Candle-Enter-Exit.png Size: 43 KB


The BDX strategy includes a rule that there should be only one trade per bar. And to enforce this the code would check for any open orders on the current bar and if there weren't any then it would permit a new trade to open. Or if there already was an open trade then it would ignore the Entry signal. However, if the trade were to be closed then we could find ourselves in a "ping-pong" situation of repeatedly opening AND closing until a new bar arrives. So, the solution is to also check for any closed orders on the current bar. That is, if the current bar has been used for either opening an order or closing an order then it is thereafter ignored. So, in the current version (19242) I added a check for closed orders after the check for open orders. Or so I thought. Turns out there is the initial check for open orders followed by another check for open orders. I was probably cutting / pasting code, got distracted, and then forgot to change the second check to closed orders. I think this is why the problem has persisted even though I had claimed it was solved.

I apologize and would like a "do-over" for 19242. In the next post I will announce the release of version 19243 which really does check for both open and closed orders on the current bar.

As for why some people experience the problem and others don't, it's hard to say. I think it is more likely to occur on H1 because the ADR offset from the midline is smaller and it would be more common for a spike to be able to span the distance from an outer band to the midline.

If you find the problem still persists for you, then please take a screen shot with the BDX indicator overlayed. We would be looking for a bar that includes both a dot (Entry) and 'x' (Exit) signals.
  • #6,386
  • Aug 31, 2019 8:11am Aug 31, 2019 8:11am
I've just now updated the software for the BDX strategy...

R4M-BDX-ADR.ex4 -- this is the indicator work horse. The EA's get their signals from the indicator.
R4M-BDX-DashboardEA.ex4 -- dashboard EA for manual trading
R4M-BDX-EA.ex4 -- single pair, auto-trading EA
R4M-BDX-AllPairs.ex4 -- multi-pair, auto-trading EA

I've uploaded new versions of the BDX software -- one indicator and 3 Expert Advisors. This is now version "2.00.19243". The software can be downloaded from the other thread -- "Custom Indicators for TMA-ADR Reversal Trading and BBMA" (https://www.forexfactory.com/showthread.php?t=920925). All indicators are bundled in a single *.zip file. Indicators must be installed under MT4's 'Indicators' folder and EAs must be installed under the 'Experts' folder. To avoid version skews, please make sure that all indicators and EAs come from the same *.zip file.

Changes:

1. I think this version fixes the bug where a price spike that spans an outer band and midline (and which includes both an Entry and Exit signal) could cause trades to be repeatedly opened and closed.

2. I've restricted supported symbols to the 28 most common forex symbols -- no metals (e.g. gold, silver), energies, commodities, exotics. The BDX strategy's magic sauce is the ADR and it's not clear to me that ADR is as reproducible in other instruments as it is in forex. The following forex symbols are supported:

AUDCAD, AUDCHF, AUDJPY, AUDNZD, AUDUSD, CADCHF, CADJPY, CHFJPY, EURAUD, EURCAD, EURCHF, EURGBP, EURJPY, EURNZD, EURUSD, GBPAUD, GBPCAD, GBPCHF, GBPJPY, GBPNZD, GBPUSD, NZDCAD, NZDCHF, NZDJPY, NZDUSD, USDCAD, USDCHF, USDJPY

Finally -- please tread carefully with this latest version. It is straight out of the oven and only minimally tested with Strategy Tester (since it's the weekend and the market is closed). I think most people understand, but it is still worth repeating -- I do not claim this is a polished product (yet). In return for sharing the software then my expectation is others will jump in and help test it and flush out the bugs. And I will then fix the bugs. In the end it's a win-win arrangement.
  • #6,387
  • Aug 31, 2019 8:14am Aug 31, 2019 8:14am
Quoting robots4me
...... So, the solution is to also check for any closed orders on the current bar. That is, if the current bar has been used for either opening an order or closing an order then it is thereafter ignored. ....
Just a question. For users who set MaxOpenTrades > 1, let's say on H1, we have 3 open positions of the same currency from different prices. If one close per bar is permitted, assuming one of the closing conditions is met (either HMA transition or TPs of all 3 positions (3 positions will have different TPs) hit on the same bar ), is EA going to wait for the next bar to close the remaining two positions?

Regards,
Blackeagle
  • #6,388
  • Aug 31, 2019 8:33am Aug 31, 2019 8:33am
Quoting Blackeagle
{quote} Just a question. For users who set MaxOpenTrades > 1, let's say on H1, we have 3 open positions of the same currency from different prices. If one close per bar is permitted, assuming one of the closing conditions is met (either HMA transition or TPs of all 3 positions (3 positions will have different TPs) hit on the same bar ), is EA going to wait for the next bar to close the remaining two positions? Regards, Blackeagle
@Blackeagle -- an Exit signal closes all open positions for that MagicNumber (i.e. symbol + timeframe + EA).
  • #6,389
  • Aug 31, 2019 10:38am Aug 31, 2019 10:38am
If I remove the older version Ea and install an newer version, the new Ea will recognize the current open trades if the magic number is unchanged right?
  • #6,390
  • Aug 31, 2019 11:00am Aug 31, 2019 11:00am
Quoting robots4me
..my expectation is others will jump in and help test it and flush out the bugs. And I will then fix the bugs. In the end it's a win-win arrangement.
Brilliant! Crowdsourcing.

To the point, here are two tests, ´42 version that you see placed multiple orders on these points:
Attached Image(s) (click to enlarge)
Click to Enlarge Name: EU42.png Size: 116 KB
Click to Enlarge Name: AJ42.png Size: 89 KB
  • #6,391
  • Aug 31, 2019 11:01am Aug 31, 2019 11:01am
and ´43 version that did not. seems to have been fixed. Brilliant!
Attached Image(s) (click to enlarge)
Click to Enlarge Name: AJ43.png Size: 127 KB
Click to Enlarge Name: EU43.png Size: 160 KB
  • #6,392
  • Aug 31, 2019 11:37am Aug 31, 2019 11:37am
Dear All,
Just to follow up on the other points above. I encourage you all to run this in strategy-tester, on a low time frame, i. 5 mins, with a narrow band. Just watch it and observe the logic. The idea is that the 5 min is a microcosmos of the bigger market, the same types of swings, ranges, spikes and all happen there over a window that is practical for tick based testing.

Its quite interesting, and I believe telling of what we will see on 1hr and 4hr, daily - over time.

My findings from having run 6 pairs (and I don't think it matters which ones as these dynamics are the same) are basically that the system shows marginal profitability over time (lots of trades and much commission paid).

Some lessons:
If the balance goes red, do not stop the system as it will recover. If you see good profit, consider stopping the system and wait until the trading zone has shifted. The bottom line on all the 6 pairs on 5 min tick data for the last three weeks is marginal profitable, I think thats somewhat random depending upon when I stopped it. Over time Its might swing around BE, or hopefully, show small net profit consistently.

I did not get significant difference with a wider band.

I did loose more with tighter stop-loss. For now it seems to be most profitable when loosers are cancelled when price hits the mid-line.

It earns most when there is a reasonable flat range.

It looses most in (i) when it starts placing (buy) orders in the first consolidation zone after a downturn, and (ii) when placing sells in the first consolidation in an upswing. It of course earns big in the peaks (and deep bottoms) of these swings, but this is mostly cancelled out from the corresponding losses. I believe the net profits are ultimately seen from the flat ranges.

I think it is worth contemplating some other techniques in addition to this. Some type of directional bias as is customary in other strategies. i.e higher MA, higher TF etc. The automation of the multipair mechanics here is really really impressive, replaces what humans just cannot execute consistently, and is worth building on.

Would be happy to contribute to testing other biases, but cannot access the buffer data from these indicators and I don't feel like decompiling this which would be like scraping the paint of a beautiful Michelangelo.

And finally, a centered MA is not "repaining" as such, but this version shows, and the also takes trading decision based upon the last 32 points on the right - and these last points are extrapolations - and will be recalculated for every new bar (until they become point number 33).

As everything else, it is a compromise, but just be aware that the channel we see when scrolling back the charts is not how it appeared real time. As such, visual back-testing is very deceiving. When it is flat, it is fine, but the swings are quite different. If you overlay this with a non-centered MA you will see a better representation of the channel as it appeared real time.

For now
  • #6,393
  • Aug 31, 2019 12:01pm Aug 31, 2019 12:01pm
Quoting Tetra144
Dear All, Just to follow up on the other points above. I encourage you all to run this in strategy-tester, on a low time frame, i. 5 mins, with a narrow band. Just watch it and observe the logic. The idea is that the 5 min is a microcosmos of the bigger market, the same types of swings, ranges, spikes and all happen there over a window that is practical for tick based testing. Its quite interesting, and I believe telling of what we will see on 1hr and 4hr, daily - over time. My findings from having run 6 pairs (and I don't think it matters which...
Dude, you are so May time period. This is not place for you.
  • #6,394
  • Aug 31, 2019 12:52pm Aug 31, 2019 12:52pm
Quoting Zkk
If I remove the older version Ea and install an newer version, the new Ea will recognize the current open trades if the magic number is unchanged right?
@Zkk -- yes, that is correct.
  • #6,395
  • Aug 31, 2019 1:28pm Aug 31, 2019 1:28pm
May is a great month of the year!

Anyways, at that time, there wasn't a fully functional machine testable algo.
And this thing is a thing of beauty. Hats of to @Robots4me

Now lets get the data speak and se if the few parameters available can be optimized. Running six simulations simultaneously now. The big TFs will take time to complete. Conventional and unconventional parameters.
All running AJ (somewhat randomly chosen)

Here is the first 5 min sim. ADR13pct, SL13. Marginally better than Sl at mid-line.
Attached Image (click to enlarge)
Click to Enlarge Name: AJ 5M_ADR13_SL13.png Size: 149 KB
  • #6,396
  • Aug 31, 2019 5:46pm Aug 31, 2019 5:46pm
This is absolutely fascinating stuff!
Running 1hr sims on several pairs. Very similar patterns.
There is something to use here for sure.

The 120 pct stop losses are a problem. They have wipeout previous months gains on both. GPB, Eg and EJ. and drawn them into deep red. Models with shorter stop losses perform better. Also, more narrow range seem to give more trades and lots to carry profit to the center line and builds more capital.

Profit-ratios is lower than I expected. Less than 50 across several sims.
Last one from me attached, three sims on 1hr with variations of channel depth and stop loss.

@majamivice sorry to have bothered you dude.

I will leave you to it.
thanks and take care!

Attached Image (click to enlarge)
Click to Enlarge Name: aj 1hr various.png Size: 349 KB
  • #6,397
  • Edited 7:38pm Aug 31, 2019 6:57pm | Edited 7:38pm
Quoting Tetra144
This is absolutely fascinating stuff! Running 1hr sims on several pairs. Very similar patterns. There is something to use here for sure. The 120 pct stop losses are a problem. They have wipeout previous months gains on both. GPB, Eg and EJ. and drawn them into deep red. Models with shorter stop losses perform better. Also, more narrow range seem to give more trades and lots to carry profit to the center line and builds more capital. Profit-ratios is lower than I expected. Less than 50 across several sims. Last one from me attached, three sims on...
You will make lots of money with this strategy IF YOU KNOW WHAT YOU ARE DOING!
  • #6,398
  • Aug 31, 2019 9:16pm Aug 31, 2019 9:16pm
Quoting Simontw
{quote} Ok fair enough. Thanks. Simon
Quoting robots4me
{quote} @Simontw -- also I found another situation where I didn't see a blue dot. But after I removed the HA indicator then I could see it. The HA transitions are used for computing Entries -- if you don't trust the algorithm then overlay the chart with the HA indicator. Otherwise I prefer to see the MT4 bands -- that's the real PA. HA candles are approximations -- smoothed and delayed.
Hello guys,

That's because the HA indicator actually "prints" over the signal dots, so it can often cover most of their area.
I think those dots should appear on foreground above everything else in the chart.

Have a nice weekend!
  • #6,399
  • Sep 1, 2019 4:17am Sep 1, 2019 4:17am
Quoting majamivice
{quote} Dude, you are so May time period. This is not place for you.
Could you explain what you mean? This is not place for what?
  • #6,400
  • Sep 1, 2019 4:34am Sep 1, 2019 4:34am
Quoting ARSpips
{quote} Could you explain what you mean? This is not place for what?
@ARSpips -- what he means is that while everyone else is on page 320, Tetra144 is still on page 22. At least Tetra144 got the message. I'm guessing you are even further behind. Please don't post unless you have something of value to add or have a relevant question.
0 traders viewing now