|
4 registered members (AndrewAMD, VoroneTZ, Quad, 1 invisible),
837
guests, and 5
spiders. |
|
Key:
Admin,
Global Mod,
Mod
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: Sphin]
#470167
12/30/17 18:08
12/30/17 18:08
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
This works! After I stored some days of .t1 data, made my first tries in backtesting the 'Simple broker arbitrage example' to find out a suitable threshold and then tried a live demo - nothing happened. While printing the spreads to csv I found out that they behave quite differently from simulation with fixed spreads so it might be better to simulate with the variable spreads stored from live/demo environment. From the manual: Both .t1 and .t6 data can contain the current ask-bid spread, which is returned by the marketVal function. For backtesting with variable spread, set Spread = marketVal() in the script. But how to store the spread in .t1 so that is accessible with marketVal()?
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: Sphin]
#470232
01/04/18 02:01
01/04/18 02:01
|
Joined: Dec 2016
Posts: 71
firecrest
Junior Member
|
Junior Member
Joined: Dec 2016
Posts: 71
|
What do you mean with "not successful" - technically or financially? Didn't the script trade at all or weren't the trades profitable?
I use the 'Simple broker arbitrage example' actually only to record spread and price (there is no commission at the used brokers) to find out a practical threshold and to view the relations between these components in a real environment. I mean financially. Technically Zorro works great. "not successful" means that the Demo accounts in both brokers are making a loss instead of gain. I use higher threshold and no trades are triggered. If I use lower threshold, trades are not profitable and losses are limited. The differences between prices of the same asset are too small (about 2- 3 pips) for profit. It can work if only one has almost zero commission and connection speed to capture this. I think another way to work around this is to trade Indices instead of FX. I have not tried out indices yet but I think it maybe possible. I am sharing what I know to the community and hope this help.
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: Grat]
#470289
01/07/18 12:53
01/07/18 12:53
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
There are some things I cannot comprehend. For searching reasons of lost trades I added this line to the simple broker arbitrage example before any enterX() or exitX() command:
print(TO_LOG, "\nPriceA: %.0f, PriceB: %.0f, Difference: %.0f", PriceA, PriceB, Difference);
Setting slippage = 0 I expected the logged prices to be the same as the traded ones but:
[4: Tue 18-01-02 04:00:00] 66.73 0 1/3 (13605.60)
PriceA: 13069, PriceB: 13412, Difference: -343
[Asset_A::L0501] Long 100@13069.00000 at 04:00:00
[Asset_B::S0502] Short 1@13204.59961 at 04:00:00
[5: Tue 18-01-02 05:00:00] 66.73 -155 1/5 (13412.70)
[6: Tue 18-01-02 06:00:00] 66.73 -131 1/5 (13197.10)
PriceA: 13294, PriceB: 13231, Difference: 62
[Asset_A::L0501] Sell 100@13294.00: +162 at 06:00:00
[Asset_B::S0502] Cover 1@13431.50: -283 at 06:00:00
While the logged PriceAs are the same at long entry and sell exit, the differences of logged PriceBs to its short entry is 208 and to its cover exit -200. The deviation is so extreme that the exit condition is simply not met because PriceA (13294) minus 'traded' PriceB (13431) is -137 and therefore still lower than 0. So the negative result is no surprise. Any hint where those deviations can result from?
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: jcl]
#470428
01/14/18 11:59
01/14/18 11:59
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
I fear it was not or not completely. But the differences have been moved to PriceA/Asset_A compared to the previous version and sometimes the values are equal.
V 1.745 on Sun 18-01-14 11:43:05
Test: BA_BTC_A_B (TICKS) 2018
Assets AssetsArb.csv
PriceA: 13842.00000, PriceB: 14353.00000, Difference: -511.00000
[Asset_A::L0101] Long 10@13405.00000 at 00:00:01
[Asset_B::S0102] Short 1@14353.00000 at 00:00:01
[1: Tue 18-01-02 01:00:00] 0000 -93.73 1/1 (14353.00)
[2: Tue 18-01-02 02:00:00] 0000 +421 2/0 (13719.20)
PriceA: 13592.00000, PriceB: 13591.50000, Difference: 0.50000
[Asset_A::L0101] Sell 10@13592.00: +12.40 at 02:52:58
[Asset_B::S0102] Cover 1@13591.40: +509 at 02:52:58
[3: Tue 18-01-02 03:00:00] 522 0 2/0 (13605.60)
[4: Tue 18-01-02 04:00:00] 522 0 2/0 (13412.70)
PriceA: 13071.00000, PriceB: 13412.70020, Difference: -341.70020
[Asset_A::L0501] Long 10@13343.00000 at 04:00:01
[Asset_B::S0502] Short 1@13412.70020 at 04:00:01
[5: Tue 18-01-02 05:00:00] 522 +10.27 3/1 (13204.70)
[6: Tue 18-01-02 06:00:00] 522 -8.10 3/1 (13231.90)
PriceA: 13293.00000, PriceB: 13231.90039, Difference: 61.09961
[Asset_A::L0501] Sell 10@13116.00: -29.00 at 06:00:01
[Asset_B::S0502] Cover 1@13231.90: +20.90 at 06:00:01
[...]
[18: Tue 18-01-02 18:00:00] 514 0 3/1 (13964.50)
PriceA: 14648.00000, PriceB: 13964.40039, Difference: 683.59961
[Asset_A::S1901] Short 10@13914.00000 at 18:00:01
[Asset_B::L1902] Long 1@13964.40039 at 18:00:01
PriceA: 13914.00000, PriceB: 13964.50000, Difference: -50.50000
[Asset_A::S1901] Cover 10@13914.00: -6.30 at 18:00:01
[Asset_B::L1902] Sell 1@13964.90: -131 at 18:00:01
(Broker B has changed its minimum trade size to 0.1, so lot factor for Broker A is now 10)
Especially the last trade seemed very strange, Price_A had fallen from 14648 to 13914 within a second? So I asked ZHistoryEditor about Asset_A.t1 and found:
2018.01.02 18:00:09.000 13915
2018.01.02 18:00:01.000 13914
2018.01.02 18:00:00.000 13914
2018.01.02 17:59:58.000 13915
2018.01.02 17:59:57.000 13918
No idea where the value of 14648 came from. I compared randomly some further Price_A's from the log with the values of Asset_A.t1 and it was always like in the sample: the prices of Zorro's entries and exits are the ones that I also found in .t1. Anything seems sometimes to go wrong with setting PriceA = priceClose().
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: jcl]
#470646
01/27/18 12:19
01/27/18 12:19
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
The bid price is negative, so: I stored .t1 that way and my .t1 in ZHistoryEditor look like
2018.01.27 00:28:34.000 177,18
2018.01.27 00:28:34.000 -175,23
2018.01.27 00:28:31.000 177,2
2018.01.27 00:28:31.000 -175,23
Then I modified the code of the BrokerArb example with
asset("Asset_A");
Spread = marketVal(); // From Manual: For backtesting with variable spread, set Spread = marketVal() in the script.
var SpreadA = Spread, PriceA = priceClose(), CommissionA = ...
asset("Asset_B");
Spread = marketVal();
var SpreadB = Spread, PriceB = priceClose(), CommissionB = ...
But both SpreadA and SpreadB remain always 0.0000. Did I miss something?
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: jcl]
#471051
02/17/18 13:53
02/17/18 13:53
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
I have a strange problem with (normal) currencies (with crypto assets all works fine), I always get the same result of -600 of each phantom trade:
{GBPUSD_B1::l6801} Long 1@1.40057 at 20:08:34
{GBPUSD_B2::s6802} Short 1@1.40113 at 20:08:34
{GBPUSD_B1::l6801} Sell 1@1.40070: -600 at 20:08:40
{GBPUSD_B2::s6802} Cover 1@1.40065: -600 at 20:08:40
[1268: Mon 18-02-05 20:09:02] (1.40118)
{GBPUSD_B1::s6901} Short 1@1.40179 at 20:09:10
{GBPUSD_B2::l6902} Long 1@1.40119 at 20:09:10
{GBPUSD_B1::s6901} Cover 1@1.40176: -600 at 20:09:17
{GBPUSD_B2::l6902} Sell 1@1.40187: -600 at 20:09:17
[1269: Mon 18-02-05 20:10:03] (1.40141)
{GBPUSD_B1::s7001} Short 1@1.40187 at 20:10:12
{GBPUSD_B2::l7002} Long 1@1.40131 at 20:10:12
{GBPUSD_B1::s7001} Cover 1@1.40181: -600 at 20:10:14
{GBPUSD_B2::l7002} Sell 1@1.40222: -599 at 20:10:14
In the asset list GBPUSD_X looks identical to the other assets:
GBPUSD_B1,1.41249,0.00047,-0.4898,0.3292,0.00010,0.080288,2.8336,400.218,1000.0,0.600,B1_Arb:GBP/USD
GBPUSD_B2,1.54901,0.00006,0.07,-0.15,0.0001,0.09129,16,0,1000,0.6,B2_Arb:GBP/USD
Any idea where I can search for the reason?
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: jcl]
#471153
02/20/18 19:02
02/20/18 19:02
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
Can confirm both, EURUSD_B1 is identified as Forex and the issues in the Forex trades from above has been gone in 1.79.3. Okay, the resolutions of trade results are now up to 7 decimals although I did not modify the assets in the asset list to more decimals {EURJPY_B1::s5701} Cover 1@132.32: -0.0070341 at 05:14:11 but I don't mind about it. 
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: jcl]
#471250
02/25/18 12:03
02/25/18 12:03
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
I have 2 further issues (1.79.3): 1. The calculated spreads from the marketVal() function seem not to be correct or not always. This effects of course only test mode, in trade mode with the spreads coming from the brokers all is ok. From Log:
2018-02-23 08:42:49.604
AssetA: EURGBP_B1, PriceA: 0.88072, SpreadA: 0.00000, CommissionA: 0.00007
From .t1 with ZHistoryEditor:
2018.02.23 08:42:49.604 0,88072
2018.02.23 08:42:49.604 -0,88064
From Log:
2018-02-23 14:59:42.400
AssetA: EURGBP_B1, PriceA: 0.87982, SpreadA: 0.00004, CommissionA: 0.00007
From .t1 with ZHistoryEditor:
2018.02.23 14:59:42.399 0,87982
2018.02.23 14:59:42.399 -0,87974
From Log:
2018-02-23 15:05:34.602
AssetA: EURGBP_B1, PriceA: 0.88017, SpreadA: 0.00000, CommissionA: 0.00007
From .t1 with ZHistoryEditor:
2018.02.23 15:05:34.602 0,88017
2018.02.23 15:05:34.602 -0,88008
From Log:
2018-02-23 15:00:5.196
AssetA: EURGBP_B1, PriceA: 0.88020, SpreadA: 0.00008, CommissionA: 0.00007
From .t1 with ZHistoryEditor:
2018.02.23 15:00:05.195 -0,8801
2018.02.23 15:00:05.195 0,8802
From Log:
2018-02-23 15:49:25.268
AssetA: EURGBP_B1, PriceA: 0.87907, SpreadA: 0.00003, CommissionA: 0.00007
From .t1 with ZHistoryEditor:
2018.02.23 15:49:25.268 0,87907
2018.02.23 15:49:25.268 -0,87899 I coded one strategy on 2 ways (one with fixed asset calls, the other one with a loop) and received 2 different results while testing. Searching for an error I surprisingly found out that both variants did exactly the same trades, only their results differ. from the 1st:
[54: Fri 18-02-23 00:54:00] (0.88342)
[EURGBP_B1::S5501] Short 1@0.88331 at 00:55:00
[EURGBP_B2::L5502] Long 1@0.88317 at 00:55:00
[55: Fri 18-02-23 00:55:00] 0000 -0.88 0/2 (0.88296)
[EURGBP_B1::S5501] Cover 1@0.88335: -0.77 at 00:55:31
[EURGBP_B2::L5502] Sell 1@0.88332: +0.0563 at 00:55:31
[491: Fri 18-02-23 08:11:00] -0.72 0 1/1 (0.88162)
[EURGBP_B1::L9201] Long 1@0.88137 at 08:11:31
[EURGBP_B2::S9202] Short 1@0.88175 at 08:11:31
[EURGBP_B1::L9201] Sell 1@0.88154: -0.53 at 08:11:51
[EURGBP_B2::S9202] Cover 1@0.88151: +0.19 at 08:11:51
from the 2nd:
[54: Fri 18-02-23 00:54:00]
[EURGBP_B2::L5501] Long 1@0.88317 at 00:55:00
[EURGBP_B1::S5502] Short 1@0.88331 at 00:55:00
[55: Fri 18-02-23 00:55:00] 0000 -0.27 0/2
[EURGBP_B1::S5502] Cover 1@0.88335: -0.24 at 00:55:31
[EURGBP_B2::L5501] Sell 1@0.88332: +0.0563 at 00:55:31
[491: Fri 18-02-23 08:11:00] -0.18 0 1/1
[EURGBP_B1::L9201] Long 1@0.88137 at 08:11:31
[EURGBP_B2::S9202] Short 1@0.88175 at 08:11:31
[EURGBP_B2::S9202] Cover 1@0.88151: +0.22 at 08:11:51
[EURGBP_B1::L9201] Sell 1@0.88154: +0.11 at 08:11:51 How can I find out what is right or better: reliable?
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: Sphin]
#471422
03/03/18 10:51
03/03/18 10:51
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
Does the FXCM API support OrderLimit? If not is there any broker known that does? As far as I think to know something like this should not happen:
[1158: Fri 18-03-02 14:17:01] (1.37795)
2018-03-02 14:17:55.4555
Asset: GBPUSD_FXCM, Price: 1.37750, Spread: 0.00001, Commission: 0.00007, TickAge: 0.00002
OrderLimit: 1.37746
[GBPUSD_FXCM::S5254] Short 1@1.37694 at 14:17:55.
How to avoid it if OrderLimit is not supported? Setting any Entry away from the actual price only to open a pending trade and using a 'PriceLimit' in a TMF?
int TMF(var PriceLimit) {
if(!TradeIsOpen) {
if(TradeIsLong && TradePriceOpen > PriceLimit)
return 1;
if(TradeIsShort && TradePriceOpen < PriceLimit)
return 1;
}
}
Is this a viable way to simulate OrderLimit?
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: Sphin]
#471476
03/05/18 19:32
03/05/18 19:32
|
Joined: Dec 2013
Posts: 568
Fuerth, DE
Sphin
User
|
User
Joined: Dec 2013
Posts: 568
Fuerth, DE
|
Sorry for quite many forum entries but I have another issue. Stop, Profit, Trail, Entry, ...
[...]All parameters above must be set after selecting the asset and before calling enterLong / enterShort. They have no effect on different assets [...] Can it be that e.g. EURUSD_B1 and EURUSD_B2 are considered as one asset anyway? I have to reset Entry explicitly after changing asset from XXX_B1 to XXX_B2 or is this intended?
|
|
|
Re: Multiple broker connection - beta testers wanted!
[Re: Veratyr]
#472818
05/25/18 02:24
05/25/18 02:24
|
Joined: Dec 2016
Posts: 71
firecrest
Junior Member
|
Junior Member
Joined: Dec 2016
Posts: 71
|
I've been testing the beta with 3 assets across 4 brokers and I've found that Zorro can't keep up with the data.
I have the Zorro window and the MT4 status windows right next to each other and watch the price on Zorro's status bar. I notice that somewhat frequently, Zorro holds still for at least a second at a time, while I can still see MT4 ticking. Zorro even keeps its little asterisk while I see the price change.
I've talked to support and they've just told me to increase TickTime (which is currently 1) but I don't think this is a good option, as broker arbitrage is extremely latency sensitive. I guess this means that broker arbitrage is out of retail traders reach.
|
|
|
|
