How do you test the reliability of broker executions

Shakone

Senior member
Messages
2,460
Likes
665
This applies I suppose to forex mainly, but spreadbetting or something off exchange could be included.

Suppose I am trading with a broker, and I am long and have a limit order at 1.2799 and 1.2800 to exit, and price rapidly moves directly on my screen to 1.2804, based on the brokers bid-ask, and yet my execution is delayed for a second (noticeable) and is then filled at 1.2799 and 1.2800.

How does one know whether this is just a fair execution, or whether my internet or computer slowed in those seconds or whether you've been ripped off a few pips?

I remember that certain brokers were fined for things like this, but how could one provide evidence to report it.Keep track of asymmetric slippage?
 
This applies I suppose to forex mainly, but spreadbetting or something off exchange could be included.

Suppose I am trading with a broker, and I am long and have a limit order at 1.2799 and 1.2800 to exit, and price rapidly moves directly on my screen to 1.2804, based on the brokers bid-ask, and yet my execution is delayed for a second (noticeable) and is then filled at 1.2799 and 1.2800.

How does one know whether this is just a fair execution, or whether my internet or computer slowed in those seconds or whether you've been ripped off a few pips?

I remember that certain brokers were fined for things like this, but how could one provide evidence to report it.Keep track of asymmetric slippage?

Can only speak from my own experience and i'm not even sure it's relevant to your question.
Some time ago I was running a no stop strategy which used limit orders for entry and regardless of fill price I then set a fixed number of points target and again used limit orders for exits. What i found was that in fast moves when price was approaching a limit close, I always got the exact price or better. When I did my totting up on the whole trade series (approx 4 months worth) the positive slippage amounted to roughly half of all spread and transactional costs. So, not to be sniffed at !

If you were using stops, you would get negatively slipped just the same if not worse.
 
depends on what exactly the order you have submitted is.

If it's a plain limit order, then what you get is what you should expect.

If it's a stop-limit order, where 1.2799 needs to be traded (or quoted) before the 1.2799 limit order is submitted (similarly for the 1.2800 orders), then you might have a case if the price moves to 1.2804 before your limit order is submitted due to latency etc.
 
This applies I suppose to forex mainly, but spreadbetting or something off exchange could be included.

Suppose I am trading with a broker, and I am long and have a limit order at 1.2799 and 1.2800 to exit, and price rapidly moves directly on my screen to 1.2804, based on the brokers bid-ask, and yet my execution is delayed for a second (noticeable) and is then filled at 1.2799 and 1.2800.

How does one know whether this is just a fair execution, or whether my internet or computer slowed in those seconds or whether you've been ripped off a few pips?

I remember that certain brokers were fined for things like this, but how could one provide evidence to report it.Keep track of asymmetric slippage?

You automate don't you?
Run a sim account parallel to the live account.
Set the sim to immediate fills and disable partial fills.
Keep track of the difference in excel.

If you want to run dedicated slippage tests,
use the same method with chump change and overtrade.
Yeah you will get caned with the spread, but as its chump change,
you build a decent sample size to gauge slippage, for small outlay.

As CV and Hakuna said, dependent on order type as well.
 
Hakuna, it is a limit order to sell placed well in advance, so already on their side, so I should be filled at the price or better. If it moves rapidly, there is typically some slippage, as CV says, maybe half a pip or whatever. If it jumps straight through to 1.2804 then it should be filled thereabouts. Of course it is possible it moved smoothly to 1.2799, 1.2800, 1.2801 all the way to 1.2804, and my internet or computer was slow. But on my screen it moved directly to the price, it paused for over a second for the order to be filled, and then filled at the limit order price. Which is no disaster, but it does make me wonder.
 
Shakone, never knew you used NT.
Check tools > options > data tab.
In realtime data section, see if following are ticked:

Save chart data as historical;
Record for market replay.

If you have any recorded feed data it will be in the following folder for replay:
C:\Documents and Settings\PC name\My Documents\NinjaTrader 7\db\data

Assuming you have the data, run your algo on the replay data at 500X with immediate fills and no partial fills.
That will give you execution prices with no slippage.
If you don't have replay data, you might be able to hack historical tick
data into .ntm files (replay files), not sure if thats possible though, never done it.
Replay is preferable though as NT will use the bid/ask correctly then.

Add comms for replay under the options > commissions > replay section - and only the minimum.
If you set levels or use the instrument manager to set comms and levels it
screws everything up with FX, NT is still primarily designed around futures contracts.
That should change with NT8.
 
Cheers LV. Record isn't currently checked. I sometimes record data for market replay, for testing purposes, but not every day. I'll try what you suggest.



Shakone, never knew you used NT.
Check tools > options > data tab.
In realtime data section, see if following are ticked:

Save chart data as historical;
Record for market replay.

If you have any recorded feed data it will be in the following folder for replay:
C:\Documents and Settings\PC name\My Documents\NinjaTrader 7\db\data

Assuming you have the data, run your algo on the replay data at 500X with immediate fills and no partial fills.
That will give you execution prices with no slippage.
If you don't have replay data, you might be able to hack historical tick
data into .ntm files (replay files), not sure if thats possible though, never done it.
Replay is preferable though as NT will use the bid/ask correctly then.

Add comms for replay under the options > commissions > replay section - and only the minimum.
If you set levels or use the instrument manager to set comms and levels it
screws everything up with FX, NT is still primarily designed around futures contracts.
That should change with NT8.
 
If you don't have replay data, you might be able to hack historical tick
data into .ntm files (replay files), not sure if thats possible though, never done it.

Checked, .ntm files are unicode, i.e. encoded.
Not worth the hassle.
 
Let me know which broker it was, which platform, which account type and what date this happened (day, hours, minutes) - assume it was EURUSD. I'll tell you what happened with 99% certainty.
 
Let me know which broker it was, which platform, which account type and what date this happened (day, hours, minutes) - assume it was EURUSD. I'll tell you what happened with 99% certainty.

Thanks for the offer, but I'm more interested in the 'how to', so that I can do these things myself, rather than someone giving me an answer.
 
Now i am to present a model broker which will have some good quality.Actually I am with a broker that is Forex-metal. They give me the following facilities 35% first deposit bonus;free automatic trading signals ; ECN accounts with low spreads ;Leverage up to 1:500. It is possible to get up to 1:1100, not just 1:500. 1:1100 leverage customers can get through the floating leverage feature inside the trader's room on Forex-Metal website.

No thanks. I get nose-bleeds with any broker offering higher than 50:1 leverage.
 
Top