FXCM Price Discrepancy

[FMtrader]

NT8 Forex charts display bid prices and I have just noticed when comparing a particular high on an NT8 chart with the bid price on the same high on FXCM's trading station (both their web based and PC based versions) that the bid price was different.

I was looking at a 5 minute GBPAUD chart and noticed a 3 pip variation beween NT8's bid price high and FXCM's bid price high. Both my NT8 chart and the FXCM web were connected to the same demo account. I did not have the PC version of Trading Station and NT8 running at the same time as I understand they use the same API and can conflict.
Nearby I also checked a swing low and in this instance the prices beween the two platforms were only 0.2 pips different.

I then did a quick check on NT7 and came up with fairly close agreement with NT8 with still the same problem compared with the FXCM Trading Station high.

Why is NinjaTrader showing different prices to FXCM's trading station on some bars when connected to the same account?

The reason I first noticed this issue was that I was stopped out of a demo trade at a price that according to NT8 wasn't hit. (Even after allowing for the bid/ask spread)

I am running NT8 RC2

Further Info: My live funded FXCM ECN account setup for NinjaTrader shows the same price issue as the FXCM demo account discussed above.

[NinjaTrader_PatrickG]

Can you tell me the exact time stamp of the bar you're referring to? Please also include the time zone which your PC / NinjaTrader is set to.

[FMtrader]

Can you tell me the exact time stamp of the bar you're referring to? Please also include the time zone which your PC / NinjaTrader is set to.

My PC is set to US Eastern time.

The NT8 time stamp on the 5 minute GBPAUD bar in question is 5.05am 19th October and the high is shown at 1.60459.

FXCM's Trading Station still shows the bid price for the same high as 1.60489. The time stamp on FXCM's platform is 8pm which is probably the server time in Sydney. (I live in Australia)

I then set up another FOREX demo account via NinjaTrader (instead of FXCM direct) and used that to log into FXCM's Trading Station. This appears to have connected me to a US based FXCM server running on US Eastern time. The time stamp on the bar is now 5am but the bid price high is still listed as 1.60489. A 1min FXCM chart still shows this same high.

NT7 disagrees slightly with NT8 and shows the same high as 1.60453 with a time stamp of 5.10am on a 5 min chart.

I have checked a number of other bars and forex pairs and generally NT8 and FXCM Trading Station agree closely with small discrepancies sometimes.

[NinjaTrader_PatrickG]

Thanks for the details.

In your FXCM connection, you can choose to use NinjaTrader's historical data servers or FXCM's. I was able to get NinjaTrader to match the high of the 5:05AM bar (1.60489) by switching to FXCM's historical data servers and reloading my historical data.

• Disconnect from FXCM
• Connections > Configure > Select your FXCM connection
• Disable 'NinjaTrader historical data server' > OK
• Reconnect
• Right-click on your chart and select 'Reload all historical data'

[FMtrader]

Thanks for the details.

In your FXCM connection, you can choose to use NinjaTrader's historical data servers or FXCM's. I was able to get NinjaTrader to match the high of the 5:05AM bar (1.60489) by switching to FXCM's historical data servers and reloading my historical data.

• Disconnect from FXCM
• Connections > Configure > Select your FXCM connection
• Disable 'NinjaTrader historical data server' > OK
• Reconnect
• Right-click on your chart and select 'Reload all historical data'

Thanks Patrick.

The plot now gets thicker! I use a number of non time based Forex charts which means that tick level data is required to build the bar. Whilst FXCM servers seem to efficiently handle historical requests for time based data they don't seem to be able to handle requests for tick based data beyond a few days. Even requests for a small amount of historical tick data take an inordinately long time to fill. As such one cannot build a longer term chart based on range bars for example. This means I must use NinjaTrader servers for historical data.
This then prompts the next question: Why is NinjaTrader's historical data for FXCM slightly different to FXCM's? Aligned with this question is why is your NT7 FXCM data slightly different to your NT8 FXCM data.

Based of my quick checks the data variations between the three sources are generally well less than a pip but the instance that spurred this thread was a 3 tick variation.

[NinjaTrader_PatrickG]

Thanks Patrick.

The plot now gets thicker! I use a number of non time based Forex charts which means that tick level data is required to build the bar. Whilst FXCM servers seem to efficiently handle historical requests for time based data they don't seem to be able to handle requests for tick based data beyond a few days. Even requests for a small amount of historical tick data take an inordinately long time to fill. As such one cannot build a longer term chart based on range bars for example. This means I must use NinjaTrader servers for historical data.
This then prompts the next question: Why is NinjaTrader's historical data for FXCM slightly different to FXCM's? Aligned with this question is why is your NT7 FXCM data slightly different to your NT8 FXCM data.

Based of my quick checks the data variations between the three sources are generally well less than a pip but the instance that spurred this thread was a 3 tick variation.

The NinjaTrader servers are from NinjaTrader Brokerage's live data which has different spreads than FXCM. This can result in slight differences in quotes.

Can you clarify what specific bar you're referring to where there is a difference in price between NinjaTrader 7 and NinjaTrader 8 while using the same real-time and historical connection?

[FMtrader]

Can you clarify what specific bar you're referring to where there is a difference in price between NinjaTrader 7 and NinjaTrader 8 while using the same real-time and historical connection?

Some examples I found in 30 seconds on GBPAUD

US Eastern Time
21st Oct: Swing Low at 8.25 am: NT8 = 1.59576, NT7 = 1.59579
21st Oct: Swing High at 4.50 am: NT8 = 1.60329, NT7 = 1.60331
20th Oct: Swing Low at 9.35 am: NT8 = 1.59411, NT7 = 1.59408
19th Oct: Swing High at 10.25am: NT8 = 1.60493, NT7 = 1.60502

As I said in my earlier posting the differences between NT7 and NT8 are small and I don't consider them to be a significant issue. It is however interesting that NT7 and NT8 are different at all.

The above values came from the same FXCM account and both used NinjaTrader for historical data.

[NinjaTrader_Brett]

FMTrader,

We have NT7 and NT8 historical data servers which record from an FXCM feed. FXCM feeds are poll based meaning they provide a snapshot at some polling interval of the current state of the ask and bid. There is no way to sync up the polling interval between different servers/clients and as such minor differences will occur between the two.

-Brett

[FMtrader]

FMTrader,

We have NT7 and NT8 historical data servers which record from an FXCM feed. FXCM feeds are poll based meaning they provide a snapshot at some polling interval of the current state of the ask and bid. There is no way to sync up the polling interval between different servers/clients and as such minor differences will occur between the two.

-Brett

Hi Brett

Considering that you record from an FXCM feed it is surprising the difference that can occur at times between historical data on FXCM's Trading Station and your historical data on NinjaTrader (7 or 8). On one GBPAUD 5 min bar this morning (Aust. time) about 50 minutes after the market opened for the week there was a 6 tick difference between FXCM historical data on their Trading Station and your data on an NT chart. The difference was enough to produce a short term swing high on FXCM but not on NT. Most of the other bars were much closer.

[NinjaTrader_Brett]

Thanks, I'll take a look. Likely expected simply due to the polling nature of the API, since we're talking subpips here and depending on how much the markets are moving I could expect seeing that kind of a difference from time to time. If it was multiple pip differences that's likely a different story though since were talking a potential difference in price inside of a second. If its possible market would move that fast inside a second then its liking the polling difference between FXCM's own servers and NinjaTrader.

-Brett