ForumSections

Trade Server Problem - page 2

New comment
Ryan L Johnson #:

Oh, I see. This is a horse of a different color.

This is the functional equivalent of a distributed high frequency trading strategy. Most broker-dealers expressly prohibit HFT. Definitely check with your broker-dealer... assuming that you even want to confess this to them.

We are way far from an HFT context.

Operations generates an opening/day, rarely two.
parragaCelebre parraga #:
We are way far from an HFT context.

Operations generates an opening/day, rarely two.

I understand that you have 200+ EA's sending the same order to the same broker-dealer within milliseconds of each other.

If this is correct, is it not possible that your broker-dealer's execution algo interprets it as HFT?

Edit: If your clients are sending large orders in that manner, they could be eroding each other's liquidity. For example, institutional traders have to "create" their own liquidity for large orders by entering opposite orders first.
Ryan L Johnson #:

I understand that you have 200+ EA's sending the same order to the same broker-dealer within milliseconds of each other.

If this is correct, is it not possible that your broker-dealer's execution algo interprets it as HFT?

Edit: If your clients are sending large orders in that manner, they could be eroding each other's liquidity. For example, institutional traders have to "create" their own liquidity for large orders by entering opposite orders first.

Forum on trading, automated trading systems and testing trading strategies

Trade Server Problem

Ryan L Johnson, 2025.12.02 17:03

I understand that you have 200+ EA's sending the same order to the same broker-dealer within milliseconds of each other.

The only precise metric is that we send all orders in a maximum of 10s time space.

Although we could control it better apllying an intentional staggered throtling system wide, there are some counter points to that

 - Would be considered treating clients differently (comercial wise bad)
 - The issue was reported by clients outside our system
 - The market volume we represent isn't that significant (yet?)

Forum on trading, automated trading systems and testing trading strategies

Trade Server Problem

Ryan L Johnson, 2025.12.02 17:03

If this is correct, is it not possible that your broker-dealer's execution algo interprets it as HFT?

Wouldn't the broker respond with any kind of more objective kind of log?

Forum on trading, automated trading systems and testing trading strategies

Trade Server Problem

Ryan L Johnson, 2025.12.02 17:03

Edit: If your clients are sending large orders in that manner, they could be eroding each other's liquidity. For example, institutional traders have to "create" their own liquidity for large orders by entering opposite orders first.
That would make sense if we represented a bigger slice of the whole cake, I assume.

Lets put it in numbers

Lets say we're talking about 7k contracts

Average of 80k/min in this market.

Forum on trading, automated trading systems and testing trading strategies

Trade Server Problem

Alain Verleyen, 2025.12.02 13:41

You need to talk with your broker. This 158 seconds is enough information for them to provide you answers about it.

You are losing your time here, only the broker can answer.
Alain Verleyen #:
You are losing your time here, only the broker can answer.
I appriciate yours and all answers nonetheless.

Will contact the broker asap.
That is a serious issue, especially the 158-second execution time! The total desync between the MT5 client and the server position is definitely the worst part.
Given you're running 200+ clients across those VPS instances, the escalating latency until you hit timeouts sounds very much like the broker might be intentionally throttling, or they simply have a capacity issue hitting a wall.
Talking to the broker with those specific 158s log entries (as Alain suggested) is the right call.
SashaD03 #:
That is a serious issue, especially the 158-second execution time! The total desync between the MT5 client and the server position is definitely the worst part.
Given you're running 200+ clients across those VPS instances, the escalating latency until you hit timeouts sounds very much like the broker might be intentionally throttling, or they simply have a capacity issue hitting a wall.
Talking to the broker with those specific 158s log entries (as Alain suggested) is the right call.
Alain was right, broker fixed it now it works seamlessly
12
New comment