I'm thinking about running a small EURGBP mean-reversion EA, and wondering whether a separate VPS just for it makes sense or whether that's overkill.
The EA:
Trades during the London session only.
Average target 4 pips, stop 6.
Around 3 trades a day.
Costs on my side: a VPS near the broker's server costs about €25 a month. With 0.1 lots and roughly 60 trades a month, that's about €0.42 per trade, or over 0.35 pips of extra cost per trade. That's a lot for a 4-pip target.
Alternatives I'm considering:
Running it on my existing VPS together with other EAs.
Running it from home, since the strategy isn't latency-sensitive (holding time 20–60 minutes).
Waiting until the account is bigger so that the VPS cost is a smaller share.
My feeling is that for a strategy that holds 20+ minutes, latency hardly matters, while stability does. A home PC with a UPS and good internet might be enough.
Anyone running EURGBP EAs? Did a dedicated VPS make a measurable difference for you?
There are two schools of thought on VPS placement, and they point to different data centers.
Approach one is to put the VPS in the same region as the broker's trading server. The logic is clean. The order leaves the terminal, reaches the broker within a millisecond or two, and the first leg of the trip is as short as it can get. Most VPS marketing is built around this idea, and for a broker whose trading server does most of the work in-house, it holds up well.
Approach two is to put the VPS near wherever the broker actually sends the flow, meaning the liquidity provider or the aggregation point that sets prices and fills orders. The argument is that the broker's front door is only one hop, and the slow part may be the hop after it. If the trading server sits in one city but fills come from a hub in another, a VPS next to the front door shaves a millisecond off a short leg and still waits on the long one.
The catch is that retail brokers rarely publish where that second hop lives. Some of it can be guessed from traceroutes and from which hosting provider the broker itself recommends, but that is inference, not fact. And a broker that internalises a lot of flow may not have a meaningful second hop for many orders at all.
So the practical answer is to test both. Rent a VPS by the hour in each candidate location, run the same terminal, and measure order-to-confirmation times and slippage on small size over several sessions, including at least one news event. Judge by the spread of results, not the average. A location that is fast on most orders but has a fat tail of slow ones is worse for scalping than a slightly slower but steady one.
Quote delivery also deserves a look. Late prices hurt as much as late orders, and the price path does not always follow the order path.
Which side has worked better for you, and did you confirm it with measured fills or just trust the ping?
