New Hope PRO
LMAX and Fast Forex Feeds: TPS vs Latency | New Hope PRO
A large TPS counter can look impressive. It does not tell you whether a useful price change reached your strategy first. Here is how to separate quote activity from latency, and what our LMAX demonstration actually shows.
What does TPS mean in a forex feed?
Here, TPS means quote ticks per second: counted updates divided by elapsed seconds. First define an update. A received message, a change of best Bid or Ask, a trade print and a software callback are different observations. A comparison is meaningless if one supplier counts messages across many symbols and another counts price changes for one pair.
In other technical discussions TPS means transactions per second. Matching-engine throughput, order-processing capacity and the delivery of quotes to your terminal are separate measurements. LMAX itself presents processing capacity and matching latency separately in its technology description. Neither figure alone describes your complete data path.
1,000 TPS does not mean 1 ms latency
At 1,000 updates per second, the average interval is 1 / 1,000 = 0.001 seconds, or 1 ms. That describes a cadence, not an update’s age when it arrives. Arrivals may be uneven: a buffered burst can deliver many updates that are already old. A quieter stream may deliver a particular useful change earlier.
Latency needs two defined endpoints and a time difference. “1,000 ticks faster” is therefore not a usable delay measurement. “This matched event arrived 3.2 ms earlier on our capture host” could be meaningful, if supported by timestamps and a clear comparison method. The IETF’s one-way delay specification explains why measurement boundaries and clock uncertainty matter.
A calculation: more updates, later arrival
Synthetic teaching example. These are not measurements of New Hope or LMAX. Observe two streams for ten seconds. Assume one genuinely shared event and synchronized clocks:
Measure | Feed A | Feed B |
|---|---|---|
Updates in 10 seconds | 20,000 | 4,000 |
Calculated update rate | 2,000 TPS | 400 TPS |
Selected event: source-to-receiver delay | 40 ms | 4 ms |
A has five times B’s update rate. Yet B receives the selected event 40 − 4 = 36 ms earlier. These facts can coexist. The example proves a distinction between measurements, not the performance of either product.
The two panels use different units. Neither bar measures trading profit.
If you only timestamp arrivals on one shared receiver, the difference still describes a relative lead for a matched event. Without a reliable source timestamp, it does not establish each stream’s absolute source-to-client latency.
Why “LMAX speed” needs a connection and a method
LMAX is a useful comparison point, but a name is not a universal speed guarantee. Its market-data access page lists different connections and configurable update frequencies, including 1 ms, 10 ms and 100 ms. Those settings are not a promise of that many unique price changes, nor a measurement from source to every client.
Specify the venue, API or platform, instrument, data depth, delivery settings and measurement host. Across different forex venues there may be no identical source event to match. A carefully defined price-response lead/lag study can still be useful, but it should not be presented as a pure network-latency test. One feed can lead during some changes and follow during others.
Watch the New Hope PRO demonstration
In 2024, New Hope was already demonstrating its multifeed approach. At 6:58, the presenter identifies a live LMAX account; the visible history includes winning and losing trades. The discussion then covers collecting ticks at 10:34, changing feed leadership at 11:15, TPS at 12:29 and software optimization at 15:05.
Terminal and trading history from our LMAX account demonstration. Open the screenshot at full size. The screenshot shows part of the trading history, not an independently verified complete account statement or a latency benchmark.
Three ways a TPS counter can mislead
Different delivery rules. Conflation combines updates, changing message counts. LSEG distinguishes full-tick and conflated services in its real-time data documentation.
Different counting boundaries. MetaQuotes explains that OnTick notifications can cover several ticks. Counting callbacks is not the same as counting tick-history records.
Different content. A stream may include depth changes, repeated prices or more instruments. Adding all messages does not produce a count of unique, useful trading signals.
TPS is still useful for understanding activity, bursts and processing load. Use it alongside timeliness and completeness. For the difference between recorded ticks and candles, see our tick-data practical guide.
How to compare feeds before making a claim
New Hope PRO includes a dedicated tool for testing quote feeds. You can connect your own feed, record ticks and compare it with a broker's quotes or another feed. This lets you check performance using your own data and look for price differences that may present arbitrage opportunities.
Example: your feed versus a broker
Connect your own quote source to New Hope PRO. For this example, call it “Your Feed 1”.
Connect the broker you want to compare it with. Call this connection “Broker 1”.
Select the same instrument on both connections. Compare equivalent prices, such as Bid with Bid and Ask with Ask.
Enable tick recording for “Your Feed 1” and “Broker 1”. Collect data over your chosen period.
Compare the recordings. Check which source shows a price change first, when the quotes diverge and how long the difference lasts. This helps you assess your feed's advantage over that broker and identify potential arbitrage entries for further analysis.
The same approach lets you compare two of your own feeds or use quotes from two different brokers. One broker can serve as the faster source and the other as the slower one. Assign those roles from the comparison results: the relationship can change with the instrument or observation period.
Keep the comparison consistent
Fix the setup. Record the instrument, quote level, connection, delivery configuration and test period.
Define the event. Separate received messages, changed Bid/Ask and matched useful events. Apply one deduplication rule.
Use a common clock. Capture both streams at the same boundary. Keep source time and receive time separate; quantify synchronization uncertainty.
Keep the inconvenient observations. Log disconnections, stale intervals and stressed sessions as well as normal trading.
Report a distribution. Include sample size, duration, median, a high percentile and how often each stream leads. One attractive screenshot is insufficient.
Then measure signal calculation, order submission, acknowledgement and execution separately. Spread, commissions, slippage, rejections and the broker’s account rules affect the trading result. A quote advantage is not automatically an executable opportunity. Our New Hope PRO setup guide helps structure those checks.
What the New Hope PRO multifeed includes
At a public, high level, multifeed means several fast quote streams combined within one product workflow. It does not mean every source leads every other venue at every instant. The specific providers and detailed composition of the New Hope multifeed are proprietary and are not disclosed publicly.
Conceptual overview, not a disclosure of the source list or selection algorithm.
The full New Hope PRO package is a one-time purchase. Continue using the purchased package without a recurring software subscription. Software updates and integrated fast quote feeds are included at no extra charge. They are also included during the evaluation’s valid one-month period; evaluation access remains time-limited. VPS hosting, broker charges and trading capital are separate.
Read the New Hope Fast Feed overview and New Hope PRO product details. Contact the team with your platform, intended instrument and VPS location. Those details help discuss a realistic setup; a TPS counter alone cannot.
For a practical test, follow our step-by-step guide to comparing quote-feed speed in New Hope PRO. It covers connecting two streams, recording ticks and comparing the arrival of matched price changes.
More articles
