How HFT Arbitrage Platform Performs

TL;DR — How it performs

Performance of the HFT Arbitrage Platform is determined by four variables: broker execution quality, VPS latency to the broker, reference-feed quality, and strategy selection. With a correctly chosen arbitrage-friendly broker, a sub-1 ms VPS in LD4/NY4/TY3/FR5, and a premium feed, the platform executes signals in 0.3–2 ms and typically achieves fill rates of 85–98% on one-leg latency, 90–99% on hedge arbitrage, and 70–90% on 2-legs latency. We do नहीं publish fixed percentage returns — anyone who does is misleading you. Real results depend on your specific broker/VPS/feed/capital combination.

When prospective users ask us “how does the platform perform?” they usually expect a percentage number — “15% per month,” “doubles your account in 90 days.” We will not give you that number, because it does not exist. Arbitrage performance is not a property of the software alone; it is a property of the full stack: the software, the broker, the VPS, the data feed, the market conditions, and the capital deployed. Change any one of those variables and the result changes.

This page explains what performance looks like in practice: the metrics we actually measure, the factors that move those metrics up or down, realistic ranges by strategy type, and how you can evaluate performance on your own setup before committing capital. If you want a one-sentence summary: the platform performs well when the execution chain around it is set up correctly, and poorly when it is not. The rest of this page explains how to tell the difference.

Live verified performance — third-party reference accounts

We maintain public reference accounts on FxBlue — an independent third-party trade-verification service — so prospective customers can see real execution data from the platform running on a correctly configured setup. FxBlue reads the account’s trade log directly from the broker and publishes it verbatim; we cannot edit or filter what it shows. Below are the current live reference accounts for our two most-used strategies.

One-leg latency arbitrage — reference account

Live statistics from our 1-leg latency reference account, updated in real time by FxBlue. Click the banner to open the full account page with trade-by-trade history, drawdown chart, and performance graph.

2-legs latency arbitrage — reference accounts

We run two parallel reference accounts for 2-legs latency so users can compare execution across different broker setups. Both are publicly verified by FxBlue.

खाता 1

खाता 2

How to read the FxBlue data

Each banner shows current equity curve, total P&L, and win rate at a glance. Clicking through opens the full FxBlue account page, which includes:

Complete trade-by-trade log with entry, exit, lot size, and P&L per trade
Monthly and daily performance breakdown
Drawdown history and maximum drawdown percentage
Win rate, average win, average loss, profit factor
Trade frequency, average holding time, and instruments traded
Account growth chart since the first published trade

Important context: these are reference accounts running on a specific broker, VPS location, feed provider, and capital base chosen by us. Your own results will differ based on your broker, VPS, feed, strategy mix, and capital. The reference accounts demonstrate that the platform executes correctly on a real-life broker — they are not a promise that your account will replicate these numbers. See the “Four factors that determine your results” section below for what changes the outcome.

Hedge arbitrage and triangular arbitrage reference accounts are provided to paid-license customers on request, because those strategies typically require broker-specific configuration that we prefer to discuss 1-on-1 rather than expose on a public verification page.

What does performance mean in arbitrage trading

In directional trading, performance usually means return on capital. In arbitrage, that metric is downstream of several more fundamental numbers. Because arbitrage trades are short-duration, small-per-trade, and high-frequency, the real performance question is not “how much did the account grow this month” but “how reliably did the software detect, send, and fill signals against the actual market.”

The metrics that matter are these:

Metric	What it measures	Healthy range
Signal detection latency	Time from feed tick arriving to arbitrage opportunity identified	< 0.5 ms
Order send latency	Time from platform decision to order leaving VPS	< 1 ms
Round-trip execution	Order send → broker ack → fill confirmation	5–30 ms (depends on broker)
Fill rate	% of sent orders that fill at the expected price	85–99% (strategy-dependent)
फिसलन	Avg price difference between signal and fill	0–0.3 pips (quality broker)
Requote rate	% of orders rejected / requoted by broker	< 2%
Profit factor	Gross profit ÷ gross loss across closed trades	> 1.5 (tunable via filters)
Broker survival time	Days/weeks before the broker flags or restricts the account	Weeks to indefinite (with anti-detection filters)

If signal detection, order send, and round-trip execution are all inside healthy ranges, the strategy will work. If any one of them is outside its range, return on capital collapses regardless of how good the strategy logic is. That is why we focus on these numbers rather than on a “monthly return” figure.

The four factors that determine your results

1. Broker execution quality

The single biggest factor. A broker with STP/ECN execution, low markup, deep liquidity, and a policy that permits arbitrage will give you fill rates above 90% and near-zero slippage. A market-maker broker with a dealing desk, internal matching, and hostile execution settings will give you fill rates below 50%, frequent requotes, and “plugins” that specifically add artificial delay to arbitrage orders. Paid-license customers receive our current list of arbitrage-permitted brokers with execution notes; this list is not public because it changes as broker policies change.

2. VPS location and network latency

Arbitrage requires the VPS to be physically collocated in the same data center as the broker’s trading server. The four datacenters that matter for forex/CFD are LD4 (Equinix London — most ECN brokers), NY4 (Equinix New York — US equities and many brokers), TY3 (Equinix Tokyo — Asian sessions), and FR5 (Equinix Frankfurt — European liquidity). Running on a consumer VPS or on a general-purpose cloud region adds 30–200 ms round-trip latency, which kills latency arbitrage and degrades all other strategies.

3. Reference feed quality

Latency arbitrage, by definition, compares a fast reference feed against a slower broker feed. If the reference feed is itself slow or derived, there is no edge to exploit. The platform supports Rithmic, CQG, Integral, LMAX, and cTrader Raw as reference sources. These are institutional-grade feeds and cost $50–$300/month; consumer aggregator feeds will not produce working latency signals. Hedge arbitrage is less feed-sensitive but still benefits from clean data on both legs.

4. Strategy selection for the account type

Not every strategy works on every account. One-leg latency requires a slow broker with permissive execution, a narrow category. Hedge arbitrage works across two brokers and is the most flexible strategy, but requires capital on both sides. 2-legs latency variant 3 is the most prop-firm-compatible variant. Triangular arbitrage requires tight pricing on all three legs of the triangle, which most retail brokers do not offer. Matching the strategy to the account is why Custom Configuration and the All Arbitrage Bundle exist.

Realistic performance ranges by strategy

The table below shows observed ranges across our customer base on correctly configured setups. These are not guarantees, and they assume an arbitrage-friendly broker + LD4/NY4 VPS + quality feed. Poor setup significantly reduces every number in this table.

रणनीति	Fill rate	Avg slippage	Trade frequency	सर्वोत्तम
एक-पैर विलंबता	85–98%	0–0.2 pips	20–200 / day	Retail accounts with arbitrage-permitted broker
2-legs latency v1	70–88%	0–0.3 pips	10–80 / day	Dual-broker retail setup
2-legs latency v2	75–90%	0–0.3 pips	15–100 / day	Reduced detection risk
2-legs latency v3	78–92%	0–0.3 pips	15–120 / day	Prop firm accounts
हेज आर्बिट्रेज	90–99%	0–0.1 pips	5–40 / day per pair	Most flexible, prop firm compatible
त्रिकोणीय मध्यस्थता	60–85%	0–0.5 pips	3–25 / day	Tight-spread ECN accounts
3-लेग विलंबता	65–82%	0–0.4 pips	5–40 / day	Advanced multi-broker setups

Ranges are observed across our active customer base on a correctly configured infrastructure. Individual results vary. These numbers describe execution behavior, not profit, and do not constitute a performance guarantee.

Why we don’t publish “% per month” returns

If you have compared arbitrage software vendors, you have probably seen screenshots of MyFxBook accounts showing “300% return in 6 months” or videos claiming “the system makes $X per day.” We will not provide those claims, and here is why:

First, return on capital is not a property of the software. A $10,000 account on broker A with VPS in LD4 will produce completely different numbers than the same $10,000 on broker B with a VPS in a consumer datacenter. Publishing one result as “the platform’s performance” is dishonest.

Second, the arbitrage-friendly broker list changes. A broker that worked in Q1 may change execution policy in Q2 and suddenly produce 40% worse fills. Any historical screenshot becomes unrepresentative within months.

Third, publishing high returns invites broker retaliation and attracts the wrong customers. Brokers monitor vendors that brag about their software; publishing specific numbers helps them identify and restrict accounts running the platform. It also attracts users who expect those numbers and blame the software when their own setup produces different results.

Fourth, vendors who publish fixed-return claims are almost always lying. The claims are typically either cherry-picked single weeks, demo-account backtests labeled as live, or outright fabricated. If a vendor guarantees X% per month, that is a signal to leave, not to buy.

What we do publish is the execution-quality ranges in the table above, plus three live FxBlue-verified reference accounts (shown near the top of this page) covering one-leg and 2-legs latency on real broker setups. FxBlue is an independent third-party trade-verification service — the data comes directly from the broker, and we cannot edit or filter it. Hedge and triangular reference accounts are available to paid customers on request. That is the honest version of “performance data”: verified third-party execution on our own accounts, not cherry-picked screenshots or promises about what your account will do.

How to evaluate performance before committing capital

The correct way to answer “will this work for me” is to test it on your intended setup. The platform supports this at four levels of commitment, from zero cost to full deployment.

Level 1 — Backtest engine (free with any paid edition)

The built-in 1-ms backtest engine replays historical tick data through your chosen strategy and broker-latency profile. You get a synthetic run showing fill rates, slippage distribution, and signal counts over whatever historical window you select. This tells you whether the logic works in principle.

Level 2 — Shareware demo ($19)

The Free edition runs the real engine on your real VPS against your real broker, restricted to EURUSD and 0.01-lot trades. This tells you whether your specific broker, VPS, and feed combination produces healthy execution metrics. Cost is $19, and the data is real.

Level 3 — Small live account

Once the shareware run shows acceptable fill rates and slippage, deploy the paid edition on a small live account ($500–$2,000) for two to four weeks before scaling. This is the only stage where you learn how the broker responds to arbitrage activity over time — which is information no backtest can give you.

Level 4 — Scaled live deployment

If the small live account holds up for 2–4 weeks with stable metrics and no broker restrictions, scale capital. This is also when prop firm accounts become worth testing, since prop firm evaluations typically last 30 days, and the small live phase validates that your setup can pass.

How does broker type changes performance

Broker type	Expected fill rate	Expected slippage	Arbitrage tolerance
True ECN (bank-liquidity, no dealing desk)	95–99%	0–0.1 pips	Permitted by policy
STP (straight-through processing)	85–95%	0–0.3 pips	Usually permitted
Hybrid STP / market maker	60–80%	0.3–1 pip	Tolerated until detected
Pure market maker with dealing desk	30–60%	1–5 pips	Prohibited / restricted
Market maker with anti-arbitrage plugin	< 20%	3–10 pips	Actively blocked

The same platform, the same strategy, and the same VPS will look like a great product on row 1 and a broken product on row 5. This is why the broker-selection list we give paid customers is the single most valuable thing they receive — more valuable than the software itself for first-time users.

How VPS location changes performance

VPS location	Typical round-trip to LD4 broker	Effect on latency arbitrage
Equinix LD4 colocation	0.1–0.5 ms	Optimal — strategy viable
London datacenter (non-colo)	2–8 ms	Marginal — hedge OK, latency degraded
Continental EU general VPS	15–30 ms	Latency arbitrage non-viable
Consumer cloud (AWS, GCP general)	25–80 ms	All strategies degraded
Home internet / US consumer VPS	80–200+ ms	Not usable for arbitrage

If the broker is located in LD4, the VPS must be in LD4. If the broker is in NY4, the VPS must be in NY4. A sub-1 ms VPS-to-broker link is not a nice-to-have for latency arbitrage — it is the difference between the strategy working and not working.

Performance on prop firm accounts

Prop firm accounts change the performance picture in two ways. First, one-leg latency arbitrage is prohibited by essentially every major prop firm — FTMO, FundedNext, The5ers, MyForexFunds — so that strategy is off the table. Second, hedge arbitrage and 2-legs latency variant 3 are typically permitted and work within the daily-drawdown and max-drawdown limits that prop firms impose.

On a correctly configured prop firm setup using hedge + 2-legs v3, fill rates are similar to retail — 88–97% on hedge, 78–90% on 2-legs v3. The constraint is not execution but risk management: prop firm rules usually cap daily loss at 4–5% and total drawdown at 8–10%, so position sizing needs to respect those limits. The platform’s risk dashboard enforces this automatically when you configure the per-day and per-trade drawdown ceilings at setup.

For a complete prop firm compatibility matrix and strategy recommendations by firm, see प्रॉप फर्म आर्बिट्रेज.

Factors that affect performance and are not in our control

We want to be explicit about what we do not control, so expectations stay aligned with reality:

Broker policy changes. A broker can change execution model, add arbitrage-detection plugins, widen spreads, or ban arbitrage at any time. We track the arbitrage-friendly list and update it, but we cannot guarantee any specific broker will remain arbitrage-permitted indefinitely.
Market volatility regime. Arbitrage opportunities are more frequent in volatile markets and less frequent in compressed-range conditions. A calm market produces fewer signals; this is not a software problem.
News events. During high-impact news (NFP, FOMC, ECB) most brokers widen spreads 5–20× and disable or restrict arbitrage-like activity. Fills during news are unreliable by design, and the platform’s session filter defaults to pausing during scheduled high-impact events.
Your capital and sizing decisions. We do not determine lot size, leverage, or risk per trade — you do. Those decisions have as much impact on results as the software.
VPS provider reliability. Network issues, VPS reboots, or ISP problems at the datacenter are outside our control. Use redundant VPS setups for serious capital.

Performance FAQ

What monthly return should I expect?

We do not publish expected monthly returns, because they depend entirely on your broker, VPS, feed, strategy, capital, and market conditions. Any vendor giving you a specific percentage — “20% per month,” “doubles your account” — is either lying or cherry-picking one favourable week. What we do publish is execution-quality ranges: fill rate, slippage, and trade frequency by strategy. Those numbers, on a correctly configured setup, are the only honest starting point. Return on capital is downstream of execution quality plus position sizing, both of which are specific to your account.

Can I see live trading statements?

Yes. Three live reference accounts are publicly verified by FxBlue — an independent third-party trade-tracking service that reads trade logs directly from the broker. The banners near the top of this page link to the full accounts: one for 1-leg latency (hftarbitrage-1leg) and two for 2-legs latency (hftarbitrage-2-legs1 और hftarbitrage-2-legs2). Each account page shows complete trade-by-trade history, drawdown, win rate, and equity curve. Hedge and triangular arbitrage reference accounts are provided to paid-license customers on request, since those strategies usually require broker-specific configuration better discussed 1-on-1.

Why are fill rates in the shareware lower than in the full version?

The Free shareware has an artificial 10 ms feed delay, is restricted to EURUSD and 0.01 lots, and runs without the anti-detection filters. Those restrictions reduce fill rates by 15–25% versus the paid editions. The purpose of the shareware is to verify that your broker + VPS + feed combination produces healthy execution on a known-restricted configuration — if metrics are acceptable in shareware, they will be meaningfully better in the paid edition.

My fill rate is low — is the software broken?

Almost always the problem is in the execution chain, not in the software. In order of frequency: (1) broker is a market maker with dealing-desk execution, (2) VPS is not colocated with the broker, (3) reference feed is a consumer aggregator rather than Rithmic/CQG/Integral/LMAX/cTrader Raw, (4) broker has added an anti-arbitrage plugin without notifying you, (5) strategy choice is wrong for the broker type. Paid-license support will walk through each of these with you. In four years of support tickets, we have never found a low-fill-rate problem that was caused by the software itself.

How long before a broker restricts an arbitrage account?

It depends on the broker and on your account behaviour. On brokers whose terms explicitly permit arbitrage, accounts run indefinitely. On brokers that tolerate arbitrage until it becomes visible, the platform’s anti-detection filters — randomized sizing, holding-time enforcement, session diversification — extend survival from days to months. On brokers that prohibit arbitrage, no filter is sufficient; you will be flagged and restricted, typically within 2–8 weeks. The broker-selection list we provide to paid customers focuses only on brokers in the first category.

What is the minimum capital to see meaningful performance?

For evaluation: $500–$2,000 on a small live account is enough to validate fill rates and broker behaviour over 2–4 weeks. For operational trading: $5,000–$25,000 is the common retail range where position sizing can respect broker minimums while still producing economically meaningful results relative to the broker/VPS/feed monthly cost. Below $500, per-trade minimums and spreads dominate the result and the numbers stop reflecting the strategy. Above $100,000 on a single retail broker, detection risk scales and multi-broker setups become preferable.

Do you offer a performance guarantee or refund if I lose money?

No — and any software vendor who offers one is either lying or running a Ponzi structure. Arbitrage trading software is a tool; trading outcomes depend on broker, market, capital, and operator decisions we do not control. We offer the $19 shareware so you can validate the software on your own broker and VPS before committing to the full license fee. We also provide a curated arbitrage-friendly broker list and set up support to paid customers, which is the closest thing to a risk-reduction mechanism that can honestly exist in this category.

Summary

The performance of an HFT Arbitrage Platform is a function of the execution chain, not a fixed property of the software. Four variables — broker execution quality, VPS colocation, reference feed quality, and strategy selection — determine whether the platform produces healthy fill rates and low slippage or degraded fill rates and high slippage. On a correctly configured setup, observed metrics fall within the ranges in the table above. On a poorly configured setup, they do not, and no amount of software tuning compensates for the wrong broker or the wrong VPS. The right way to evaluate performance is to test on your own setup at increasing levels of capital: backtest, shareware, small live, scaled live. The wrong way is to rely on a vendor’s percentage claim.

Want to test performance on your own broker and VPS?

Start with $19 Shareware
संस्करणों की तुलना करें