<\/p>\n
<\/p>\n
Client Story \u00b7 Multi-Account Arbitrage \u00b7 12-Month Review<\/span><\/p>\n An honest writeup of what twelve months of multi-account hybrid masking actually feels like, what brokers actually check (and what they don’t), and the operational layer that arbitrage software vendors don’t talk about.<\/p>\n Guest post by a BJF Trading Group client and multi-account arbitrage operator. Identity withheld at the author’s request. Names of brokers and exchanges anonymized throughout to keep current relationships intact.<\/p>\n<\/div>\n <\/p>\n I have been running BJF’s Phantom Drift with a Hybrid Masking strategy across four broker accounts for about twelve months. Boris asked if I would share what that actually looks like from the operator’s side, and I agreed, on the condition that my identity stays out of it. So here we are.<\/p>\n I am going to skip the part where arbitrage software vendors say their stuff is amazing. You can read the BJF site for that. What I want to write up is the layer underneath that nobody advertises: the operational reality of running multiple accounts, what brokers actually check (it is less than you think), and the parts that genuinely bored me into stability.<\/p>\n If you are considering jumping into multi-account arbitrage, this is the write-up I wish I had read before I started.<\/p>\n <\/p>\n Six years of trading before this. Started with the usual: trend-following systems, mean-reversion robots, news strategies. Made some, lost some, the net was probably break-even after costs. The honest assessment from year four was that most retail strategies do not survive contact with the broker’s execution layer.<\/p>\n Arbitrage was on my radar, but I assumed it was either fake (a marketing pitch attached to whatever software was being sold that month) or only viable for institutional players with co-located servers. Both assumptions were wrong, but it took me about a year of reading and watching to realize that.<\/p>\n The turning point was the BJF research material. Specifically, the “Using Arbitrage Strategies in Trading on Financial Markets”. The numbers in those papers matched the academic literature on retail forex microstructure (Hasbrouck and Saar on low-latency trading, Aquilina and Budish on the HFT arms race). Once I cross-checked the BJF methodology against the academic side, the picture stopped looking like marketing and started looking like a documented market structure that I had been ignoring.<\/p>\n “Once you cross-check the vendor’s methodology against the academic literature and the numbers reconcile, the question stops being ‘is this real’ and starts being ‘why am I not on the inside of this’.”<\/p><\/blockquote>\n I bought SharpTrader Pro about a month later. Started with a single account on a single broker. Ran it for three months on a minimum lot, just learning the configuration parameters and the workflow. By month four, I had added a second account on a second broker. By month seven, four accounts on four brokers, with Phantom Drift and hybrid masking layered in.<\/p>\n <\/p>\n This is where most articles about arbitrage end. They tell you about the strategy, maybe the parameters, maybe a backtest. Then they leave you to discover for yourself what running four accounts actually looks like operationally.<\/p>\n Here is what it actually looks like.<\/p>\n I want to be clear about the basics first: all four accounts are opened honestly in my own name (three of them) and in a family member’s name for one of them. No fake identities, no synthetic data. I opened accounts at four different brokers in different jurisdictions, exactly as their TOS allows. Multi-broker operation is not the same as identity manipulation, and the difference matters legally and operationally.<\/p>\n Each broker account needs:<\/p>\n I am going to list these specifically because every single one of them is the kind of thing that seems obvious in writing but is genuinely easy to forget when you are setting up account three at midnight.<\/p>\n Operational mistakes worth\u00a0<\/span>avoiding. Same<\/strong><\/span>\u00a0email inbox for all four brokers.<\/strong> Easier in the short term, terrible in the medium term. Notification noise piles up, password reset flows get confused, and the broker’s customer system links you across accounts even though they are at different brokers. Use a separate inbox per broker.<\/p>\n Same VPS instance for multiple accounts.<\/strong> Two broker accounts that log in from the same VPS IP are flagged immediately. Use a separate VPS per broker. The cost difference for four VPSes versus one is trivial relative to the operational protection.<\/p>\n Same trading session timestamps.<\/strong> If all four accounts log in within the same five-minute window every morning, you have created a temporal fingerprint that anti-arbitrage plugins flag. Stagger the login schedule deliberately.<\/p>\n Skipping the account-aging step.<\/strong> A brand-new account that immediately runs automated activity is a flag. Open the account, deposit modestly, run a handful of manual trades for two to three weeks before deploying any automation.<\/p>\n<\/div>\n I did three of these four in the first month. Specifically, the email separation, the VPS provider, and the timestamps. None of them caused immediate consequences, but each was a point of friction that would have surfaced eventually. I cleaned them up before the second month.<\/p>\n <\/p>\n \u270f Editor’s note from BJF Trading Group: The<\/span>\u00a0views expressed in this guest post are the author’s own and do not in every case reflect BJF Trading Group’s editorial position. We want to flag two specific points where our view differs from the author’s operational approach.<\/p>\n On separate payment methods and IP addresses across different brokers.<\/strong> The author treats per-broker funding and IP separation as standard hygiene. In our view, this is overstated for the legitimate multi-broker case. Data exchange between competing retail brokers is restricted by data protection law (GDPR in the EU, similar frameworks elsewhere) and is realistically possible only when two brokers share a common owner. Outside that scenario, broker A simply has no lawful access to broker B’s customer records, and using the same payment method or the same IP across two unrelated brokers does not by itself create a cross-broker identity link. Per-broker IP separation is sensible for technical reasons specific to anti-arbitrage detection within a single broker (multiple accounts hitting the same broker from one IP is a clear flag), but the cross-broker concern that the author implies is largely theoretical under current regulatory frameworks.<\/p>\n On starting with four broker accounts.<\/strong> The author scaled from 1 to 4 accounts over 7 months. We think starting with two accounts (one per broker, two brokers in different jurisdictions) is the healthier learning curve for most readers. Two accounts is enough to learn the operational workflow without overwhelming the onboarding capacity in the first quarter. Four accounts are the right footprint for a year-two operator who has stabilized on two. It is not the right footprint to bring up from scratch in a single setup phase.<\/p>\n Read the rest of the writeup with these caveats in mind. The author’s 12 months of operational experience is genuine, but a single trader’s path is not a recipe.<\/p>\n<\/div>\n <\/p>\n This part is comedy in retrospect.<\/p>\n Four browser windows. Four VPS sessions drop intermittently due to uneven residential proxy quality. Four phone notifications about deposit confirmations from four different banks. Four different broker dashboards with four different terminologies for the same concept. Four different sets of margin requirements that I keep forgetting.<\/p>\n I missed a margin call on account two in week one because I was watching the wrong dashboard. Closed out of two open positions at a loss. Lost about as much as I had paid for the software the previous quarter. Lesson learned.<\/p>\n By week three, I had built a spreadsheet that tracked: account name, broker, login schedule, margin level, open positions, last execution timestamp, and latest VPS health check. Every login starts with the spreadsheet. Every logout ends with the spreadsheet. It is the most boring workflow I have ever set up, which is precisely why it works.<\/p>\n <\/p>\n This is the part nobody markets, but it is the most important thing I have to say.<\/p>\n After about three weeks of the workflow I described above, everything stabilized. The accounts run themselves. The software does the heavy lifting. My job became reduced to:<\/p>\n The total operational time is about an hour per day, with most of it spent on the morning check. Some days it is twenty minutes. Sundays are zero.<\/p>\n This is not exciting. It is not the trading lifestyle that gets posted on Instagram. It is closer to running a small business than it is to “trading”. And that is exactly why it works.<\/p>\n <\/p>\n This is the part I find most interesting after twelve months.<\/p>\n The marketing pitch for hybrid masking is “make your activity look normal so the broker does not flag you as an arbitrageur”. The reality is more nuanced. Most retail brokers do not have a human compliance team that sits there watching your account. They have plugins. The plugins are not particularly clever. They look for specific patterns, and the patterns are not what you would intuitively think.<\/p>\n From watching what does and does not get flagged across my four accounts, here is what I have learned about what brokers’ anti-arbitrage layers actually catch.<\/p>\n This is the more interesting list. These are gaps in the broker-side detection layer that hybrid masking exploits by design.<\/p>\n The hybrid masking strategy is essentially a layer that converts a “pure arbitrage pattern” into a “messy human-looking pattern” without sacrificing too much of the captured edge. The math on how much edge you sacrifice is in the BJF research paper on backtest fidelity, which is worth reading if you want the theoretical grounding.<\/p>\n <\/p>\n I am not going to pretend that nothing has ever gone wrong. In month four, one of my four brokers froze account three for a “compliance review”. The trigger, as far as I can reconstruct, was a deposit pattern that matched what their AML system considered high-risk. Nothing to do with arbitrage. The account was in good standing on the trading side.<\/p>\n The review took three business days. They asked for a refreshed copy of my ID, a utility bill from the last 90 days, a brief explanation of my source of funds, and a copy of my recent trading statements from any other accounts I held with regulated brokers.<\/p>\n I provided everything from a pre-prepared folder I had set up at the start. The trading statements from the other three accounts showed mixed strategies (which is what hybrid masking produces). The compliance officer accepted the package and released the account. No restrictions added. No follow-up.<\/p>\n Have your story ready before they ask. Have a refreshed ID, recent utilities, and a one-page document explaining your trading background, prepared and stored where you can send it within 30 minutes of a request. The slower you respond to a compliance request, the more attention you attract.<\/p>\n I keep a folder labelled “compliance refresh package” on each VPS. It contains the documents I need to send if a broker ever asks. I have used it once. It paid for itself the first time.<\/p>\n<\/div>\n <\/p>\n Every multi-account problem you will eventually have to solve is easier to solve when you only have one account to break. Run a single account with a single broker for at least 2 months before adding a second. The instinct is to scale fast. Resist it.<\/p>\n Co-location at LD4 or NY4 sounds technical and serious. It is. But a cheap LD4 VPS that drops connection three times a day is worse than a stable VPS in a generic data center. Stability beats milliseconds-of-latency advantage for retail-scale arbitrage. Pay for the good provider.<\/p>\n A brand-new account that immediately runs sub-100 ms execution patterns is a compliance flag the moment you start trading. Open the account, deposit modestly, run a handful of manual trades for a week or two before deploying any automation. Let the account appear to be a normal customer long enough to register in the broker’s CRM as “real”.<\/p>\n One bank card across four broker accounts is the single most reliable identity-linking vector that compliance systems use. If you can split funding across two banks (or two crypto exchanges for crypto venues), do it. If you cannot, at least space out the funding events.<\/p>\n I cannot stress this enough. I started a spreadsheet in week three. I now have twelve months of data on broker-level slippage, execution latency, opportunity density, and per-account behaviour. When I want to know whether broker A has gotten more aggressive on slippage recently, I can answer that in 30 seconds. When I am evaluating whether to add a fifth account on a new broker, I have a baseline to compare against. The spreadsheet is more valuable than the software, almost.<\/p>\n <\/p>\n SharpTrader Pro is the software I run. Phantom Drift and hybrid masking are configurable modes within it. BJF’s free public arbitrage stats dashboard convinced me that arbitrage was real before I committed.<\/p>\n SharpTrader Pro \u2192<\/a> <\/p>\n If I had to summarise twelve months of multi-account arbitrage in three sentences, it would be:<\/p>\n The software does the heavy lifting.<\/strong> Phantom Drift, hybrid masking, the underlying execution engine: these are solved problems. Buying the software solves the trading half of the equation.<\/p>\n The operator’s job is operational hygiene.<\/strong> Account separation, IP separation, funding separation, identity separation, and schedule separation. These are not glamorous, but they are the entire game on the operator’s side.<\/p>\n It gets boring fast. That is the goal.<\/strong> If your multi-account arbitrage setup is exciting day-to-day, something is wrong. Boring is what stability looks like. Boring is what eight or nine months of compounding looks like. Boring is what you want.<\/p>\n I am going to keep running this setup. I have no plans to scale to a fifth account in the next six months because I am happy with the operational load at four. If anything changes materially, I might write a follow-up. In the meantime, that is the year-one report.<\/p>\n <\/p>\n New methodology papers, broker execution quality updates, and operator notes like this one, delivered when we publish them. No promotional spam.<\/p>\nYear one running Phantom Drift across four broker accounts<\/em><\/h2>\n
How I got here<\/h2>\n
The setup nobody tells you about<\/h2>\n
The four-account stack<\/h3>\n
\n
Things I almost screwed up in the first month<\/h3>\n
Weeks one and two are chaos<\/h2>\n
By week four, it is just routine<\/h2>\n
\n
What brokers actually check (and what they don’t)<\/h2>\n
What plugins consistently catch<\/h3>\n
\n
What plugins consistently miss<\/h3>\n
\n
When one broker flagged me anyway<\/h2>\n
The lesson<\/h3>\n
Five things I wish I had known on day one<\/h2>\n
1. Start with one account, not four.<\/h3>\n
2. VPS quality matters more than VPS location.<\/h3>\n
3. Account aging matters.<\/h3>\n
4. Funding source diversity is more important than it looks.<\/h3>\n
5. Document everything as you go.<\/h3>\n
If you are considering your own setup<\/em><\/h3>\n
\nOr watch the live dashboard first<\/a><\/p>\n<\/div>\nClosing thoughts<\/h2>\n
Subscribe to BJF research<\/h3>\n