The cryptocurrency market is fragmented: the same coin on different exchanges costs differently at the same moment. This difference—the spread—is the bread and butter of an arbitrageur. The scheme is classic: buy cheaper on one platform, sell more expensive on another. The problem is that such windows of opportunity last only seconds, and their size rarely exceeds fractions of a percent. Manual trading loses to machines here with a crash.
The Arbitron project offers a systematic solution to this problem, taking on the routine of monitoring and execution. My analysis shows that this is not just another scanner, but a full-fledged infrastructure for delta-neutral trading, where earnings are generated not from price predictions, but from the math of spreads and funding rates.
Why speed is not a luxury, but a condition for survival
A typical inter-exchange spread is 0.01–0.5% and disappears within seconds. The key problem is not in spotting the gap, but in being able to take advantage of it before other participants close it. This is exactly where most traders lose money, mistaking the "raw" difference in quotes for net profit.
On the path from the scanner screen to the actual balance, the spread is "eaten up" by four factors: taker fees on both legs of the trade, insufficient order book depth for the required volume, slippage during execution, and the decay of the spread itself. This is why the so-called "pretty" percentages on regular scanners are most often an illusion that has nothing to do with reality.
Infrastructure: milliseconds decide everything
Arbitron connects to 20 exchanges and tracks about 10,000 trading pairs. But the key difference lies in the approach to data. The platform keeps the full order book in memory, not just the best price. This allows calculating the executable price for a specific volume, for example, $500 or $1000, rather than for a "paper" order of $2 sitting at the top of the book.
The service architecture is tailored to the physics of the market. Each user gets a dedicated AWS server with a static IP in the region closest to the exchange engines. Measurements show that a request to Binance from Tokyo takes ~23 ms, while from Singapore it is already 206 ms. During this time, the order book of a liquid futures contract manages to fully refresh. The trading core is written in Rust, which is critical for minimizing latency on the order execution path.
Backtest as a filter of reality
The project's scanner deserves special attention. Instead of momentary figures, it shows the result of a backtest with a 5-minute delay. The platform runs the last 8 hours of quotes through the same strategy as in real trading, taking into account the trader's personal fees (VIP level, token discounts). This filters out false signals that would not survive even a second of verification.
Each opportunity receives a reliability score from 0 to 100, which takes into account market depth, the number of successful cycles, and the profit margin. Anything below 50 points is already a "yellow flag," a signal that the pair may be too thin and risky.
Risk management and security
The system provides three levels of position protection: a hard stop, a soft exit, and an automatic safety mechanism 3% before liquidation. Particularly noteworthy is the delisting detection mechanism—this is the scourge of arbitrageurs, when one leg of a trade disappears, leaving the second without a hedge. The platform monitors official announcements from Binance, Bybit, Bitget, and OKX, and also cross-checks instrument lists every 5 minutes.
From a security standpoint, the model is non-custodial. API keys do not have withdrawal rights and are stored encrypted (AES-256-GCM) using AWS KMS. A separate encryption key is created for each user, which eliminates the compromise of all data through a single point of failure.
Monetization: pay only for results
The project's financial model looks mature. Pricing differs not by trading volume, but by functionality. On the Trader plan ($99/month), the fee is 35% of realized profit. However, the calculation happens once a week, and losses are carried forward to future periods. If you lost $100 in the first week and earned $100 in the second—you pay nothing. This is an honest approach that aligns the interests of the platform and the trader.
My verdict: Arbitron is a professional tool that addresses the main problem of arbitrage—the gap between theory and execution. It does not promise easy money, but automates complex mechanics, requiring the user to understand the basics of futures trading and be willing to configure strategies. For those ready to spend time learning, this could become a reliable source of steady income independent of market direction. However, for beginners expecting a "wealth button," there is nothing to do here—the $3000 entry threshold and the need to understand strategy parameters are a deliberate filtering of the audience.