TitanArb — A Live-Capable DeFi Arbitrage Engine on Arbitrum One

Hi everyone,

I’ve been building a DeFi arbitrage engine on Arbitrum One called TitanArb, and I’d really like to get feedback from people who understand the Arbitrum ecosystem, DEX infrastructure, MEV, and execution systems.

TitanArb is written primarily in Go and is designed around real-time market state, cross-DEX opportunity discovery, bounded multi-hop routing, executable quotes, trade-size optimization, simulation, and atomic execution using Aave V3 Flash Loans.

The current system works with Uniswap V3 and Camelot V3 / Algebra, evaluates 2/3/4-hop routes, and includes:

real-time WSS market updates

dirty-pool based state refresh

pair intelligence and dynamic market selection

route scoring and route memory

cross-venue opportunity detection

bounded quote and optimizer budgets

gas, DEX fees, price impact, slippage, and flash-loan premium in profitability calculations

multi-provider RPC routing and latency-aware failover

pre-execution simulation

live atomic execution with on-chain safety checks

One of the most interesting challenges has been avoiding both extremes: evaluating everything and overwhelming the RPC/runtime, or filtering so aggressively that potentially useful opportunities never reach the economics stage.

I’m currently spending a lot of time on opportunity discovery quality, particularly understanding exactly where routes are rejected throughout the pipeline and whether the system is being too conservative before an opportunity reaches executable pricing.

The repository is public:

I’d especially appreciate feedback on:

route discovery and candidate selection

market/pair universe design

cross-DEX arbitrage on Arbitrum

RPC/state freshness architecture

MEV and execution assumptions

areas where the system may be filtering opportunities too early

additional DEXs or market structures worth supporting

Critical feedback is very welcome. I’m much more interested in finding weaknesses in the architecture than simply promoting the project.