Updates on plutarc
When I wrote the launch post, plutarc connected to one exchange, had 28 strategy components, and ran on a subscription model.
The pace since has been less a sprint than a sustained reckoning with what the platform needed to become — not in vision, which was always clear, but in coverage, in depth, in the kind of operational honesty that only reveals itself under actual use. Bybit came first, then Binance, then Kraken Futures, each properly integrated rather than approximated: candle fetchers, instrument discovery, latency probes, correct inverse contract arithmetic. The strategy component library has grown from 28 to 48, with additions that reflect genuine expressiveness gaps rather than padding — funding rate velocity, open interest divergence, HTF support and resistance zones, a volatility regime classifier that distinguishes between a slow market and a genuinely dangerous one. A fifth component role was added: the risk gate, which sits between signal and execution and blocks entries when the conditions don’t warrant the exposure. You can now add the same component multiple times with different parameters and independent roles, because the same tool tuned differently is sometimes two different tools entirely.
Billing changed, and I think it changed correctly. The subscription model made a certain kind of sense at launch. Usage-based prepaid credits makes more sense now — per-hour, transparent, pay for only what you use, switch deployment infrastructures, run strategies on infra that is designed to support it; swing traders can ignore latency concerns, so they run on cheaper options, scalping may need colocation, best efforts have been made, based on what is publicly available, to get you in the same building as the exchange matching engine, the latency numbers speak for themselves, and configurable Telegram notifications to keep you up to date on how your bot is performing.
The compute layer expanded more than I anticipated. Hetzner was joined by AWS EC2 in Tokyo and Singapore, then Vultr across APAC, then AWS and Vultr in London and Dublin — with the AWS instances migrated from ARM to x86 for lower exchange latency along the way. Regional pricing is now explicit, you can find more details on pricing page.
The bot architecture has also been reworked; every bot is stateless, and designed from the foundation with extensibility in mind, bots now report their IPs to the dashboard for full whitelisting support, plutarc encourages this behaviour, and finally, bot versioning is live — each bot reports its running version on the dashboard, updates are opt-in, and they apply in-place on the same instance so your exchange IP whitelist stays intact.
The backtest tooling has become something I am quietly proud of. Long and short breakdowns, expectancy, consecutive streak analysis, hold time distributions, a proper date range picker, sortable trade tables with CSV export. These are not cosmetic additions. They are the difference between knowing a strategy has a positive expectancy and understanding the conditions under which it earns it — and the conditions under which it doesn’t.
The launch post said plutarc is infrastructure you operate, not capital you hand over. That has not changed. Every effort is being made to stay true to the slogan, “Algorithmic Trading, Orchestrated”, and align the product towards this goal. What has changed is the surface area of what you can express within it, the number of markets you can reach, and the precision with which you can understand what is working. The platform is already substantially different from what launched. There is considerably more to come.
