DCA engine
fires when Fear & Greed < 25
One AI brain uses hundreds of audited signals for 24/7 automatic execution, applying the same logic to each client pro rata by capital.
Request accessOrders always go out limit, never at market, and the stop is registered at the exchange the moment the position opens.
market signals scored before each decision
trading engines working in parallel on the same account
with a watchdog that restarts any process that stops1
1 Servers run in more than one region.
assets traded on Kraken, Binance and OKX2
2 Futures on BTC, ETH and SOL only.
The API connection is created with
no withdrawal permission
so the funds never leave the client's own exchange account
Five engines compete for capital: the brain funds each cycle and a controller enforces ceilings. Three are shown here; the deck covers two more.
Ceiling: no engine passes 40% of the account. Positions and free balance are reconciled against the exchange every cycle, and if the balance book and the venue disagree the desk stops adding risk.
fires when Fear & Greed < 25
recentres past 5% drift
covers trend regimes
one ceiling per engine
matched to the exchange
Every order is limit. Every stop is registered at the exchange.
No engine can send a market order, and no stop lives only inside our own servers.
Coordinated pushes across BTC, ETH and SOL are detected within 60 seconds; the desk stops new positions until the move settles.
Limit orders sit 1.5%, 2.5% and 3.5% below price, so a violent wick fills at our level instead of being chased after it passes.
If price moves more than 5% outside the working range, the grid recentres on the new price instead of trading the old band.
One document covers architecture, risk controls, engine rules and the operating record. Access requests are answered in person, not automated.
Same document, same figures, two languages. Nothing in it is a projection of returns.