Case Study — Algorithmic Trading Engine

The Touch-Fire Strategy Engine

One pluggable parent/child engine — five families of algorithms. Smart order routing, iceberg, synthetic stop orders, event-driven options trading and cross-exchange crypto arbitrage, all built on a single, recoverable strategy architecture. Delivered 2015–2018.

Executive Summary

Rather than building a one-off system for each algorithm, Touch-Fire built a common Strategy Engine Container Model: a parent/child order architecture in which a parent order is routed by its StrategyId and RoutingType to the relevant strategy, which then generates and manages child orders into one or more venues. Every child links back to its parent, so the whole strategy can be displayed, recovered after a restart, and fully audited.

On this one foundation, Touch-Fire delivered a growing catalogue of production algorithms — from a smart order router to route to our off-exchange dark pool matching engine, through iceberg and synthetic stop orders, to an event-driven options algo triggered by earthquake signals and a cross-exchange cryptocurrency arbitrage engine.

The Engine Core

An order is recognised as a strategy order by its routing type and directed to the strategy engine instead of straight to the matching engine or exchange. The engine subscribes to prices, monitors trigger conditions, and emits child orders — keeping parent state durable so it can be rebuilt on startup.

Parent Order (StrategyId) → Strategy Engine → Child Orders → Match / OSE / Remote Venue

Careful warm startup sequencing — recover child orders, relink them to parents, then re-enable order and execution-report flow — ensures the engine never processes a live fill while still rebuilding its state. Every routing decision logs its inputs so the logic can be reconstructed after the event.

The Algorithm Catalogue

The SOR compares the best bid/offer across a primary market and related remote venues (for example OSE against SGX and CME on the Nikkei 225) and routes child orders to the best price. Its defining innovation is that, because Touch-Fire's own matching engine is one of the markets, the router has shared memory access to latest matching engine prices — eliminating one whole side of execution risk.

  • Builds a linked list of secondary instruments per primary instrument at startup, for instant cross-venue price comparison.
  • Executes immediately against local depth, and places fill-and-kill child orders to remote venues for the remainder.
  • Handles partial fills, rejects and leg risk by re-evaluating the remainder against local depth.

The same parent/child model powers liquidity-friendly and protective order types:

  • Iceberg orders reveal only part of a large order to the market at a time, slicing child orders behind a single managed parent.
  • Synthetic stop orders monitor price and trigger into the live market on stop-loss or take-profit conditions. Triggers can fire on last traded price, best bid or best offer (via the order's qualifier), with the client explicitly choosing stop-loss versus take-profit to avoid ambiguity in fast markets.

As a demonstration of the engine's flexibility: an algorithm that places options orders automatically the instant a major earthquake is detected. An external signal device feeds trigger parameters into the algo immediately upon detection.

  • The algo continuously pre-selects target options by underlying, put/call class, delivery and premium threshold, keeping the selection hot as prices move.
  • On a valid signal it instantly sends orders for the pre-selected options, sizing each from the client's assets, a configurable ratio, and the earthquake's impact ratio. Repeated trigger protection included.
  • Runs as a long-lived good-till-cancel parent order, so it is always armed and ready.

An arbitrage engine that tracks prices across more than a dozen global cryptocurrency exchanges and acts when a profitable spread opens up. It reuses the SOR's related-instrument machinery to iterate venues and compare opposite-side prices.

  • Connectivity designed across major venues of the era — Bitstamp, GDAX, Kraken, Poloniex, BitFlyer and more — over FIX, WebSocket and REST.
  • Evaluates each opportunity net of trading fees and execution quality, then places simultaneous buy and sell legs, managing the pairs to avoid a half-filled position.
  • Built-in "rogue algo" protection — position checks, throttles and accumulated-P&L guards — to prevent runaway trading.

Results & Impact

A single parent/child engine that turned each new algorithm from a ground-up build into a focused, repeatable extension — iceberg, SOR, stop, earthquake and arbitrage all share the same core. Running the smart order router colocated with the matching engine removed an entire leg of execution risk — a structural advantage few vendors can offer. Durable parent state, disciplined startup recovery, full decision logging and rogue-algo safeguards make the engine dependable in live, regulated operation.

Related Capabilities

The microsecond matching engine with embedded smart order router. The full institutional crypto exchange where much of this multi-venue work was proven. Integrating a third-party algo engine to Japanese futures markets.

Building a Custom Algorithm?

Touch-Fire's strategy engine has hosted everything from smart order routing to event-driven options trading. Talk to us about the algorithm you need.

Contact us about algorithmic trading.