Engine Simulation

We simulated and compared randomly generated order books run through Elektro and a standard auction-based engine.

Test results for Configuration 1: crossed markets in puts, calls and forwards show:

• Elektro ME significantly outperforms a standard auction-based ME

• Increasing the number of contracts increases Elektro’s outperformance

• Elektro ME outperformed standard auction-based ME by 44%, under crossed puts calls, and forwards market configuration

• Elektro ME outperformed standard auction-based ME by 52%, under crossed puts calls, and forwards market configuration

• Elektro ME creates significant liquidity that otherwise would not exist under:

  • An uncrossed puts, calls, and forwards market configuration

  • An uncrossed puts, calls, forwards, and binary and barrier options market configuration

  • A standard auction-based ME, by definition, cannot find matches under these circumstances

Configuration 2: Uncrossed markets in puts, calls and forwards

• The outperformance of Elektro ME increases in illiquid markets. In uncrossed markets, simple matching finds few or no matches and the value of pooling liquidity across instruments increases.

• In our simulation, Elektro ME outperformed the simple matching engine by 2,149%.

Mixed Integer Linear Programming (MIP) Optimization

Elektro allows for organic continuing inclusion of new products in ways that naturally do NOT cannibalize existing liquidity; to the contrary, more products create synergistically MORE liquidity among existing contracts. The ability to incorporate statically replicable structured products and collateral swap ones open up huge markets that are currently unattainable. This is done via Mixed Integer Linear Programming (MIP) Optimization. MIP optimization allows searching for clearing prices and associated sets of consistent orders satisfying them that maximize executed volume and satisfy best execution requirements.

Maximizing volume is challenging, as it involves optimizing a continuous quantity (volume), subject to discrete constraints. MIP utilizes advanced heuristics to perform an efficient search of the solution space, solving the problem using augmented Linear Programming tools (Simplex algorithm):

• Branch-and-bound methods enumerate possible values for integer variables and prune away provably un-optimal ones

• Cut generators augment the problem with additional constraints, which guide the standard, continuous-valued Linear Program solutions toward integer values

At current, Elektro is capable of processing auctions of 3,000 orders within 5 seconds, with scope for additional optimization to further improve auction processing time.