So, where does the idea of product managing incentives in complex systems come from? For the past two decades, I have worked in electronic trading. This includes working on the strategy, products, practices, and market structure that are features of the current exchange landscape. During this time trading migrated from the physical exchange floor with shouting, hand-signaling brokers, to humming computer-based order matching engines hosted in datacenters.
Fairness, the Intended Outcome
One of the primary products offered by exchanges is modern matching engines. Most match buy/sell orders based on a price/time algorithm. For example, the buy/sell orders with the highest/lowest prices match first. When there are multiple buy/sell orders at the same bid/offer price, the oldest gets priority in the queue.
This is deemed to be a “fair” methodology. That is to say, it rewards those orders that are willing to trade at the most aggressive prices in the order of their arrival at the exchange.
Racing, the Unintended Incentive
The simple incentive to be first in line has created some unintended consequences: namely the preeminence of speed.
Trading firms sought an edge and exchanges responded (to that incentive) by offering products that promote speed, at a premium price to slower products. For example, the most expensive trading infrastructure and data feeds now measure speed in nanoseconds (billionths, i.e. 0.000000001, of a second) and disproportionately benefit customers that rely upon speed.
Speed-driven customers enjoy the advantage. However, a zero-sum arms race has emerged among them to procure ever-faster technology and data. Those who can’t compete on the basis of speed may feel the system is unfair. Those who can resent their escalating costs.
This suboptimal racing outcome presents an opportunity to re-examine the incentives that prioritized speed.
New Incentives, A Product-Driven Approach
A product-driven approach is for competing exchanges to offer alternative matching models to optimize for a different definition of “fairness”. One such approach is to deliberately slow things down with a “speedbump”. Other options include frequent batch auctions that prioritize time in defined intervals that neutralize the benefit of speed in the absolute, or a size/time algorithm that prioritizes orders based on their relative sizes. Each of these models requires trade-offs. They create a set of incentives that may benefit some categories of customers more than others. The key is, the product manager is in a position to decide which incentives to prioritize, and let participants choose where to trade according to their preference.
Ultimately, there are many competing interests and constraints in a complex system, especially one that is heavily regulated. No single entity, let alone product manager, can independently impose structural change.
However, thoughtful product managers can start by examining negative incentives and structures that present opportunities to create different, positive incentives through product design. Resulting competitive advantages may be fleeting if competitors follow suit. However, the overall system can evolve to be structurally improved around the newly embedded positive incentives. Negative outcomes can be the result of actors responding individually to incentives. Similarly, individual actors creating new incentives (itself a response to incentive) create positive outcomes.
About the speaker
Serial entrepreneur, adviser and investor with experience building successful, disruptive technology startups. Industry speaker and panelist. Domain expertise in regulated financial markets, market structure, trading technology, electronic marketplaces and exchanges, blockchain and distributed ledger technology, cryptocurrencies and crypto assets. International experience in North America, Latin America, Europe and Asia spanning product, strategy, operations, and business development. 12 market/exchange launches in multiple jurisdictions (N America, Europe, Asia) and asset classes (equities, futures, precious metals, cryptocurrencies).