How to be 99.9999% accurate when your system is only 49% accurate

99.9999 win percentage is an odd statistic

As a recent Securities and Exchange filing reveals, the company, operated by a litany of some of the exchange world’s top executives, boasts that out of 1,485 trading days it has only one losing day. This is the key statistics that left those familiar with algorithmic trading scratching their heads.

On the surface this 99.9999 win percentage is a rather unworldly performance statistic in the world of algorithmic trading.

Virtu Financial is not a trend follower

The most popular managed futures strategy, trend following, has an average win percentage near 55 percent. Trend following might not be the best algorithmic strategy to compare to Virtu, however, as the firm claims in its S-1 that their trading “is designed to be non-directional, non-speculative and market neutral.” Micro-trend following and benefiting from market moves in one direction is a popular high frequency trading strategy, but based on their S-1 this is not the primary strategy.

This doesn’t explain the win percentage.

The highest win percentage of all managed futures strategies, near 75 percent, is short volatility, which is also the least popular strategy. While the strategy is known to win most of the time, the key statistic is to understand its small win size and large loss size. In managed futures the size of a trader’s wins can often be more important than how often they win. In the case of short volatility, while they win most of the time, when they lose they lose big – with an average loss size that is close to double that of a discretionary trading category, for example.

While risk in Virtu may exhibit strong downside volatility during crisis, much in the same way market crashes bankrupt many individual market makers in the golden days of the trading floor, comparing Virtu’s strategy to a short volatility strategy is inaccurate.

Perhaps the most applicable managed futures strategy to benchmark might be the relative value / spread arbitrage category. The spread-arb strategy has a high win percentage, near 60 percent, and it also has the best win size / loss size differential. The strategy works by buying one product and then selling a related product. The directional strategy works when a market environment of price relationship dislocation occurs.

While the fit isn’t perfect, nonetheless the most relevant managed futures strategy for which to compare Virtu is its direction-less, market neutral approach taken by certain spread-arb CTAs. The primary difference being Virtu doesn’t hold positions for directional profit. What they do, much like a short term trend follower, is take a position and then immediately lay off risk in a hedge. For instance, they may buy oil and then immediately hedge that position in another market and perhaps even using a derivatives product with different product specifications. In the olden days one could simply describe this as a “market making” strategy, but in the new school world of high frequency trading, separating two-sided liquidity providers from directional trend followers has oddly become more difficult.

99.9999 percent daily win percentage overshadows 49 percent intra-day win percentage, highlighting the importance of win size

When comparing Virtu to known managed futures strategies, the 99.9999 win percentage sticks out like a sore thumb – until you read the next punch line in the most recent S-1. That is when the firm reveals that its win percentage on an intra-day basis is 49 percent.

This puts the pieces of the puzzle together. The 99.9999 percent win percentage needs to be considered in light of the 49 percent intra-day win percentage. This highlights the fact that Virtu, by logical default, is benefiting from size of win. Just like a trend follower, it isn’t always win percentage that matters most but size of win and controlling loss. This is likely the secret sauce inside Virtu’s success.

This article originally appeared on Virtual Walk and was authored by Mark Melin.

The post How to be 99.9999% accurate when your system is only 49% accurate appeared first on MetaTrader Expert Advisor.

Source:: How to be 99.9999% accurate when your system is only 49% accurate