It’s better to be lucky than smart*

An extract from our 2015 Annual Letter

*but its easier to be smart twice than lucky twice.

Since G2’s launch in 2011, markets have delivered a wild ride. Indeed, Mr Market has lived up to his schizophrenic reputation from one year to the next, if not from quarter to quarter.


In 2012, large cap and low volatility stocks dominated globally. In 2013 small cap stocks and the reflation trade was the place to be. While in 2014 high yield stocks and deflation was the better bet. 

Not to be out done, 2015 required a focus on mega cap stocks in developed markets – everything else was … messy.

At the time our first annual letter went to press, The Guardian reported that a ginger moggy cat (‘Orlando’) picked better stocks than professionals in their annual investment challenge.

“It's time to crack open the Whiskas," said one money manager, "the cat's got talent."  Sadly, Orlando, failed to live up to expectations afterwards and has entered feline exile.

A key factor which helps sort skill from luck is sample size. This is not just a matter of time; what really matters is the number of trials. For instance, The Guardians ‘challenge’ involved picking 5 stocks – hardly a decent sample – with no disrespect to Orlando.

In contrast, Poker (or chess or golf) champions have played hundreds of hands before making the final table. Similarly, G2’s US Alpha strategy has generated over 800 individual matched trades over the past three years (not counting delta trades).

Three years into our live track record we can begin to get a sense of whether we are winning, or at least not losing:

1.      Reliability:  Hit ratio (#Wins/#All Trades)

Research by Inalytics on the performance of 215 fund managers revealed that the average hit ratio was 49.6%. So managers’ stock picks fell in value just over half of the time. “Good managers” had hit ratios of 51.3%.

G2’s hit ratio, based on over 800 matched trades since inception, stands at 56.6%. Despite the fact this is “better than good”, we remain circumspect: the system is wrong 43.4% of the time!

2.      Positive Skew:  Win Loss ratio ($Won/$Lost)

The best stock picking strategies make more on their winners than they give up on their losers. For G2 this is indeed the case, with winners outperforming losers by 1.4x.  This compares well against the “good manager” benchmark of 1.1x, again based on Inalytics research.

3.      Performance:  True alpha after backing out beta, size, and value exposures

Performance is one thing, but adding value is another. One way to measure the true active returns of a strategy is to risk-adjust its performance (beta, size) as well as removing common factor exposures (value, momentum).

This so-called Jensen’s alpha is +4.1% for G2’s US Alpha strategy annualized over the past three years. This provides some evidence that G2’s process is punching above its risk weight and common factor weightings.

While these outcome metrics are important to track, above all, G2 continues to focus on the nuts and bolts of its process: a good process provides the highest probability of success over time.

As Michael Mauboussin points out, success in a probabilistic field (like investing), requires a constant search for favourable odds, a disciplined weighing of probabilities and overcoming cognitive biases. We think a systematic process driven by machine learning and anchored by fundamentals is a good way to get there.

That is until Orlando, the fabulous feline, lifts his game.