Looking Forward not Backwards at HP

“I skate to where the puck is going to be, not where it has been” – Wayne Gretzky

Asset Management is, in many ways, oriented to backward looking information. It makes sense since it’s a very performance driven marketplace and past performance is how we evaluate investors. Most investor letters generally focus on explaining the last quarter’s or year’s performance, and portfolio statements show performance YTD and for the trailing 1 year, 5 years and even 10 years. Fund rating companies show past performance vs. peers and relative to the market, and emerging fund managers need a track record to be taken seriously.

Past performance in many ways is the only valid tool we have to determine which fund managers to give our money to. Since there are no better gauges for future results, it’s the best factor we have. It’s far from ideal though, and even prospectus’ warn people, “past performance is no guarantee of future results”.

The same holds true for the tools that professional investors have at their disposal to support their investment decisions. Market data terminals are very good at organizing masses of financial information into a usable format, but most of this information is also backward looking, such as trailing P/E ratios or company results from the previous quarter. Price charts in all their iterations show us the recent and historical performance, all up until the last tick. Money managers are overwhelmed with market data and analytics that show where the puck has been, but few help determine where it is going.

EidoSearch leverages content based search technology to generate return and volatility projections from Big Financial Data. In plain English, we generate predictive analytics. Our technology has previously been applied to deep scientific fields like Neurosciences, helping data scientists find relationships in brain activity and cognitive functions. Taking the current pattern from an EEG, and comparing it to similar patients in the past, has proven predictive to the types of cognitive skills or issues that a patient could be faced with.

We have now applied this powerful technology (we can accurately search through 100 million patterns in 1 second) to financial time series data like price trends. Our technology looks not just at recent history, but sifts through the entire past to capture actual returns from statistically similar environments – the basis of our projected return distributions. We’ve validated, through 7.6 trillion pattern comparisons and 5 million predictions, that this is predictive to future returns and volatility across time horizons from 1 day to 1 year forward.

Professional investors have quantified and automated parts of the investment process effectively in the past, e.g. with trading and order implementation (TCA and Algos). Now the tools are available to quantify when to build or trim a position, when to lock in gains, identify stocks that may be mispriced and more generally to better understand forward return distributions and the impact on profitability and risk management.

For this week’s market call, we found an energy company with 19x the upside to downside and 85% probability the stock will be ABOVE its current price in 1 month.

In 70 of 81 historical instances the stock is in the green in the next one month per the Bar chart below:

Have a great week!