This video recording was showcased at the online Bloomberg App Portal event last December 15th. This explains the CI Futures, a highly accurate app that forecasts commodities, currency pairs, and equity indices. Our CEO and founder expands on the AI and machine learning technology behind this and how our clients experience success by adding CI Futures in their arsenal of planning tools. Bloomberg Terminal clients can access this right from their dashboards by typing APPS CI<GO>
Hi everyone. This is Tony Nash with Complete Intelligence. I’d like to walk through our app CI Futures that is on the Bloomberg App Portal.
Complete Intelligence is based in Houston, Texas and we’ve built a completely automated, globally integrated artificial intelligence platform that looks at cost and revenue forecasting. We look at markets and what we’re looking at is better ways to help people understand future prices and costs.
One of the key issues we’ve seenwith Covid and with all the volatility and uncertainty in the market is companies and investors trying to understand how they plan for that environment. Over the past year, we’ve had a number of companies come to work with us to look at their revenue planning, to look at their exposure to commodities, and to look at their overall exposure to markets.
We’ve used CI Futures which is available on the Bloomberg App Portal for example with a global chemicals firm. And with that we’ve helped them reduce their forecast error rate for one of their major products to 4.4 percent on an absolute percentage error basis. At the same time, we’ve helped a global mining firm reduce their error rates on their gold forecasts by 38 percent. As you can see, those two companies have done extraordinarily well in equity markets this year. Whether it’s directly a result of that, we’re not exactly sure. But the message here is that companies that take analytics seriously, who deploy say artificial intelligence solutions can respond better to markets and to customers.
As we’ve put together CI Futures and as Complete Intelligence has grown as a company, we’ve really built the company on the basis that the status quo planning process and status quo forecasting process is just flawed. Many companies and many investors rely on consensus forecasts which are very similar to forward curves. And consensus forecasts have a fairly high error rate, 20 percent, in some cases more, in some cases slightly less. But generally, consensus forecasts have a high error rate.
People take those in when they’re planning their portfolios or planning their investments or their budgets as a benchmark for their activity. What we’ve seen is with Complete Intelligence, we have on average of 4.6 absolute percent error rate with our forecasts across currencies, commodities, and equity indices. Those that are available on the Bloomberg App Portal.
We’ve helped companies really plan and save millions of dollars and we help investors understand their risk exposure. When we look at consensus data, a real tangible comparison we do is look at the full year 2019 consensus forecast to our Complete Intelligence forecasts. These are actual consensus forecasts taken out of the report referenced below this table. And we look at the consensus forecasts in the yellow column. The Complete Intelligence forecasts in the blue and actual forecasts in the gray.
When we look at the right-hand side of this table, we see that Complete Intelligence beats consensus forecasts 88 percent of the time. There’s a tangible metric there to show how we’ve performed across say crude oil, industrial metals, and precious metals. When we do this, Complete Intelligence, CI Futures is really powered on publicly available data. We do two things.
The first thing we do is take in publicly available data. It’s about 15 and a half billion data items and we’ve built a very large ecosystem to understand how the world economy and how global markets work. We run that process twice a month and we’re taking billions of calculations to understand price action. We work on what’s called an Ensemble Approach.
We take fundamental methodologies. We have technical methodologies as well and we look at the best configuration of those methodologies to forecast the assets that are within CI Futures. It’s also possible to take in custom assets from investors and from companies so that we can help them understand their specific items that they’re looking to forecast, whether it’s a cost, whether it’s an investment asset, whether it’s a sales revenue item, something like that. We can take those in from customers and we can put them on a closed login for customers so that they can understand the forecast outlook for those assets.
With our methodology, as I said we’re using an Ensemble Approach with billions of items within the context. We actually utilize about 10 thousand scenarios for every asset we forecast. So twice a month, on the first of the month and the 15th of the month, we’re gathering all the publicly available data that we use. We’re testing every line item
every month. We’re weighting that line item and reconfiguring the algorithm and then we’re reforecasting those assets. With every asset, the market movements, the movements in the world economy, the movements and things like world trade and so on and so forth are taken into account so that we can understand what that cost or what that asset price will be every month for the next 12 months.
So CI Futures is a subscription service on the Bloomberg App Portal. We have a very simple interface. What we want to do is present users with a monthly interval outlook over the next 12 months. We provide high, base, and low scenarios within that forecast context. Now, the base and low scenarios are one standard deviation applied equally over 12 months. We’re not running discreet scenarios for a high and a low scenario. We’re taking the baseline, the likely scenario and and putting a one standard deviation band around it.
What’s really key for us here is that 97 of our forecasts typically have less than 10 percent absolute error. 84% have less than 5% absolute error. We really focus on error performance and with every iteration of our forecast, we’re collecting those error data. We’re understanding how we perform to market and where necessary, we’re making system-wide changes to take account for say volatility or say government intervention and so on and so forth. What’s key to understand here is we are not manually adjusting any individual asset within CI Futures.
We continue to iterate the overall ecosystem of CI Futures so that we can best account for volatility. For example after Covid, we introduced much more volatility-specific methodologies so that we could identify ways the volatility would impact market prices. With CI Futures, we’re really trying to help companies and investors do more with less. We have a subscription available for individuals on the Bloomberg App Portal.
We’re trying to help people improve net income, improve their cash flow, and improve their valuation or market cap. It’s really important for us to help our clients and to continue to understand how these assets move. If you can check us out on the Bloomberg App Portal, we’d really appreciate it. Thanks very much.