dCPI
Chainlink Fall Hackathon '22
Excited to share our submission to Chainlink’s Fall Hackathon!
Inspiration
Economic data releases have become sensationalized and have dramatic effects on markets as we have seen over the past year with CPI readings. It's strange that the world has access to so much data on current market conditions, yet we fully rely on the FED to release their official number on indicators like inflation. The traditional FED YoY CPI reading is well known as a lagging indicator. With the emergence of alternatives such as truflation’s YoY CPI reading, we thought there must be a way to forecast the FED’s reading with this new source of live data. Why must inflation be calculated on such a backward looking basis rather than a live data point we can all adjust to day by day? We at dCPI think inflation readings should be much more accessible. This is why we set out to find an accurate forecast of the FED’s CPI readings leveraging truflation’s open source data and backed it up with a tamper resistant format using smart contracts and Chainlink oracles.
What it does
dCPI takes the relative % change month to month of truflation’s YoY CPI readings and applies it to the Federal Reserve's official data points to forecast official readings three months in advance. This three month window exists because the Fed needs to take time to compile data and calculate CPI compared to truflation which is a live feed of the year over year change in pricing data. After applying this lag to Truflation’s data, we were able to create an inflation model that accurately predicted the Fed’s official YoY CPI readings for 6 consecutive months. The accuracy with which it predicted was within 1 tenth of a percent. We are very proud of this discovery and look forward to seeing its performance going forward, as it turned out it could predict CPI readings 3 months in advance with the same accuracy.
How we built it
We built our model using historical data provided by truflation and monitored it for some time. As it became more and more clear with each reading being within 1 tenth of a percentage point of the official FED reading, we decided to make a public, tamper resistant version of our model using truflation’s Chainlink Oracle.
Challenges we ran into
The biggest challenge that we, among others, had was trying to figure out how to utilize truflation’s data effectively. Calculatively it was difficult to assess, as different metrics of CPI can vary and sector weightings can also differ. It was already confusing since truflation’s absolute number always seemed much higher than the FED’s official reading. But it was always clear that if truflation was increasing, then the FED’s number was likely to increase as well. This held true conversely: if truflation was decreasing, then the FED’s was likely to decrease. Something was always off between the two metrics however, and it seemed that truflation was actually a leading indicator of the FED’s readings. When we put the peak inflation months of truflation and the FED’s data in sync, the data fell into place. It was a three month difference in peak months which makes sense when considering that the FED’s data is likely compiled over the following three months (the length of one quarter) and then released well in advance for them to plan accordingly.
Why we took our discovery public
We decided to take our discovery public rather than keeping the advantage in CPI data readings. Ultimately, we thought it more prudent to make the information public because it could contribute to making economic indicators like CPI more transparent, or at the very least more accountable. Economic readings have become sensationalized and can cause dramatic swings in markets on their release dates. There are groups of people who are aware of the readings before the general public, and we believe this is inherently unfair. The general public and investors should not be holding their breath each month as a government agency releases a number that will dramatically affect their lives. It is an archaic system and we should be leveraging open source data such as that provided by truflation to make markets and the economy fairer for all its participants.
What we learned
Truth is accurate.
What's next for dCPI
We hope that dCPI can become the industry standard as an accurate, up-to-date, and honest economic indicator.