Long-Only Performance Since July 11, 2011

Higher Returns, Lower Risk

 PredictorS&P 500
Average Daily Return-0.0185%-0.0003%
Standard Deviation1.1339%1.8017%
 
Total Return-3.4783%-2.3359%
Beta0.5821

Strictly Live Statistics

Most competing traders and algorithm designers pride themselves on over-fitted backtesting results, but we chose otherwise. Since July 11, 2011, our system has reported all of its desired trades at 5:00am on the day of trading, set to be executed at 9:30am. Short of time travel, this makes it physically impossible for our system to "cheat" as a result of curve-fitting, programming errors, hindsight, or experimenter bias, often found in backtesting reports from technical analysis.

We have achieved higher average returns at lower standard deviations than the market in both live and backtested settings, suggesting that our 11+ years of backtesting is not from over-fitting random price patterns, but rather from finding a reliable predictor of price change from news analysis. We track the algorithm's live performance using third-party virtual portfolio managers, ensuring that had we actually conducted the trades that the system reports at 5am every morning, we would have been able to beat the market by the margins reported above. We do not incorporate trading fees into our performance as they are impossible to predict or estimate without actually conducting the trades. Furthermore, we only average 4 trades per week and this would not have a significant impact on the reported results.

We currently operate a long-only model that holds a subset of the current S&P 500 constituents. While the model is susceptible to failing markets, its prior outperformance of the S&P 500 implies that one could obtain riskless profit by shorting the market and longing our portfolio. Subscriptions to what the model is holding are available for a small fee.