In a joint paper with Christoph Schlegel (Flashbots), @AliTslm (@EntropyAdvisors), Ko Sunghun and @totorovirus (Matroos), we look into the performance of TimeBoost fast lane bidders, in particular, the correlation between the bids and profit estimations - 5s markouts. Short 🧵:

@AliTslm @EntropyAdvisors @totorovirus The correlation between winning bids and markouts is weak across bidders, suggesting that bids are a noisy predictor of extracted value. The correlation slightly improves when comparing paid bids (the second highest bid) instead of winning bids to markouts.

@AliTslm @EntropyAdvisors @totorovirus We attribute this to the fact that the auction is more of a common value type, as opposed to private value type auction. In all settings, the relative order of the three most frequent bidders' performance remains the same, together with their absolute profits.

@AliTslm @EntropyAdvisors @totorovirus Bids and markouts aggregated over long time intervals exhibit much higher correlation, indicating that bidders detect trends much better than identify when the high arbitrage value is exactly available.

@AliTslm @EntropyAdvisors @totorovirus One possible explanation for this is the fact that the correlation between previous minute markouts and current minute bids is significant, suggesting that the previous minute markouts is used to predict the next minute value when bidding.

@AliTslm @EntropyAdvisors @totorovirus Bids and markouts on shorter time intervals are correlated even weaker than the full minute interval. We attribute this counterintuitive finding to a “cold start“ problem.