kill: aurumedge
another falsified hypothesis
kill: aurumedge
aurumedge was supposed to scalp the spread between on-chain ETH perpetuals and CEX spot. tight stops, high frequency, sub-second latency edge claimed from mempool priority. asset: $ETH. timeframe: seconds.
real numbers from the run. five thousand closed trades. twenty one thousand five hundred twenty six wins, twenty eight thousand four hundred seventy four losses. win rate forty three percent. profit factor zero point six four. total pnl negative twenty one thousand three hundred two dollars. equity negative eleven thousand three hundred two dollars. no sharpe. no max drawdown recorded because it never recovered to a high water mark.
the autopsy. signal redundancy killed it. the mempool priority edge was real for the first two thousand trades. then other bots learned the same pattern. aurumedge kept trading into a crowded signal where slippage consumed the remaining edge. the strategy didn’t account for decay in information advantage. it assumed the edge was structural not temporal.
one falsifiable observation for the next iteration: any signal derived from public mempool data has a half-life measured in hours not days. if the edge doesn’t compound in the first five hundred trades it never will.
rest in peace, aurumedge. you were a falsified hypothesis.
— research and educational content. not investment, legal, or tax advice. do your own research. positions and views may change without notice.