Yesterday I shared my first results.
Today โ things got real ๐
๐ What changed:
โ Before:
โข 0 trades
โข Wide spreads โ no execution
โข Bot looked โsmartโโฆ but did nothing
โ
Now:
โข 20+ trades in session
โข Balanced BUY/SELL cycles
โข Real market interaction
But hereโs the twistโฆ
๐ฐ PnL Result:
โข Trade PnL: positive
โข Total PnL: slightly negative
๐ค Why?
Because of one brutal truth:
โ ๏ธ Fees > Edge
Even with perfect execution, small spreads (0.1โ0.2%) get eaten by fees.
๐ฅ Biggest realizations so far:
1๏ธโฃ A bot that doesnโt trade = useless
2๏ธโฃ A bot that trades too cheap = also useless
3๏ธโฃ Execution is step 1โฆ but EDGE is step 2
๐ Market context:
โข SOL moving in tight range (~82โ84)
โข Low volatility โ harder to capture profit
โข High competition โ spreads must be precise
๐ง What I changed:
โ Removed artificial spread distortion
โ Fixed inventory skew behavior
โ Increased effective spread (0.25%+)
โ Focused on PnL per TIME, not per trade
๐ Current goal:
๐ Fewer trades
๐ Higher profit per cycle
๐ Positive net PnL after fees
๐ก My current hypothesis:
โMore fills โ more profit
Better edge + controlled fills = profitโ
๐ค Question for traders & bot runners:
What matters more for you?
1๏ธโฃ High frequency (many trades, small edge)
2๏ธโฃ Lower frequency (fewer trades, higher edge)
๐ Next step:
Testing dynamic spreads based on volatility (ATR-based)
Will update results soon ๐
#Hummingbot #MarketMaking #SOL #CryptoTrading #AlgoTrading