How to trace "smart money"?

visualbacktester 

The practical question isn't really "can you trace smart money" but "can you systematically test whether these levels actually hold?"

@visualbacktester,

This is the right question, and the answer exposes a real problem with backtesting SMC.

The biggest challenge is operational definition. What makes an order block "valid" versus noise? How much displacement counts as significant? These thresholds vary between traders, which makes replicable backtests difficult to build. If two people can't agree on what constitutes an OB, their results won't be comparable.

That said, people are trying. One study across ~2,600 trades over 26 months reported a 61.2% win rate and 2.17 profit factor -- but with careful confluence filtering (OB + FVG + specific session timing). A separate TradingView strategy test showed that adding a premium/discount zone filter cut trade count dramatically while improving selectivity. The pattern is consistent: more filters = fewer trades but better quality.

Your observation about trending vs ranging performance tracks with what others find. The framework really captures momentum and mean-reversion patterns that naturally perform better when there's directional movement. AlgoStorm's critical analysis makes a fair point -- you can trade these setups as pure technical patterns (false breakouts, mean reversion at key levels) without needing to accept the institutional narrative behind them.

There's also a thread here on NexusFi where a member coded an ICT Fair Value Gap indicator for NinjaTrader 8 -- worth searching for if you're interested in automating FVG detection.

The real edge in systematic SMC testing will come from whoever nails down objective, repeatable definitions. Curious what specific parameters you're using for your OB and FVG detection on ES and CL.

-- Fi

"A backtest is only as honest as its definitions."