I agree with this. I think there is an inevitable period of high uncertainty in the transition from backtesting/paper-trading to live trading. Parameters whose optimal value seemed set or unknown take on more solidity as the "unseen" data gets run through the algorithm. It is definitely best to take it slowly during this phase. It gets a little bit easier to know what to look for in each new market, but there are still going to be growing pains, so erring on the side of caution makes the most sense.
I simply love the feedback I've gotten, when I try to filter low volume days such as these past weeks. Not much of trades were done and helped me save quite a bit..
I saw the ATR's earlier in the year drop to 2, the herding instincts on the futures, etc. I'm primarily a swing trader but was looking for other markets and wound up with a big ass filter, kinda hit me over the head. If you got a swing trading system with a 70% win rate, it made sense to trade that particular stock intraday also, odds are on your side. Kinda hard to backtest this stuff, just got bored with the YM,NQ,ES,EUR, etc. But, if the intention is to focus effort on just a couple trading vehicles (herding), it's worthless, as well as this post. Various filters.