This is more a prompt for a discussion that a black and white answer but would value your opinions.
It is generally suggested a long backtest with lots of trades are more likely to be a robust system. Which i do agree with.
However, i think that depends on the timeframe of the trade. For example, say you had an intraday system that trades a couple of times a day. Would we be eliminating a lot of potentially good systems by making the "go-live" criteria to produce consistent results for the last 10 years? Fantastic if that is the case but the intraday stucture of the market changes so rapidly. You could get some high level blunt tool that probably generates alpha over 10 years, but to get some really stand out results i'd argue a strategy that has performed extremely well over the last 6 months would likely be more suited to todays market.
The half life of the system is likely to be far quicker for an intraday system performing well over the last 6 months than the last 10 years, but i suggest also probably more in tune with the next couple of months.
If a system had generated >500 trades over the past year with good results would you be happy to trade it live with the ambition of it surviving a good 4-5 months.
Or is long term historical results still a must?
Would be interested in a few others opinions.