Most short-term systematic trading I've seen last from a few seconds to within a single trading session. Anything longer can be called long-term because overnight risk has to be taken into account. The advantage of short-term systems, 1. can be backtested more vigorously with more results to analyze from 2. lower margin -> higher leverage 3. no overnight risk The advantage of long-term systems, 1. more tools available (e.g. commitment of traders, open interest, short interest ratio, etc.) 2. lower transaction cost 3. can include complex trading strategies (combination of options, futures, and underlying, etc.) Both can be very profitable.
This is a interesting topic except I think we need to break it into four groups 1) Intra-day trading , up to 2 day hold 2) Short term trading 2-7 day hold 3) Intermediate term 7-20 day hold 4) long term 20 day to six months When developing mechanical systems I have found different strategies break themselves down in these natural timeframes. Another issue is the short term to intermediate term systems are normally counter trend. They might be volatility breakout system for example but the setup is off of a counter trend move in the scope of a trend. Other methods like intermarket analysis which are intermediate term are true countertrend. Long term system are by nature more likely to be trend following. This sets up the issue of what are you trading ? 1) I would say for stocks, you want to be short to intermediate term unless you are using fundamental analysis as part of your system approach. 2) True commodities , longer and intermediate term are better. I have not seen too many shorter term system average more that $200.00 a trade on a basket of real commodities. 3) If you are trading cash forex , then longer term means you need to understand the effect of interest rates and how they effect the P/L. I will make some other comments soon ,when I have more time.
I would echo what others have said, especially man's point about compressing the same number of trades to "converge" to profitability faster. You can generally reduce variability if you can shorten your time frame effectively, add more trades, and still retain your edge. This positive effect of shortening your time frame is especially pronounced when compounding your profits. The one thing I would caution against is thinking you can backtest a short term system with less data as it will be statistically significant due to having more trades in a shorter time period. You need to remember that markets can often move in regimes and that some intraday systems are more susceptible to erosion if their profit per trade is especially small. For example, you may have a high freq system that trades 2000 times in the period of 6 months and shows profitability. 2000 trades is a lot, but dude, its still only 6 months! Gotta be careful with this kind of thing.