evo34
82 posts
msg #104162
- Ignore evo34
modified |
1/2/2012 1:20:22 PM
1) Allow backtests of longer than 2 years.
2) In the backtests results, offer more useful daily PnL stats:
a) Average trades per day.
b) AVerage holding period per trade.
c) Median profit/trade.
d) Avg. daily PnL per position.
e) Median daily PnL per position.
f) Standard devation per trade.
g) Sharpe (daily)
h) R^2 of equity curve
i) Avg. annualized PnL per position
3) Offer 'point-in-time' characteristics of stocks. E.g., record the inst. ownership pct. from Sep 2010, and be able to use this data in a backtest, where the data will be pulled according to the current point in time of teh backtest. For these types of historical variables, add: exchange, industry, inst. ownership, div yield, short interest (pct.), market cap, etc.
5) Import past earnings dates and results into StockFetcher (from WallStreetHorizons, e.g.), then make the date and quality of EPS usable fields in backtests. Also, get analyst upgrade/downgrade data from theflyonthewall.com, and integrate it into SF. E.g., "show all stocks where >=3 analysts upgraded it over the past 2 weeks"
|