Have been on Quantopian for a while. Really enjoy the community. This is my first algo which is a momentum strategy (buy winners, and sell losers). I am trying to summarise my recent research here.
Some key items in this algo:
1. Ranking based on three factors: Size: market capital > $2B; Value: low ev_to_ebitda; Momentum: Past 2 to 12 month cumulative return
2. Trend filter: only open trade with SPY_current_price > SPY_mvag(200), if SPY_current_price < SPY_mvag(200) clean positions and switch to TLT(Treasury ETF)
3. Trade: Then long top 30 stocks in the ranking
4. Rebalance: monthly (first day of each month)
5. Weigh: equal weight
Other things factor, want to investigate:
1. Ranking logic (Tricky): Size(large, mid, small), value(book_to_market, evit_ev, P/E etc), momentum(trend signal, past 3 months modified log return slope(from Stock on the move), etc), other(low beta, low volatility, etc )
2. Weight: EW, VW, ERC, MVO
3. Rebalance frequency: weekly, monthly, quarterly? (monthly seams reasonable considering commission, and momentum continuous)
Ideas are from several literature(not limited):
1. Value and Momentum Everywhere, Clliford Asness ect 2013.
2. Stocks on the Move: Beating the Market with Hedge Fund Momentum Strategies, Andreas F. Clenow 2015
Code implementation are based on several previous Quantopians post(not limited):
1. Equity Long-Short everywhere by Simon Thornington
2. EV/EBITDA value, then momentum by Johnny Wu
3. Risk Budgeting to impove performance
Will go through some backtests, and show some thoughts. Since this is my first algo, please advice if something goes wrong.