Momentum Application To Sector Rotation Strategies

Sector rotation strategies consider that economy runs in cycles while different sectors perform differently in the different part of the cycle. Fidelity’s article briefs this idea pretty well.

According to GICS classification on sectors, Chinese market has the following sector indexes, which covers 80% market cap in Chinese stock market with companies of large/mid/small cap sizes. Some of the indexes have missing market data or static definition from the back test platform, in which case I would use replacement sector indexes.

1. #000928.XSHG    中证能源    000032.XSHG as replacement.
2. #000929.XSHG    中证材料    000033.XSHG as replacement.
3. #000930.XSHG    中证工业    000034.XSHG as replacement.
4. #000931.XSHG    中证可选    000035.XSHG as replacement.
5. #000932.XSHG    中证消费    2009-07-03  2009-07-03  ZZXF
6. #000933.XSHG    中证医药    2009-07-03  2009-07-03  ZZYY
7. #000934.XSHG    中证金融    2009-07-03  2009-07-03  ZZJR
8. #000935.XSHG    中证信息    2009-07-03  2009-07-03  ZZXX
9. #000936.XSHG    中证电信    2009-07-03  2009-07-03  ZZDX
10 #000937.XSHG    中证公用    2009-07-03  2009-07-03  ZZGY

Trend Filters
Trend filters will apply to each sector indexes with the following actions:
1. Rank each index with momentum indicators and we only operator on top half indexes. (E.g. use WRSI to measure the momentum; filter out whose value lower than 55; trim down the list to 5.)
2. Index with strong momentum requires to check general filters, such as:
2.a. MA(10)>MA(120)
2.b. ichimoku is above trend; and ichimoku is travelling upward.
2.c. balancer>baseline, and baseline has good strength (>55).
2.d. price is above baseline.
3. Index in BULL state will have full risk allocation.
4. Index in BEAR state will only sell stocks while buying is forbidden.
5. Max number of stocks is 12. (Or 10?)

Stock Filters
1. Build stock universe for each index.
2. Rank stocks with momentum indicator and take top half.
3. Trim list to SIZE*5. (SIZE is the number of stocks to build positions, e.g. 2).
4. Ichimoku>Trend, and Ichimoku is travelling up, with price above Ichimoku.
5. Baseline travels up, with strength>50.

Stock Signals – Bull
1. EMA(12) crosses above EMA(26), while strength>50
2. Or, price crosses above EMA(26)
3. Or, RSI impulse occurs.

Stock Signals – Bear
1. EMA(12) is below EMA(26) with strength<50.
2. Price closes below Ichimoku.

Basically sector indexes who are in BULL shape will allow to sell stocks, and then buy stocks with signals, to have fully allocated risks.

Sector indexes who are in BEAR shape will only allow sell for cash, and disallow any buying, to reduce risks.

Risk Management
1. Stock will get liquidated on next open if it closes below Ichimoku.
2. Stock drops 20% will get liquidated.
3. Stock will be allocated risks based on it’s risk level (e.g. ATR), so that each component has same risk impact to the portfolio.

ichimoku=ema(60), trend=ema(120), balancer=ema(12), baseline=ema(26)
indextrend
1/ wstrength>55 and wbaseline trend up
2/ sma(10)>sma(120)
3/ ichimoku>trend, and ichimoku moves up
4/ balancer>baseline and strength>55
5/ price>baseline

stockscreener(index,stocklist)
ranks=rank_table(stock)
filter for wsharpe>0 only, from which only score>median is picked.
filter for price>sma(120) only as final check.
take top10 from the list.

stockranks(stock)
histd=300d of ohlc
histw=resample(histd)
histw(25+8)
logreturn=pct_change
mean=logreturn.mean()
sigma=logreturn.std()
wsharpe=mean/sigma


stockfilters
ichimoku: ichimoku>trend; ichimoku moves up; price above ichimoku; weeklystrength>55
baseline: baseline moves up; strength>50
wrsi	:wstrength>50; wbaseline.

stocksignals(stock)

strength>50, macd crosses above 0.

stockseller:
1. gracefully: balancer<baseline and strength<50
2. stoploss: price<ichimoku

Performance
Return: During 2014 to 2020 (7 years), the rotation strategy has 198% (the composite index HS300 has 123% for the same period).
ROR: 17%
Sharpe: 0.95
Sortino: 1.2
MDD: 27%

Reference:
1. Sector Rotation: https://school.stockcharts.com/doku.php?id=market_analysis:sector_rotation_analysis
2. 美林时钟: https://www.zhihu.com/question/23117153
3. Merrill Lynch Investment Cycle: https://www.stat.berkeley.edu/~aldous/Research/Ugrad/Tiantian_first_draft.pdf