This will be an interesting backtest - Instead of price if we combine another factor like profitability (say reduce initial universe to stocks that have ROE > ROE.median). How does that impact drawdowns.
Been retail investor myself just freeing tests of survivorship bias has been a time consuming exercise. Getting fundamental historical data and then ensuring test is free of look-ahead bias will be even more time consuming.
Its tough to automate low-frequency fundamental data with high-frequency pricing data like that. Besides, a universe of Quality stocks makes sense only if your holding period is long. But if you use a stop-loss, then does Quality really matter?
And if you are not using a position-level stop-loss, then market beta itself will, from time to time, hand you a 30% drawdown.
A quick short-hand would be to just take the top 200/300 by market cap and trust "efficient markets" to do bubble up reasonably good quality stocks. This is what the actual implementation of ATH and HTP does.
This will be an interesting backtest - Instead of price if we combine another factor like profitability (say reduce initial universe to stocks that have ROE > ROE.median). How does that impact drawdowns.
Been retail investor myself just freeing tests of survivorship bias has been a time consuming exercise. Getting fundamental historical data and then ensuring test is free of look-ahead bias will be even more time consuming.
Its tough to automate low-frequency fundamental data with high-frequency pricing data like that. Besides, a universe of Quality stocks makes sense only if your holding period is long. But if you use a stop-loss, then does Quality really matter?
And if you are not using a position-level stop-loss, then market beta itself will, from time to time, hand you a 30% drawdown.
A quick short-hand would be to just take the top 200/300 by market cap and trust "efficient markets" to do bubble up reasonably good quality stocks. This is what the actual implementation of ATH and HTP does.
Seems to be a variant on the short term SMA as compared to the 200 DMA.
In this case specifically using “peak price” and returns so a more refined trend tracking filter.
Yup. Need to see how the after-cost metrics pan out once it goes through a cycle. The paper doesn't discuss markets other than US. And US is unique...