Some minor system update and general system description
On Friday, May 28, I did a new screening. I did a minor change this time. Instead of using a “positive 1y Share Buyback Rate” as a KO prescreen, I added it to my list of red flag penalty counts. To better understand this, I should go back and describe my process a bit better.
First, I screen on gurufocus.com for profitable, global stocks (based on E(exNRI), opCF and FCF), which are listed on European exchanges. Then I clean and prepare the data and finally I calculate three main ranks:
1. Valuation Composite (anything x/P, first combined/ranked to correlated subclusters and then combined/ranked to the final Composite)
2. Performance Composite (3-1m, 6-1m and 12-1m Momentum combined/subranked and then paired with 60-12m Longterm Reversal)
3. Accounting Composite (equalweighted: Asset Growth, Piotrosky F-Score, Accruals, External Financing, amongst others)
These three ranks are then "basically" equal-weighted (there is a slight Value bias in case of a "draw") to generate the Buy signal.
In these 3 main composite ranks I only use factor metrics which have a "linear" return distribution, meaning that historic returns increase/decrease steadily from the highest to the lowest decile or vice versa. Unfortunately for many "proven" factors, this is not the case. Often, the lowest deciles are the main driver of the L/S spread and for high deciles, the return distribution is basically flat (see also my Factor Cheatsheet on my Twitter page).
Ranking on these factors can be misleading and kicks out many stocks unjustly. These factors are, however, not useless for long-only investors. Their nature makes them the perfect "Red Flag Screens". OSAM described a similar concept here:
https://osam.com/pdfs/research/58_Alpha-Within-Factors-Nov-2018.pdf
Depending on the return distribution of factors, which are interesting for the "Red Flag Screen", I count penealty points for a stock if it ranks in the lowest 1 or 2 deciles of given factor metrics. For each "Redflag" penalty point I reduce the final "Multifactor-Signal" by a given percentage instead of applying a KO screen.
That procedure might seem shady, but it basically is an insurance against being "too strict". In most cases, I will buy a stock with 0 red flags anyway but if at any time I “run out of" 0-red-flag stocks (e.g. due to having to skip too many sector-concentrated or illiquid alternatives), I want to have them in "reasonable range" as next best alternative. The final list looks like this (screening of May 28, yellow stocks are currently in my portfolio):
You can best compare my system to a Casting Show: The Prescreen basically covers the application procedure and first casting rounds for the applicants. Then in the main event on stage, the 3 main composite ranks build the jury and the “red flag” count is somewhat comparable to the audience (it has no major impact most of the time but if everyone in the audience in booing the candidate, the jurors will kick you out).
After the ranking, I go through the list and buy the highest ranked stock for which
I don't have already 12 stocks in the same currency/country
I don't have already 5 stocks in the same sector
I don't have already 3 stocks in the same industry
The Bid-Ask Spread doesn't exceed 1% of the stock price
The stock is liquid enough that I can buy the stock within minutes with the given position size.
I currently have 20 positions, each of them with a holding period of 6 months before checking/rebalancing and selling/buying. The rebalancing dates are spread over a 6-month cycle meaning one position is rebalanced each ~183/20=~9 days.
PLEASE NOTE:
The ranking is somewhat "optimized" for 6-month rebalancing. None of this is investment advice. The stocks in the screening are only held in a diversified portfolio and bear high risk of capital loss. THESE ARE NO STOCKS TO HOLD FOR THE LONG-TERM. These are stocks traded in a systematic strategy which has to be followed over the long-term and which will have long periods of underperformance.
Now that you know my system better, back to the recent adjustment of my prescreen:
I did kick out my "positive Buyback Yield" criteria in the prescreen and added it to the Red Flag Counter instead. This adjustment increased the number of stocks in my screen from usually 2500-3500 stocks to 5600 stocks. Main reason for the adjustment: If I wanted to check the ranking of a famous stock in my system, it often was not included in my ranking due to an often tiny negative Buyback Yield (positive Stock Issuance) of -0.05% or something like this. I know that the change also has an impact on my system on the whole, e.g. more stocks change the Momentum landscape but I still decided to do it.
As long as you stay in the same style, minor adaptions to the system are okay in my opinion, especially if it is such a minor quality-of-life update.
Kind regards,
Your Non-Prophet.