Furthermore a thought through mechanical strategy applied without flexing, will in most instances beat a mechanical strategy that is then flexed by humans for their expertise. This ties into the seminal, (but imho – rather dull book) “Thinking fast and slow” by Daniel Kahneman. It would seem that most of us let our biases allow us to reduce the overall effectiveness of a mechanical strategy. Though it could just be that the urge to change the mechanically derived list is less a bias than a reflection of the sample population needing to validate their roles by making a change.
Mr O’Shaughnessey in his book does look at the overall effectiveness of a number of screening factors in creating an above market return and reviews these both in terms of individual effectiveness but also in a number of cases as a multi factor screen. My understanding of his book is that certain individual screens can be very effective, but the best results can come from a combination of the best factors, a multi factor screen.
Mr O’Shaughnessey lays out his best screen as being a combination of both Value factors and Trend/Momentum factors.
Specifically he sets out the following value parameters
-Price to book
-Price to earnings
-Price to sales
-EBITDA/EV
-Price to cashflow
-Shareholder yield
And combines this with the best 6 month price appreciation.
To implement this screen you have to rank all the shares in your population against the value parameters individually and then from the top decile you select the ones with the best 6 month momentum.
Given that some of these shares are lowly valued for good reason it is recommended that between 25- 50 different shares are bought each time.
It should also be noted that the best returns normally come from smaller company sets, though with potentially more volatility in the holding period.
I was interested in trying in part to replicate this screen but wanted to achieve certain points.
1 – I wanted this screen to be genuinely mechanical. The data had to present easily to me without me going through self-calculation, which I felt might lead to me intruding a bias.
2- The screen had to be fundamentally simple for me to operate.
3 – I wanted to see if I could live with the decision to buy, step back and not understood what I had bought and to just sell it when a date rather than price arrived. In short could I live with potentially buying crud and letting the market take it.
Not selling on price is a reasonably big deal for me as my standard approach is largely Value based and inherent within this is a sale price for everything.
To meet my criteria I used a screen that I could easily create in Stockopedia, one of the information services I subscribe too;
My screen has the following factors
EV/EBITDA . I have set this as a minimum of 0.1 and a maximum of 10
Price to Turnover. I have set this at a maximum of 4
In the version of “What Works on Wall Street” that I have the best single factor is EV/EBITDA with strong consideration being given to Price to Sales/Turnover. Historically P/S was a better factor now superseded. To keep my screen simple these are the only 2 Value factors I am using, and instead of applying a decile based approach I am using them as filters by applying levels as above.
I have used the Relative to Index (6 months) approach and I have set this at a minimum of 5% to apply a level of Momentum requirement to my holdings. This means that the filter requires shares to be beating the main index by 5% over the last 6 months so there is some relative momentum there.
I have also used 6 months as a trading period for these holdings. 6 months is set as this is the time utilised by O’Shaughnessey as the momentum period. It should be noted that the main holding period discussed in the book is 5 years. But in my case by the time I have finished a 5 year experiment I will have forgotten that I started.
Market Cap is set at a minimum of £10 million. (I did not do this initially but found that some tiny companies got into the selection and were not tradeable in £2,000 amounts, and £2,000 was my initially intended trading size).
I then run this screen on all shares listed on the LSE and based on this sort those that pass the filter by EV/EBITDA. I then buy the 5 lowest EV/EBITDA shares provided I can buy the shares at the £ value I desire in one lot. I put in a market cap filter when I started this in January 2017 as from my initial screen I found a number of buys that I could not get in a sensible amount. If I cannot buy the share in a sensible (by my definition) lot I move on to the next one.
One of the more difficult parts of this for me was that of my initial 5 buys I had already looked at 2. One was interesting for further research THAL, the other I had rejected LAKE so buying and holding against my judgement was an issue. As was holding for the prescribed 6 months even when information on the companies was in the public domain and based on this I would normally have dumped a holding. Also difficult for me were shares that routinely could move, both up and down by double digits, sometimes in the same day.
Given I am only buying 5 holdings rather than 25+ this is another parameter to consider. But I am restricting this as either I have to commit a serious level of my money to the experiment £50k+, (25 holdings at £2k each) or I am going to be killing the results with dealing costs.
My first 6 month period ran from 04/01/17 to 03/06/17. Results are as follows
Cadogan Petroleum 20,000 shares for £1961.65 sold £1,590.05 Loss £371.6
Cellcast 70,000 shares for £2192.45 Sold £2,438.05. Profit £245.6
Lakehouse 6,000 shares for £2,022.19 Sold £2,454.05. Profit £431.86
Thalassa 4,000 for £2,087.95 Sold £2,841.38. Profit £753.43
Urals Energy 3,000 shares for £2,201.95 Sold £3,438.35. Profit £1,236.4
(These figures reflect all transaction cost).
So for an initial investment of £10,466.19, I have returned £2,295.56. A return of about 22% in 6 months. But with high volatility.
I am concerned by both the volatility in the portfolio, and the individual spread in the results. Over 50% of the gain was from one company. That said I have decided to do another 6 month trial. I have decided to add some additional funds and rather than use these to add more holdings I have placed larger positions. I am also considering whether I should hold more shares by applying the screen on a more regular basis. For instance doing a screen every 3 months. This would still be for a 6 month run but at any point in time I would have 2 sets of 5. If I get really happy with the screen it could be run monthly which gets me 6 sets of 5, which sits within the 25-50 holdings recommended in the book.
Details of the shares I bought are on the transactions page.