On 26 March I sold Telford shares which Stockopedia (SP) rated as a Value Trap at the time at 263p.
Today they are now rated as a Super Stock with a Stock Rank of 93 and the price is 351p.
As the results for the year to 31.3.19 were pretty much in line with SP's forecast how sound are either of SP's assessments?
Maybe the thing to have done would have been to set some alarms to tell you when the style changed or any of the factors improved??
Carey
As the results for the year to 31.3.19 were pretty much in line with SP's forecast how sound are either of SP's assessments?
Stockopedia doesn't give forecasts or assessments. Stockranks, from which Stock Styles are derived, are statistical, they don't predict anything about specific stocks. They can either be used to blindly select a portfolio or as an initial filter before doing more research. What they should never be used for is the sole basis for share selection.
More generally, Stockranks are generated based on the factors which academic research has shown are predictive of performance, so they're not mediated by human judgement. However, we don't know the secret sauce of the algorithms that Stockopedia uses to generate the rankings, so there's an element of trust involved in using them. Still, by perusing the Stock Reports from before and after a change in rank or style it's generally possible to figure out why something has changed.
The big change in Telford Homes (LON:TEF) was in the Quality and Momentum Ranks - so presumably recent updates have affected those. If you want to understand what happened then you probably need to dig into whatever changed to impact those ranks.
timarr
To my mind the StockRanks are a statistical description of probability so you can use them in one of two ways :
1. You can use them "as is " across a diversified portfolio (for example Ed's NAPs portfolio). Some of the 'calls' will prove to be right and some will prove to be wrong, but on average you should be ahead of the game (over time).
or
2. You can use them as a reference point for further research on a given stock.
As you say in late March Telford Homes (LON:TEF) was regarded as a "value trap" (I have used the archived report from 7th April as a reference point.)
A comparision between then and now looks like this
| 7-Apr | 18-Jul | |
| SR | 54 | 95 |
| QR | 37 | 85 |
| SR | 96 | 92 |
| SR | 28 | 60 |
It seems to me that the former low momentum rank was clearly deserved. The share price was in a clear downtrend and the earnings forecasts were in retreat.
Even now the momentum rank is only a modest 60, the share price momentum has certainly improved, the earnings momentum probably less so. I'm not sure of the precise cut-offs for the "styles" but I think TEF is very close to being contrarian rather than super-stock.
It is somewhat harder to track back on the reasons for the change in Quality Rank - from 37 to 85!
One factor must be that the Piotroski F-Score has gone from a sub par '5' to a strong '7' (or is it 8?)
I noted on another thread recently that the QualityRank and more particularly the F-Score is much more volatile than I would expect being heavily reliant on one year numbers.
So for me if I am going to use the SR as a guide to further research on a single stock it is absolutely the Quality Rank on which I would want to kick the tyres.
I actually have a number of stocks where I believe the "true" QualityRank should be significantly higher than that reported. As a statement of statistical probability the QR is still probably "right" but in terms of the specifics I am relatively confident that I am "right".
So the biggest question I guess in terms of the 'transformation' of Telford Homes (LON:TEF) is whether a QR of 37 or 85 best describes the business. (I have no particular view having not looked in any detail, although I would note that perhaps Housebuilders are near a cyclical peak and generically the fast moving QR might be overly complimentary).
To be honest, the fact CBRE have made an offer for them may have had some impact too, I don't think Stock Ranks are yet capable of predicting bids :)
timarr
Here's how the stock ranks evolved recently for Telford Homes (LON:TEF)
So yes, both Q and M fell under 50 for a while (which would make it a Value Trap) and then bounced back with a vengeance.
The stock ranks themselves are not faulty in the sense that they are simply a weighted average of the ranked values of certain metrics.
However it is worth remembering that high Stock Rank has a win:lose ratio of about 60:40 and low stock rank has a win:lose ratio of 40:60. It is not a particularly strong effect and at times the advantage can disappear altogether (as it did in late 2018).
To be honest, the fact CBRE have made an offer for them may have had some impact too, I don't think Stock Ranks are yet capable of predicting bids :)Yeah go for the low hanging fruit timarr - why not? ;-)
timarr
That passes for my "DOH" moment of not having seen that pachyderm in the booth.
In fairness though whilst that no doubt has contributed to momentum, the big shift in recorded quality is still of interest.
These Stocko ranks are a mechanical calculation.
On that basis, we should treat them purely as a statistic indicator and not as an individual stock picking tool.
Because on average they have some predictive basis for a group of stocks with a given score, but for any individual stock not so. They require a large sample to work. The process would be to buy 50-100 stocks of a certain score and realise the average result. Any individual stock's score is too prone to a sample of one error.
Also, have they been calibrated across different market conditions? bull, bear, flat ? You would need to know a lot more about these scores before you incorporated them into actual investment decisions with real money.
Appreciate the momentum idea has a strong basis in research, although a moving average likely gives you the same result. The big quant funds do this all day long. Also there are an endless range of EFTs now with specific styles you can invest it.
Myself, I wouldn't build a portfolio around scores, especially if I don't know how the exact stats behind them, otherwise, when the stocks drop, how do I know if it's a natural pull back or the model's wrong? At least if I've formed my own hypothesis for buying a stock, I can challenge it if the conditions change.
Hi Gromley
Yes, I didn't spot it either :)
The bid has clearly contributed to Momentum, even though the premium was derisory. The changes to Quality reflect the full year results at the end of May. In general I think it's a useful reminder of how mechanical the rankings are. Sticking labels like "Value Trap" or "High Flyer" on them obscures the underlying machinery.
In general using ranks and styles is fine if you know what they mean, but otherwise they can be quite misleading ...
timarr
I'm surprised that the quality rank can move from 35 to 85 in what looks like a couple of weeks. Personally I think true quality ought to be reasonably enduring. On that basis, I would expect the algorithm to use several years data for this measure & therefore be less volatile.
Have not looked into this any further, since I tend to only briefly look at the ranks, before moving onto other measures.
The subject of rapid changes to the QR did come up in another recent thread here .
Whilst I too am slightly uneasy about the rapdity of change and the (over) reliance on one year data in many of the QR components, it is worth reading what Ed Croft had to say on the subject.
Yes, I do understand this. The "Franchise Factor" - which is the more "Buffett-esque" part of the Quality Rank is perhaps 50-60% of the overall weighting. The fundamental trend component (F-Score) is 30-40% which is more volatile year to year. There are many who just want access to the Franchise Factor Rank - even though it's actually less predictive of returns - it's perhaps a better proxy for finding long term buy and hold compounders.(my bolding)
So the QR is a composite of the more enduring measures that conceptually we'd perhaps associate with the term and the more 'flighty' measures (particularly F-Score). In fact it is these latter measures that are more predictive of returns, I presume this means over a 6--12 month horizon that the Stocko data usually tests.
I would speculate therefore that over longer periods the more enduring measures possibly give better performance (although it is that much harder to test for Longer Term Buy and Hold returns). Presumably also these more stable measures could be expected to provide better downside protection.
Future developments potentially will give a greater ability to build ones own weighted measure, but in the meantime if you want to get a feel for the more "traditional" measure of quality you can separately screen for it.
Looking at Telford's data, as shown by Stockopedia, suggests there is some subjective factor being used for calculating the Quality Rank which is NOT supported by such data.
Can someone explain what I have missed to come to this conclusion if they disagree with it?
Is there a subjective component in the QR rank? I don't think there is.
I don't have data for all the metrics used in the Quality Rank calculation so I've substituted for some of them (Opn Mgn for Net Mgn and ROCE 5y average for ROCE LT average) in the plot below. I've also normalised the changes of all these metrics into a rank between 0-100 which is how Stockopedia would see them before it averages them for the Quality Rank. (Although Beneish gets ranked high -> low so Stocko would subtract the value here from 100. It gets confusing!)
It looks like the big swing on the Piotroski F score is the main cause of the large change in QR. I believe that this metric has a higher weight in the average than other metrics so it will have a disproportionate effect too. FCF LT/Assets also rose but to such a great extent.
Unfortunately to understand why Piotroski F has changed you would need to see the historical values of its constituents - which I don't have!
Also why was Quality Rank much lower before April when Piotroski F was also high? Not entirely sure! It would be really useful if Stocko themselves provided the tools to drill down to answer questions like this.
Looking at Telford's data, as shown by Stockopedia, suggests there is some subjective factor being used for calculating the Quality Rank which is NOT supported by such data.Taking a slightly different tack to Nick's answer, I would say that if you want us to challenge your view that the QR might be subjective it would be helpful if you suggest why you think that (I don't mean that to be an aggressive comment, btw although I can see why it could be interpreted as such).
Can someone explain what I have missed to come to this conclusion if they disagree with it?
The current contributors to the QR are :
Stocko doesn't actually make it easy to find that data currently, but to get there if you call up the stock report then :
1. click on "Homebuilding & Construction Supplies" in the second menu bar.
2. Find Telford in the list
3. Click on the QR number.
You will then see the view presented above - you can then click on the individual elements to see the calculation (Actually for F,M & Z scores you need to click on the links in the main stock report)
I am very confident that the QR is NOT in anyway subjective. However, whether it is "right" probably is a more subjective subject.
