Consensus forecasts aren't always what they used to be

14th Apr '18 by HumourMe

17 comments 5161 reads

Edit: Nothing to see here. My misunderstanding. Please move along.

I used to take comfort from steadily rising consensus forecasts, however, there is an element of historical revision. I think visibility of how erratic they can be reduces the importance of the trend, especially if for a point in time, the trend can be subsequently revised, several months after the fact.

I spotted this because I sold in October Vitec (LON:VTC), after tightening stops as a result of a falling broker consensus. In March my historic notes and a review of why I exited suddenly didn't make sense until I reviewed the history.

Below is what happened to the trend as we roll forward one month at a time.

I'm wondering how this affects calculations based on the movement in forecasts?

The revision occurs as a result of a transition between 2017 EPS and 2018 EPS, however, this only highlights the cause and doesn't reflect the figure that was forecast at that point in time.

Unlock with Email

Unlock with Google

Already have an account?
Login here

PhilHofGrowYourDough

14th Apr '18

It appears to be the switch over from 2017 EPS estimates to 2018. The switch occurred in the last two images as you can see the label on the left changes.

HumourMe

Hi Phil; I was just editing that in(!) but I don't understand how this should affect a historical forecast. I'm clearly missing something!

If you look at a stock report to the left of the Broker Consensus Trend image you have extracted you'll see a panel called Broker Forecasts. You'll see this is split into two separate years and the figures will be different. I guess Stockopedia has to work out how to deal with the transition of that data into the panel on the right hand side. So at some stage I guess they switch from FY + 1 (2017) to FY + 2 (2018). I'm guessing that is what you are seeing?

To be clear the first 5 images in your post show the trend for the 2017 EPS estimates and the last two images show the trend for the 2018 EPS estimates. Understandably they are different.

When asking the question "what was the consensus in October?" I'm surprised that the answer changes, depending on when it is subsequently asked. I think that like other figures it should be ideally 'fixed' at a point in time.
One way around this would be to use a rolling forecast and not revise the value subsequently.
Another would be to highlight the discontinuity in the chart, or have two series, one capturing the historic estimate and the other a revised figure.

But you aren't being specific enough with your question.

Are you asking "what was the consensus for 2017 EPS estimates in October?" because it's different to the 2018 EPS estimates

You could actually see the consensus estimates for both 2017 EPS & 2018 EPS by looking at the Broker Forecasts panel on the October historical stock report, accessed via the Print button (which is a bit weird)

andrewdb

HumourMe

Does your approach work for you?
I have been struggling to find a good (or personally satisfying) exit rule.

sharw

Since subscribing to Stockopedia I have found the future estimates to be one of the most unreliable parts of the system and I always check these against other sources.

Take, for instance Pebble Beach Systems (LON:PEB) (formerly Vislink). After it ran into trouble estimates went down and then dried up. I cannot find any estimates elsewhere now - not even on Reuters from which I thought Stocko got its data. However, it shows eps for 2017 (not yet announced) as 1.3p. For 2018 it shows eps 2.1 for a bargain p/e of 1.31!

In Jan. 17 the estimate was 1.4 and in March and April 1.55 before coming down to 1.3 in May after the 2016 results had been published. It has remained at 1.3 since.

However, here is the strange phenomenon that confirms HumourMe's suspicions: from July onward the 1.4 and then 1.55 as the forecasts made from Jan - Apr last year have crept forward so that if you look at today's sheet you will see the 1.4 as being forecast in October and the 1.55 as Nov 17 to Jan 18. Bonkers.

Looking at the historical stockreports for Pebble Beach Systems (LON:PEB), I'd agree that something doesnt quite add up.
Perhaps raise a defect report?

Yep that's the nub of it. At the end of October there was a consensus figure reported. In November the October consensus remained the same. In December, October it began to nudge up.

What I thought I was seeing was the consensus in October, then the consensus in November etc; once 'closed' the historic forecast was fixed and assuming so I though it a trend in historical consensus. That isn't what is being shown.

What is shown I think, updated with the latest figures, is what the consensus would have been in October, given subsequent forecasts post October. It isn't apparent to me that this is immediately useful but I may be missing something.

Effortless Cool

It would make a lot more sense to put both years' forecasts on the same chart. It would also be useful to be able to add broker forecasts for prior years, to see whether the directors tend to underpromise and overdeliver or vice versa.

It's not what's being shown in your series of graphs because they are two sets of graphs!

2016 EPS trend & 2017 EPS trend. It seems to me that you are looking at the latest 2018 graph and saying why doesnt it tie in with the 2017 graph.

There is no one consensus, the brokers will be running mulitple sets for each financial year.

So they will have broker estimates for 2017, 18, 19, 20 and Stocko could show graphs for the changing trends in each of these financial years over time, but they only present one graph on the stockreport.

What is being shown is an estimate of the 2018 EPS of 75.6 in October 2017 as confirmed by the value in the October 2017 Stock report. You are expecting to see a 2017 EPS value there but it is a graph of 2018 EPS trends.

I think scrollable charts for each year would be good ... a bit like the scrollable Guru Screen qualified on the stockreport

Damian Cannon

The other thing to note (apart from the change in year) is that the first two graphs show that data was taken from 5 brokers - this then switches to 4 brokers for all of the other charts. This might explain why the Oct and Nov numbers for 2017 change?

Does your approach work for you?
I have been struggling to find a good (or personally satisfying) exit rule.

Some of the time.

This year, after 13 closed trades (1 still open) I am 6% down. Last year was easy but rather frenetic ... I aspire to being a position trader, following trends.

The trouble with exits is that there are so many potential rules. Technically x% off an absolute high is okay for trend followers, a lower low can also work (a variation of this is a break to the downside after a period of consolidation). I find weekly charts more useful for this assessment. Some use a close below a moving average that has previously shown support. In the opposite direction 'going parabolic' is taken by some as a sign, although I've never pulled that off. Some people exit on increased volatility, I guess I do too if it establishes a clear range.

Meeting a fundamental target also works for some and with VTC it was the downward revision of forecast, in combination with a period of consolidation that spooked me and provoked the exit.

The current markets I'm not finding easy to enter (requiring breakouts), with most positions taken been largely stopped out by increased volatility. At the moment, on a weekly basis I briefly review market direction, add and remove shares from watch lists and after reviewing charts don't place orders. From time to time I review previous entries/exits to see if with hindsight my selection criteria were on balance valid and if I could have exited 'better' using a 'new' consistent rule.

I'm thinking of non-trading as capital preservation, although fear of missing out is always there.

It is a good opportunity to revisit trading books, especially those that emphasize trading with the market :-)

It's not what's being shown in your series of graphs because they are two sets of graphs!

2016 EPS trend & 2017 EPS trend. It seems to me that you are looking at the latest 2018 graph and saying why doesnt it tie in with the 2017 graph.

I get it now. It just isn't what I thought it was. Thanks for your patience!

Consensus forecasts aren't always what they used to be

Unlock the rest of this article with a 14 day trial

About the Author

HumourMe

17 comments

Unlock the metrics for all these stocks and more...

Unlock the metrics for all these stocks and more...