A letter to my panic stricken self

8th Aug '18 by Muggles

10 comments 3467 reads

Introduction

I’m currently reading The Little Book of Behavioural Investing, and one of the key lessons is having a plan and sticking with it when emotions want to take over. So, with that in mind, here is a letter to my panic stricken self when the market goes bear.

In writing this, I discovered a flaw in my rules. My current 52 week high rule might do well to get me out when the market is doing down, but it will keep me out when it starts to recover. This is something I’ll need to formally review next month. For the purposes of this, however, I’m assuming a ‘temporary’ rule based on the market bottoming out and starting to recover.

Letter

Dear me

Sit down, shut up, and grab a coffee. Yes, you can afford to still spend money. No, you are not a terrible investor (ok, you’re not good, but you’ll be ok). Now take a deep breath, and follow the plan.

Rule 1: Don’t change the plan
You came up with this plan for a reason: when it comes to picking stocks, judging bottoms, calling tops, and making decisions under pressure, you are at best, taking an educated guess, or more likely, making exactly the wrong call. Don’t change the plan. No excuses

Rule 2: Follow your rules
You’ve got two types of buy and sell rules. Ones based on stock ranks, and ones based on price. The StockRank ones are ranks, so there will always be stocks ranked with great momentum ranks, even when the market is going down the drain. However, your price markers will be sensitive to a spiralling market. If everything is off their highs, and there is nothing coming up, that means there is nothing to buy, so hold the cash, and review weekly. The only exception to this is forgoing shares that are close to being excluded because of their price rules (i.e going down and close to % off 52 week high limit) – if they are going down and within 5% of this rule, you can forgo that share.

(Note to self - you already sometimes give yourself 5% desecration with this rule for shares heading down. Time to just make this part of your usual rules?)

Rule 3: Use the built in flexibility
You’ve got permission within the…

Unlock with Email

Unlock with Google

Already have an account?
Login here

Nick Ray

8th Aug '18

Your comments about the 52-week-high rule not working when the market is down is interesting. There is a lot to be said for using relative values rather than absolute values - you just want your stocks to be performing better than their peers.

I explored this a bit in a recent post where I looked at the curious fact that the median is very often the "best" value to use for the cutoff over a wide range of metrics:

https://www.stockopedia.com/co...

So I thought I would have a quick look at the way that the median 25-week high behaves as the market moves up and down. Apologies for the ugliness of the following graph which I put together as a very quick look at what happens:

The green line is a normalised version of how the AIM100 index behaved over the last 2.5 years, and below it is the median value for the 52-week-high for a basket of stocks. I have added the vertical blue lines as a guides for the eye to see how the dips align with changes in the median 52-week high metric.

On this quick look I would say that there is a prima facie case that this approach would indeed adjust the metric quite well to suit market conditions.

iwright7

9th Aug '18

Nick,

Another interesting observation, which also fits with my propensity to buy High QM shares near their 52 week high.

As a counter to this I bookmarked a paper that seems to show that combining earnings momentum with far distance to 52 week high (e.g. % vs. 52 w High less than or equal to -10) produced outstanding results over a 10 year period from 2005 -2015. Perhaps you could take a look at it and bring your statistical powers to bear?

Thanks, Ian

https://ideas.repec.org/p/zbw/iwqwdp/122015.html

gus 1065

Hi Jm.

Some interesting thoughts - always good to have a plan ahead of time. Various commentators advocate this as a way of trying to take the emotion out of dealing with the stress of a market correction.

On your Rule 5 about the merits or otherwise of looking at Value shares, my inuition would suggest that in a correction these may perform relatively well compared to highly rated Shares with a high M or G score. (Simple idea being that if your already in the basement, there’s less far to fall than if you’re already on the top floor). Conversely, one might hope high Q shares would have good defensive properties.

I had a quick look through the Stocko archives to see if there was anything relevant on the point (“in a crash, which dominant stock factor/stock type suffers best/worst”) but couldn’t find anything. Wondered if anyone else recalled reading something on here.

Gus.

Ian

I had a quick look at the paper and to cut to the chase I couldn't reproduce a benefit. Combining two factors and looking to at their joint probability distribution requires a contingency table. For the median return I got something which looks like this (bigger numbers are better):

That clearly shows you want above median earnings momentum and above median "% vs 52w High" but that for very high earnings momentum the exact value of the "% vs 52w High" does not matter very much.

However if I look at the 30% quantile (i.e. trying to get a return above the 30% quantile) the table looks like this:

This shows that for protection against bad outcomes it is worth having good earnings momentum and above-median "% vs 52w High". I won't bother you with the 70% quantile plot but that shows that for the better-than-median outcomes the importance of "% vs 52w High" greatly diminishes - i.e. you can have high or low values and it won't affect the outcome very much.

So what is the effect they are seeing? My quick instinct is that by choosing below-median "% vs 52w High" what they are doing is increasing the volatility of the outcomes - so there are more extreme outcomes (good and bad.) - which means they are exploiting tail effects. For me, this is somewhere I would not want to be. On the other hand, they looked at a range of "pain" and drawdown measures and claimed that their approach caused less drawdown - which is not what you would expect with higher volatility. So there is room for doubt about whether my instinct is right here.

At which point it partly comes down to personal preference. I have gradually come to dislike methods which increase volatility of outcomes because they tend to be more unstable and prone to data mining effects - and they are quite stressful to implement too. But other people will prefer them. It boils down to the question:

Which would you rather have?:

1. the highest possible mean return (and accept that around that mean you might have big swings from year to year) -or-

2. the highest probability of a return exceeding some baseline value (such as the 30% quantile).

The first option will appeal to people who go for concentrated portfolios; choose conviction stocks (alpha); use GMM (Geometric mean maximisation) etc.

The second option will appeal to those who prefer to diversify; use factor-based (beta) selection; and maximise the Sharpe Ratio rather than the Geometric Mean.

Gromley

On your Rule 5 about the merits or otherwise of looking at Value shares, my inuition would suggest that in a correction these may perform relatively well compared to highly rated Shares with a high M or G score. (Simple idea being that if your already in the basement, there’s less far to fall than if you’re already on the top floor).

That's a tempting argument Gus, but one that in my experience doesn't seem to hold true (I should stress that is a personal observation, not the result of any scientific study).

Often-times I think "unloved" value shares can all to easily become "thoroughly detested". In a parallel observation of the course of a number of years I noticed that companies periodically recurred as value opportunities causing me to muse that some companies have characteristics that make them prone to being unloved. I haven't made any kind of study of what those characteristics might be so it's not in itself a particularly useful observation except as a bit of a reminder that apparently irrational under-valuation can become more-so in the near term.

I had a quick look through the Stocko archives to see if there was anything relevant on the point (“in a crash, which dominant stock factor/stock type suffers best/worst”) but couldn’t find anything. Wondered if anyone else recalled reading something on here.

In the comments section (comment 6) of Ed Crofts five year StockRanks review Howard Marks posted an interesting quant view from JP Morgan. I overlaid the StockRank styles on in post 12 (although there is a typo the top right hand box should be labelled "turnaround") .

If this study is to be believed, then in the early part of a downturn you want to be in "High Flyer" [QM] stocks (presumably quality being defensive and momentum being to some degree persistent).

In the latter stages of a downturn, value begins to work and the target would be "Superstocks" [QVM].

In the early stage recovery Value on it is own comes to the fore "Value Traps being favoured. Interestingly this is the only market phase where a "losing style" seems to be preferred (although as I noted if you successfully get in at this phase, you'll probably winning whatever you pick).

Finally in the late stages of growth "Turnarounds" VM are preferred. Arguably perhaps that means keeping those value traps that have performed, but remain in value territory and ditching those where the value has been "outed" together with those that really were "traps".

One difficult here though is clearly to understand whether we are in the late stage of expansion [VM] or the early stage of slowdown [QM]. Either way Momentum still seems to be important. Maybe though "on the way down" a high momentum rank would be a better filter that proximity to 52 week high?

I should just point out that the JPM study specifically refers to the economic cycle rather than the stock market cycle, it's generally taken to be the case that the stock market cycle precedes the economic cycle. I'm not clear if that is sufficient to suggest that the quant model should be offset by a quartile if considering against the market phase?

Nick - What an excellent piece of work - In relation to my investing preference it is definitely less stress, lower volatility and a higher probability winning even it it reduces the annual gain.

Many thanks again. Ian

Muggles

11th Aug '18

Hi Nick

Thanks very much for your posts. I went back and read your recent blog and it is fascinating. Your point about setting cut offs too high in your blog really got me thinking. I've tried to construct this with the view that, since I'm not an expert, keeping things simple with a few clear rules which I understand and have a clear rationale is best. However, you've made me wonder whether rather than thinking "this didn't quite work - I need to raise my thresholds" might end up removing more winners than losers from the pool.

I've also spent some time getting my head around your work on the 52 week high metric. Originally, I set it as a proxy for "sell if a share has lost more than x% of it's maximum held value", but in it's by form, I was just pointing out to myself it had the simple result of taking me out of the market if everything is down 10/15/20%. I'd love to know how you got the data on median 52 week high, and if it is something I can get from Stockopedia, as this sounds like a potentially useful addition that would also work when the market starts to recover but is still well off where it was before the slump.

Thanks again!

Hi Gus

I had originally had the same thought - surely if it is already really cheep, it can't that much cheeper. Indeed, my first draft had me doing exactly that. Then I went and had a look at how different combinations of stockranks had performed since they started, and to my surprise found Value suffered the worst drops when the market dropped. The flip side was they had great gains after. However, during the down turn, QM seemed to be one of the best styles to be in, so I decided to just say to keep my rules as is. Not particularly indepth evidence I know, but until I find better, it seems to point in this direction. Thanks again for your thoughts!

Hi Gromley

Thanks for your thoughts, and particularly the bit about JP Morgan. I'd anticipated that as the market changed, I'd need to shift my weightings between QV and M, but I didn't know how. I'll be book marking this as something to guide my thinking when I'm revisiting how I'm weighting my buy rules. Cheers!

Mike888

In general, you would assume that value stocks take less of a beating than expensive stocks, I believe this to be true.

The key to understanding this is to determine whether a value stock is quite rightly cheap or underpriced for no logical reason. If it is the former the stock will fall dramatically, quite simply because if it is performing poorly in a bull market there is no hope for it in a bear market. Conversely, if a value stock is underpriced for reasons of a lack of coverage\visibility, historic failings now resolved etc this stock may well drop in sympathy with the market, but it would probably not drop as dramatically and would, in the fullness of time, probably be a good one to purchase in a recovery.

The skill is knowing the difference between value and cheap rubbish, metrics and ratio's rarely provide that depth of insight.

A letter to my panic stricken self

Unlock the rest of this article with a 14 day trial

About the Author

Muggles

10 comments

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

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