SIF Folio: Why nine months could be the optimum holding period
Before I get started, I’d just like to point you towards Jack Brumby’s piece from last week, which covers three potential recovery plays. Jack also included two charts which contrast Google search levels for “corona” with the performance of the FTSE All-Share index this year.
In short, since hitting peak panic in mid-March, interest in the pandemic appears to have fallen and investors have pumped cash back into the market. In the real world, we’re seeing a growing number of listed companies finding ways to get working again. That’s also the feeling in the area where I live - more local businesses are finding ways to reopen.
This week I’ve decided to step back and look at one important aspect of the rules which govern my SIF Folio: my holding period.
The story so far
When I launched SIF in April 2016, my minimum holding period was six months. Stocks could only stay in the portfolio for longer if they continued to satisfy all of my screening criteria.
In 2017, I decided to increase my standard holding period to nine months. Although I didn’t have much statistical support for this change, two factors made me think it might be a good idea:
Data refresh: Nine months should be long enough to ensure that all companies publish at least one set of accounts during my holding period. This should ensure that the quantitative data for each firm will be ‘refreshed’ at least once during my holding period. This provides a greater opportunity for strong or improving performance to be reflected in a stock’s valuation. (Of course, the opposite is also true, as bad figures can trigger a sell off.)
Relative strength: My screen uses one-year relative strength (RS1y) as a measure of momentum. This looks for shares that have outperformed the market over the last 12 months. Stockopedia’s research indicates that this measure is most predictive over a 6-24 month timeframe. So my six month holding period seemed like it might be a little too short.
Since 2017, I’ve left my holding period unchanged at nine months. Stocks can stay in the portfolio for longer if they continue to pass all of my screening tests, but I’ve found this uncommon in practice.
Backtesting: is 9 months too long?
One reason why I rarely make changes to my screen is that I have no way of backtesting them. However, I’ve now built up four years of trading records for SIF. This represents 80 completed trades - where I’ve bought and sold a stock. This doesn’t include current open holdings.
Although 80 is still a fairly small sample in statistical terms, it’s better than anything I’ve had before. So I’ve been using a Google spreadsheet and the GOOGLEFINANCE function to model what might have happened if I’d held stocks for different lengths of time.
I used the original purchase date and share price for each stock SIF has held since April 2016 to model different holding periods. Here are the results:
Holding period (months) | Average return* | Annualised return* |
3 | 5.30% | 21.51% |
4 | 5.50% | 16.74% |
5 | 6.75% | 16.43% |
6 | 7.50% | 15.21% |
7 | 8.40% | 14.60% |
8 | 8.70% | 13.23% |
9 | 11.30% | 15.28% |
10 | 9.70% | 11.80% |
11 | 10.30% | 11.39% |
12 | 9.80% | 9.80% |
13 | 9.30% | 8.70% |
14 | 8.95% | 7.78% |
15 | 9.65% | 7.83% |
*Including estimated trading costs, excluding dividends. Stocks that were subject to takeover bids/mergers are also excluded.
Looking at the average returns column appears to vindicate my decision to hold stocks for nine months. But when you annualise these returns to compare figures for different timeframes, a different picture emerges.
On the face of it, holding shares for just three months would have generated the greatest returns for SIF.
Changing from six to nine months does not seem likely to have generated any benefit, with near-identical annualised returns of 15%.
Holding stocks for longer than nine months would apparently have reduced my returns.
Dividends: My records show that dividends have added about 2.5% per year to my returns. The majority of these have been earned across a nine month holding period, increasing the average total return for this period to just under 18%.
I think it’s fair to say that with a three month holding period, I would have missed most of these payouts. So I’d suggest that dividends have proved a low-risk way to improve returns without more intensive trading.
Performance vs market
The problem with testing a relatively small sample of stocks over a short period of time is that these results will have been influenced by market conditions.
If we consider the portfolio’s performance since its inception, it’s clear that some periods delivered strong gains. These were often linked to wider market movements:
To get a more accurate idea of the impact of different holding periods, we need to try and strip out wider market movements.
I extended my Google spreadsheet to compare the gain or loss for each stock with the movement of the FTSE All-Share index over the same period. I used the ASX because this is the official benchmark for the SIF folio.
Let’s take a look at how the numbers shaped up:
Holding period (months) | Average return vs index | Annualised return vs index |
3 | 3.00% | 12.17% |
4 | 2.70% | 8.21% |
5 | 3.42% | 8.32% |
6 | 3.74% | 7.59% |
7 | 4.50% | 7.82% |
8 | 4.55% | 6.92% |
9 | 6.83% | 9.23% |
10 | 6.17% | 7.51% |
11 | 7.00% | 7.74% |
12 | 5.70% | 5.70% |
13 | 5.91% | 5.53% |
14 | 4.60% | 4.00% |
15 | 5.23% | 4.24% |
The picture looks similar to start with. Three month holding periods show the greatest annualised gains versus the index. The numbers then decline, except that the average market outperformance over nine months is notably higher than for six months (or any other period except three months).
Why I’m staying with nine months
Disclaimer: I want to point out that these projected returns are only relevant (if at all) in the context of the ruleset I use to select stocks for SIF (you can see the SIF screen here). Please DYOR.
Having said that, I think these numbers suggest two tentative conclusions for SIF:
A holding period of three months might improve returns;
Nine months is the best holding period, except three months.
I could use this result to justify changing to a three month holding period, but I’m not going to. At least not right now. I can see too many problems with such regular trading:
Higher transaction costs
Loss of dividends
Difficulty finding enough stocks to hold in the portfolio and maintain diversification
Greater volatility and exposure to short-term market shocks
Too far removed from the fundamental-led approach I am targeting
I’m going to leave SIF’s nine month holding period unchanged.
But I’d be very interested to know how subscribers who follow rules-based systems handle the question of when to sell. Do you have quantitative rules for selling? Or fixed timeframes, or some other approach?
Disclaimer
This is not financial advice. Our content is intended to be used and must be used for information and education purposes only. Please read our disclaimer and terms and conditions to understand our obligations.
9 comments
it's a bit easier to see your table of returns as a plot:
I can't help but feel that the plots are saying "shorter time frame is better" but you are rejecting it because it is not what you want to do. On the annualised plots there is a bump at 9 months but it not the prevailing trend.
Now to be honest I understand your view. You set out some good reasons why short holding periods are not a good idea. So what might be happening here? My suspicion is that your screen is not finding stocks which are actually in tune with the principles you set out.
When I look at the mini-graphs of the stocks the screen is currently choosing over 1 year, 3 years and 10 years I can't really see anything particularly exciting or special. However over 6 months and 1 month the stocks look perkier (A bit harder to tell just at the moment of course.)
I strongly suspect that your rules are being influenced by short-term effects (1-6 months) but your natural style is to go for longer holding periods.
I'm going to indulge in a little speculation and suggest that you would be more at home with stocks with good quality metrics over the 3-5 year time-frame rather than the emphasis on value over the current year.
Hi Roland,
Interesting article.
For me, a 3 month holding period would lead to far too much administration and time spent researching potential purchases and sales. And, as mentioned previously by others, the increase in transaction costs over a 9 month holding period would be a huge handicap on your portfolio's returns. You would be transferring a large part of your hard earned savings to brokers (via commission), market makers (via bid-offer spreads) and the government (via 0.5% stamp duty on each purchase).
Hi Roland , thank you for your article.
What do you use to decide your entry point for a trade?
Hi win13,
I select stocks to buy for SIF using my SIF stock screen:
https://www.stockopedia.com/sc...
The screen uses a set of fundamental rules and the results are ordered by StockRank. If a stock satisfies my diversification requirements and passes my sanity checks, then I add it to the portfolio (maximum one new stocks per week).
Regards,
Roland
Thank you .
Hi Roland. Good thought process.
Made me consider trading costs v frequency of trades. I assumed a portfolio of £100,000 consisting of 30 investments and some reasonable trading costs. I assumed that frequency of trade did not improve the average return.
Trading frequently i.e. only holding for 3 months v's infrequently i.e. 1 to 2 years could halve the final fund value in trading costs alone. Over 20 years the additional cost would be £160,000 in my simple calculations. Clearly this is because the costs of trading are removed from your fund and therefore the loss is compounded over the years.
The big assumption here is that frequency does not improve return. But it does highlight that you need to make a much better return if you want to trade frequently.
I would not want to hold stocks for less than 9 months, in my above scenario, to save trading losses.
PS. I was intrigued to see a clear peak in your returns with 9 month trading. Have you been able to verify that this is real or an aberration of your own trading patterns?
Hi Robert,
Thanks for your comment. I agree that the costs of trading more frequently would probably be a serious impediment. I'm not comfortable with the idea at all, at least not with this strategy.
The modelled results for nine month holding periods that I included in this article are very close to the actual returns generated by the portfolio over the last four years:
Average profit (inc dividends + costs): 11.9%
Average annualised return (inc dividends + costs): 16.9%
If we subtract an average dividend contribution of 2.5%, these figures are c.9.5% and 14.4%.
Obviously my long-term average returns do not yet reflect the impact of this year's market crash, as I haven't closed any positions since February. So these averages could come down a bit.
I should also reiterate that my comments are only valid (if at all) in the context of the rules I use for SIF. Everyone's circumstances and results will differ.
Regards,
Roland
If you're fully invested at all times, won't your dividend income be braodly unaffected, regardless of the holding period per individual stock?
Hi karlmarx1,
I agree that if the portfolio was in theory this might be true and dividend income would be roughly constant.
But in practice, the portfolio has rarely been fully invested. When the market crash started, SIF was 40% in cash. Even now, it's around 25%.
A second point is that I think Sod's Law would mean that with such short holding periods, I'd probably time my buys/sells so that I missed a lot of dividends.
Regards,
Roland