Trailing, Rolling, Annualised, Discrete?

Prism

BritishInvestor said:

Thrugelmir said:

BritishInvestor said:

Thrugelmir said:

Against the benchmark that the fund uses. If there's an appropriate one of course. 

I would suggest that very few people know how to confirm the benchmark is appropriate.

Funds normally dislose and compare to their choosen benchmark. Which is an established market index. 

Yep, but that doesn't change my point. You need to select an appropriate benchmark. For a fund that is invested in US large-cap growth/tech, MSCI World is arguably about as relevant as the wholesale price of cockles.

If you don't use the benchmark chosen by the fund manager it is difficult to know at what point you select another. Since the only way of outperforming the index is to actively move away from it, at what point is the benchmark incorrect and you need to look for another if there even is one. Quite a few of the well know active fund managers say that they don't benchmark themselves against an index because they are so different, but are basically required to pick something similar. 

I am not sure really how I would benchmark Scottish Mortgage for example with its widely different allocation to the world index plus 25% private equity. The same with most of the wealth preservation trusts. I guess they are benchmarked against each other.

Prism

To try and answer the initial question I would tend to use 10 or 5 year annualised for an initial glance but then use charting to compare during certain dips and peaks. For example on annualised figures many of the Baillie Gifford funds look amazing but using a chart to compare performance you find the real story.

JohnWinder

It's a great question which I can't give a systematic answer to, sorry.

Firstly, you couldn't do justice in analysing a fund's performance if you didn't take into account the riskiness of the fund. A fund that underperforms another might be taking less risk. Which do you want? Most folk want the biggest returns for the smallest risk, so can one really ignore risk when evaluating performance?

I think it helps to distinguish 'passive' funds which attempt to track an index, from other funds. For the former, which should be taking the risk of the market they're in, no more and no less, the best way would surely be to see how closely it tracks the index over several years, and ideally understand why it does/doesn't.

For the latter, the benchmarks some funds choose make it impossible for me to evaluate them because I can't picture their risk. It's easier if they say they'll beat the government bond index, and then load the fund with 80% government bonds and 20% large cap equity; of course it will usually beat its benchmark without much risk.

A problem with annual, rolling etc performance measures is the start/end dates; if there's a crash on January 3, then Jan 1 to Jan 1 performance will be very different from Jan 5 to Jan 5. And for rolling 5 year returns, the first year of the first 5 year period goes into the calculation only once; but the 3 year performance is part of the first, second and third 5 year rolling return, so some years can have a big impact (or none actually). Annual performance measures, like all summary information, hide a lot of detail. If you want to see some of the detail it might be better to look at a graph of £100 invested at time zero, and follow its growth/decline over subsequent years, making visible all the ups and downs as well as overall gain/loss.

cfw1994

BritishInvestor said:

cfw1994 said:

threlkeld53 said:

Could someone please explain what is the best way to analyse a fund's past performance.

For example looking at a fund's history on iweb, they usually give discrete, trailing and annualised percentages for 3, 5 and 10 years. I sometimes see the word 'rolling' also, so I wondered what is the best way between these 4 sets of data to give the best indication of any particular fund.

Well, I am perhaps very amateur at this, but I would start by checking the 1/3/5/10 year numbers.   Trustnet or similar to find the factsheets.
That's a decent starting point, before you delve into whether there are differences, etc: see how that fund is comparing with things you do perhaps know, or just other funds to compare.
The benchmark thing is a red herring, IMHO....different funds chose to use different benchmarks (although it might be reassuring to know you are in the top quartile, I guess)

The benchmark isn't a red herring. People may buy a fund because it
1. It has vastly outperformed its benchmark
2. It has great recent returns and appears on platform best buy lists
3. It has outperformed "similar" funds
whereas all of these should be reasons to dig a little further before buying. 

Like I said - might be reassuring to a few....but unless you know the benchmark in some detail and are only checking funds that submit to the identical benchmark, I still maintain it is a red herring.   
I have worked with IT benchmarks in my past: great to show certain things, but never the whole picture at all!
I think Linton nails it here:

Linton said:

I don't see there is any point in comparing a fund against a benchmark if the benchmark is not actually investible or does not truly represent the fund.  What do you do if the performance is poor? You really need to compare against other real alternatives that meet the objectives for wanting a particular fund.

When comparing performance I only glance at 1yr/3yr/5 yr/10yr because these figures can be strongly influenced by what happened over a short time period possibly recently or possibly 5-10 years ago.  So when I really want check performance of a fund against other possibilities with the intention to buy  I prefer to look at  the individual year by year values to see how they vary.  This may show up one fund being more volatile than another or a consistent pattern of outperformance.  Another worthwhile check is to see how they performed during market crashes. Sadly though the last real crash in 2008 is too long ago for much to be learnt except for sectors such as Wealth Preservation where the funds reason for existence is to have a viable long term strategy.

Having identified a significant performance difference, before one buys it is important to identify the cause.  Usually it is because the funds invest differently.  eg some high performing funds in general sectors are actually focussed towards small companies or tech.  If you want a small company or tech fund, fine. But if you don't there is no point in buying one because of apparent high performance relative to what you really want.

I would agree with the second paragraph, BUT for quick comparisons between a large number of funds, the 1/3/5/10yr numbers can give you a decent quick comparison.  I did that 15+ years back when scouring 80 funds to decide which ones should make a shortlist for deeper investigation!
Definitely drill more into what the funds are investing in, check how they compare in bad years v good ones, etc.....but use the headline numbers as a starting point!
None of us (to my knowledge!!) are investment analysts - if you are, please step in with your thoughts for the 'layman' to follow!

BritishInvestor

cfw1994 said:

BritishInvestor said:

cfw1994 said:

threlkeld53 said:

Could someone please explain what is the best way to analyse a fund's past performance.

For example looking at a fund's history on iweb, they usually give discrete, trailing and annualised percentages for 3, 5 and 10 years. I sometimes see the word 'rolling' also, so I wondered what is the best way between these 4 sets of data to give the best indication of any particular fund.

Well, I am perhaps very amateur at this, but I would start by checking the 1/3/5/10 year numbers.   Trustnet or similar to find the factsheets.
That's a decent starting point, before you delve into whether there are differences, etc: see how that fund is comparing with things you do perhaps know, or just other funds to compare.
The benchmark thing is a red herring, IMHO....different funds chose to use different benchmarks (although it might be reassuring to know you are in the top quartile, I guess)

The benchmark isn't a red herring. People may buy a fund because it
1. It has vastly outperformed its benchmark
2. It has great recent returns and appears on platform best buy lists
3. It has outperformed "similar" funds
whereas all of these should be reasons to dig a little further before buying. 

Like I said - might be reassuring to a few....but unless you know the benchmark in some detail and are only checking funds that submit to the identical benchmark, I still maintain it is a red herring.   
I have worked with IT benchmarks in my past: great to show certain things, but never the whole picture at all!
I think Linton nails it here:

Linton said:

I don't see there is any point in comparing a fund against a benchmark if the benchmark is not actually investible or does not truly represent the fund.  What do you do if the performance is poor? You really need to compare against other real alternatives that meet the objectives for wanting a particular fund.

When comparing performance I only glance at 1yr/3yr/5 yr/10yr because these figures can be strongly influenced by what happened over a short time period possibly recently or possibly 5-10 years ago.  So when I really want check performance of a fund against other possibilities with the intention to buy  I prefer to look at  the individual year by year values to see how they vary.  This may show up one fund being more volatile than another or a consistent pattern of outperformance.  Another worthwhile check is to see how they performed during market crashes. Sadly though the last real crash in 2008 is too long ago for much to be learnt except for sectors such as Wealth Preservation where the funds reason for existence is to have a viable long term strategy.

Having identified a significant performance difference, before one buys it is important to identify the cause.  Usually it is because the funds invest differently.  eg some high performing funds in general sectors are actually focussed towards small companies or tech.  If you want a small company or tech fund, fine. But if you don't there is no point in buying one because of apparent high performance relative to what you really want.

I would agree with the second paragraph, BUT for quick comparisons between a large number of funds, the 1/3/5/10yr numbers can give you a decent quick comparison.  I did that 15+ years back when scouring 80 funds to decide which ones should make a shortlist for deeper investigation!
Definitely drill more into what the funds are investing in, check how they compare in bad years v good ones, etc.....but use the headline numbers as a starting point!
None of us (to my knowledge!!) are investment analysts - if you are, please step in with your thoughts for the 'layman' to follow!

There are a few things to be aware of with investing
1. A decade is too short a time period to determine anything meaningful.
2. Diversification is key.
3. Risk and return are linked (as a first-order approximation)
4. The markets are broadly efficient.
5. The investment management industry is primarily a marketing machine, with participants attempting to differentiate themselves in a crowded marketplace. They want you to forget about 1-4 (or I should say, most do)!!

Investors broadly fall into two camps
1.Those that build up knowledge around items 1-4 and don't fall for the marketing spin of #5. Let's call those the "what has always worked" investors. 
2. Those that chase returns. Let's call those the "what is working now" investors. 

Obviously, the investment management industry are going to have a lot more luck with the second cohort (which is probably 95% of private investors). There are lots of cheeky tricks to take advantage of the "what is working now" group. Maybe an industry insider should write a book on it.

Example of a "what is working now" portfolio
1. A portfolio of bestselling fund managers who would typically currently be holding large-cap US growth/tech.
2. Poor diversification (on a sector, style, geographic and asset class basis).
3. If bonds are held, these would be something that has "worked recently", such as long-dated inflation-linked bonds.

cfw1994

Mmmm,....”A decade is too short a time period to determine anything meaningful”.
Whatever you say   

BritishInvestor

cfw1994 said:

Mmmm,....”A decade is too short a time period to determine anything meaningful”.
Whatever you say   

It's not me that says it, it's what the evidence points to. Look at the last decade as a perfect example.
It is so much easier, and profitable, to appeal to the #2 audience.

threlkeld53

I started this thread, so thanks for all the replies. They were all interesting. I guess then that apart from taking into account benchmarks, it's worth looking at the 1/3/5/10 year figure for all the cumulative, annualised, and discrete performances and add in one's own initiative. 

Whilst on this subject, are any of you familiar with iweb's site? Like on Trustnet's, it's possible to look at the data as above, using 1,3,5,10 years, plus the current year to date (ytd). However on iweb's site I don't understand their graph for the 1 year figures. Using Vanguard VLS60 as an example, it shows a 100% starting point for 26.2.20 and ending with 124% on 26.2.21. But VLS60 for the past year is around +6.5% not 24%. The iweb site does that for all funds for the PAST ONE YEAR OPTION on their graph, always it seems about 3 to 5 times more than what the fund has grown in the past year.