Trustnet FE scores - anomaly?

aroominyork

I was looking at Trustnet's table for Japan and noticed that HSBC Japan Index is FE91, while Fidelity Index Japan is FE75, suggesting Fidelity is significantly less volatile. Trustnet explains "The scoring system uses a minimum of 18 months and a maximum of 3 years of weekly total returns to measure the volatility of an instrument relative to the UK leading 100 shares. All values are rebased to Sterling, for UK Risk Scores. But less weight is given to the older returns, so that the score is more sensitive to recent events, but not excessively so." Yet looking at a 3 year chart there is little difference between them.

I thought maybe HSBC has fewer holdings so its ups and downs are more extreme (even if not noticeable on the chart) but in fact it's the opposite - HSBC has 529 holdings while Fidelity has 272 holdings. So what explains the difference in FE scores? HSBC holds some mid-caps which Fidelity does not, yet any increased volatility caused by them does not seem reflected in the chart.

masonic

There are several possible explanations for this, but the most likely is that (as you have identified) HSBC has a larger number of holdings, so likely has a lower average market cap. Smaller companies are typically more volatile, and both the volatility and performance in your pasted plot post-April is visibly different and suggests that the HSBC fund has not given such a smooth ride as the Fidelity fund. Another possible contributing factor could be that the available data used for the risk score of the two funds spans different dates.

aroominyork

HSBC more volatile since April? Nope, it doesn't look like it.

masonic

Then perhaps it is the latter explanation. Those calculating the scores would be able to provide you with a definitive answer.

Edit: It should be noted that for the HSBC fund "This fund does not subscribe to Trustnet", and no risk score is displayed on the fund's Trustnet page, so is the score you have seen specific to that fund, or some sort of estimate based on its sector?

aroominyork

This is where a 99-styled person would usually chip in to explain...

masonic

aroominyork said:

This is where a 99-styled person would usually chip in to explain...

He's still around, so don't lose hope

Thrugelmir

masonic said:

Edit: It should be noted that for the HSBC fund "This fund does not subscribe to Trustnet", and no risk score is displayed on the fund's Trustnet page, so is the score you have seen specific to that fund, or some sort of estimate based on its sector?

Trustnet provide a ranking for all funds in the summary page.  Probably a weighted average as an FE of 91 sits around the middle of the 80 funds benchmarked. Why promote a fund a fund that doesn't subscribe and pay fees to Trustnet! 

Linton

Perhaps it is relevent that the two funds track different indexes.  Fidelity Japan Index tracks the MSCI Japan (net total return) index whereas the HSBC fund tracks the FTSE Japan Index.  Also both use derivatives but they presumably use different ones.  The Fidelity fund info states that it may use currency hedging whereas HSBC  doesnt seem to mention it.  

So it is conceivable that the funds behave differently on a day to day basis.  If you look at the trustnet graphs for different individual months ISTM the HSBC tracker can at times be much more volatile day-to-day than Fidelity.

aroominyork

Interesting, Linton. However even if there is more day-to-day volatility in HSBC, FE scores are based on weekly total returns so those would, to a reasonable degree, be smoothed out. Below is Trustnet for three months and each vertical line represents a week - I see nothing to explain the difference between FE91 and FE75 if each intersection with a vertical line is taken as a data point. And remember, FE does not measure volatility compared to the index each fund seeks to track (ie MSCI for one, FTSE for the other) but it measures against the FTSE100 so we are comparing like with like.

masonic said:

It should be noted that for the HSBC fund "This fund does not subscribe to Trustnet", and no risk score is displayed on the fund's Trustnet page, so is the score you have seen specific to that fund, or some sort of estimate based on its sector?

Specific to the fund. Even though FE scores are not posted on the page of non-subscribing funds, they are shown when running tables of the full sector, and are also shown if you include a non-subscribing fund in the Portfolio tool. (Re. Thrugelmir suggesting it is an average, I do not think so - in the past I have included high/low volatility non-subscribing funds in the Portfolio tool and I think it showed an appropriately high/low FE score.)

Linton

aroominyork said:

Interesting, Linton. However even if there is more day-to-day volatility in HSBC, FE scores are based on weekly total returns so those would, to a reasonable degree, be smoothed out. Below is Trustnet for three months and each vertical line represents a week - I see nothing to explain the difference between FE91 and FE75. And remember, FE does not measure volatility compared to the index each fund seeks to track (ie MSCI for one, FTSE for the other) but it measures against the FTSE100 so we are comparing like with like.

Can you give any reference to how you believe the Risk Score operates? From my reading of the very small amount of information that Trustnet provides:

1) The volatility over the previous 3 years is recalculated on a weekly basis.  That does not mean that only weekly data points are used.
2) The calculated value of the volatility is scaled so that the FTSE100 comes out as 100 and cash as 0.  That does not mean that the funds are compared with the FTSE100 at the detailed level.  Volatility is not measured against something else, it is purely based on the performance graph of the fund under consideration.

See https://www.investopedia.com/terms/v/volatility.asp. A common measure of volatility is the standard deviation of the data points about an average. Whether that is the measure used by Trustnet and if so what the average used in this case is I dont know.  

aroominyork

Both are good points, Linton. I have not looked further than the small amount of info Trustnet provide - the difference between the two just looks wrong!