You trust M*'s statistics more than mine. Good idea. Seriously. Just as I would trust M*'s suggestion that 12% of funds beat VBINX over your claim that just 7.4% did. It's large discrepancies like that (1.5x) which raise credibility issues.
Over time, you'll find that a fair number of people here don't trust M* either, at least when it comes to which funds to count. You just hinted at that yourself with PRWCX.
Though closing a fund for
years strikes me as a funny way to gather assets. Sooner or later you'll learn that TROW is the most investor-friendly publicly traded investment company (yeah, I know, faint praise).
Regarding which funds to count, you dismissed my selection of VBIAX as being the same fund as VBINX. Fine with me, I agree with your thought that funds should only be counted once. But then you turned right around and referenced a M* figure (12th percentile) based on 501 moderate allocation "funds" with ten year records.
Two of the funds in that list of ten year funds are VBIAX and VBINX. You were the one who picked the list of funds, I just selected a fund off your sanctioned list.
You don't have to believe me. The 148 unique fund figure I cited came straight from M*. You can get it yourself. There may be around 500 moderate fund
share classes with ten year records, but there are only around 150
funds (using your understanding of funds - counting VBIAX and VBINX as one fund).
One doesn't need a M* premium membership to figure this out. Just use their
basic fund screener. Though it is limited to returning 200 funds (i.e. share classes), one can coax all 500-ish out by asking for them in three pieces - 4* and 5* funds, 3* funds, and then 2* & 1* funds. (Any ten year fund should have a star rating, so that captures all of them.)
Ask for moderate allocation funds with ten year performance exceeding (-10%), and do the star breakdown described. The neat thing about this exercise is that because the results come back alphabetically, it's very easy to scan for unique funds (as opposed to share classes). While this will still double count some funds (if a single fund has one share class with 3 stars and another with 4 starts), it gets you in the right ballpark.
It will show how exaggerated that M* 500 fund figure is, at least if you're not going to allow me to pick VBIAX :-) (i.e. if you're going to stick with counting unique funds only once).
I count around 70 unique 4 and 5 star funds (out of 191 share classes,
including three instances of Vanguard Balanced Index). I could be off by a couple of unique funds as I counted quickly. The point is to see, in front of your eyes, M*'s (not my) data - how many unique funds there are, how some funds get counted many more times than others. Just look at that first page - it's nearly filled with different share classes of a single fund - American Funds American Balanced.
Along with 76 unique 3 star funds (out of 176 share classes) , and 43 unique 1-2* funds (out of 124 share classes), this totals 189 potentially unique funds out of 491 share classes.
We can eliminate much of the double counting by performing two more screens - checking for 2 and 3 star funds (and subtracting any fund that shows up with both 2 and 3 stars), and then checking 3 and 4 star funds.
Unfortunately, there are more than 200 2* and 3* funds combined, but even working with the 200 returned, there appear to be at least 18 duplicates there, and at least 19 duplicates between 3* and 4* share classes. Subtract 37 from 189, and we get 152 - darn close to the 148 unique funds that I said came from M* data.
Notice what I did. I showed how to reproduce the results and provided the methodology - what the definitions were, what was being counted. This adds to both comprehension and credibility. I showed that you too can come up with figures similar to the ones I gave above.
This exercise demonstrates just how much shrinkage there is going from share classes to unique funds. It offers a glimpse into how much distortion there is, when a single fund like American Funds American Balanced can get counted 16 times, while a fund like T. Rowe Price Cap Ap is counted only twice. This insight can lead to better understanding of what the numbers mean (and what the odds really are, and what factors can improve those odds).
Don't follow your 322 funds. They're not unique funds (by your definition), and they're skewed (I hope by now you're at least considering the possibility that by using Fidelity's screener to eliminate closed funds, you may be distorting your results by excluding funds like PRWCX and Wellington). Spend some time figuring out what you do want to observe, find a way to do it accurately.