How can you tell if an approach or a tilt index provides alpha?

Manage
1

I investigated the strategy underlying these two indices:

Fool-100 (https://www.solactive.com/indices/?index=DE000SL0EY00)

Founder-led BOSS-ETF (https://www.solactive.com/indices/?index=DE000SLA21M5)

Both outperform the market by a wide margin, but they are also cap-weighted. How is it possible to determine whether it is the Fool-100, or Founder-led companies, that drive the return, or if it is simply because a majority of the companies are large cap, and hence benefit from the broader momentum in the S & P500? ( Large Cap)?

Or how to determine how much of the return is attributable to alpha and how much is due to market and cap weighting?

2

You can create a equal weight index of the constituents on this site and compare it to those indexes.

3

Here is a start:
BOSS ETF
Fool-100

Both ETFs outperformed QQQ–BOSS by a lot, and Fool-100 by a little–and would have done a lot better had they used equal weightings.

However, I using most recent holdings of their respective ETFs so there is a strong risk of look ahead bias. Indeed, the most recent two years saw under-performance from both vs QQQ.

Is there a way to use historical holdings? I know that there was some talk about that.

4

[quote]
However, I am using most recent holdings of their respective ETFs so there is a strong risk of look ahead bias.
[/quote]Using the most recent holdings from an ETF for backtesting is a NO-NO. This will give survivorship biased returns.

For BOSS one can get historic holdings from 11/30/2019 (updated every 3 months) here:
https://whalewisdom.com/stock/boss

5

Georg, did you ever find an easy way to import historical holdings into this site?

Georg, as you know, your link provides names but not tickers. It is a lot of work to convert names to tickers. Did you find any shortcuts?

Furthermore, did you find an easier way to convert old tickers to the format used here with a ^nn in the ticker name?

6

There is no easy way to do this.
You have to upload the historic holdings with dates into a stock factor - cusip numbers work best.

That link used to have cusip numbers for the stocks. I have a query in as to why this has been removed.

7

Okay, here are some facts:

  1. Each ETF is required by US regulations to file quarterly holdings with the SEC. The SEC form is called NPORT-P. I assume that with a little more work I should be able to automate the process of downloading these forms.

  2. With a free (for now) account at https://docoh.com/fund/S000056121/global-x-founder-run-companies-etf/holdings-history, you can download all historical holdings for this (and other) ETFs which include tickers!

8

This link works better than mine. Remember the dates are actual filing dates. Point-in-time for us is 2 months later. So the list of 10/31/2021 would be PIT as of 12/31/2021.

9

Python code to create csv from docoh download (in clipboard):

import pandas as pd

df = pd.read_clipboard()
df.index = df['Ticker']
df : pd.DataFrame = df.drop('Ticker', axis=1)
txt = ''
for date, col in df.items():
    for ticker, amt in col.items():
        txt += f'\n{date}\t{ticker}\t{1 if isinstance(amt, str) else 0}'
txt = txt.strip()
print(txt)
10

Here is the performance of BOSS using historical ETF holdings:
https://www.portfolio123.com/port_summary.jsp?portid=1669252

Underperforming QQQ

NB: Some tickers were not available in docoh. Probably less than 5%.

11

Here is the performance over the same period of 20 stocks from Fidelity Founders Fund FIFNX with P123 Core Sentiment ranking system.
Much better performance.

FIFNX.png
FIFNX.png1567×909 185 KB

12

Here’s another fund that takes the same approach, but with a higher proportion of midcap stocks and 450 companies: https://docoh.com/fund/S000067868/frc-founders-index-fund / holdings-history

https://funddocs.filepoint.live/firstrepublic/foundersindex/

13

Azouz, nice work! Can you make your universe public?

14

Done. The universe is based on a custom stock factor.
Attached the file I used to import data for that stock factor.

StockFactor_ETF_BOSS.csv (24.7 KB)

15

Thanks!