ETFreplay.com is a research, analysis and backtesting website for Exchange Traded
Funds.

ETFreplay’s tools are designed to allow investors to find, test and pursue a robust
and repeatable process for gaining exposure to up-trends while avoiding large drawdowns.

What is an investment process?

An investment process is a set of well-defined steps undertaken in a consistent
manner to create and maintain an appropriate portfolio designed to meet your investment
goals. A good process balances the need for a level of return with the risk of loss
(drawdowns).

How many ETFs does ETFreplay cover?

ETFreplay covers more than 98% of all U.S. ETP (Exchange Traded Product) assets
and we routinely add new ETFs.

However, before adding new funds we require that they have approximately $100 million
in assets or at least sufficient asset momentum that $100 million will be reached
in the near future.

This requirement is necessary because if an ETF doesn't trade volume then the closing
price will not accurately reflect the underlying index, which in turn will compromise
any backtest results.

How do I decide which ETFs to include in my backtest portfolio(s)?

We generally assume that investors have their own ideas on which asset classes,
sectors, countries, regions or industries they want to monitor.

However, if you are agnostic about the universe of ETFs, portfolios with a good
broad mix of asset classes, such as the well known Permanent Portfolio, Ivy Portfolio or our own Sample portfolio
provide a decent starting point.

Once this defined universe is in place, the tools on ETFreplay can be used to find,
test and pursue a robust and repeatable process for gaining exposure to up-trends
while avoiding large drawdowns.

Do returns on ETFreplay include dividends?

Yes, all returns and calculations (including moving averages) on ETFreplay are Total
Return, which accounts for the receipt and reinvestment of dividends and distributions.

Does ETFreplay use calendar months or trading days in its calculations?

We only use trading days and start with the convention of 252 trading days per year,
therefore:

6-months is 252/2 = 126 trading days

3-months is 252/4 = 63 trading days etc.

This means that when, for instance, 3-month returns are chosen on the Screener,
the calculation will always count back 63 trading days, making comparisons over
time consistent.

How does the ETF Screener calculate Relative Strength?

The ETF Screener is a statistical model loosely based
on the Sharpe Ratio, which measures reward per unit of risk. The Screener takes
this concept and decomposes it into three separate factors:

Higher timeframe total return (ReturnA)

Lower timeframe total return (ReturnB)

Volatility

From these three factors, and the weights you assign to them, the overall rank is
calculated.
Both the lookback periods and the weight of each factor can be changed. For example,
if you want to rank the ETFs in a list by only 6-month total return, then:

Set ReturnA to '6-Months' and set its weight to 100%

Set the weights of ReturnB and Volatility to zero

Click 'Run Model'

The full process employed by the Screener to rank ETFs is explained in How The ETF Screener Works (subscribers only)

Volatility – What is it and why is it important?

Volatility is the annualized standard deviation of daily returns.

i.e. 20-day Volatility is the standard deviation of the past 20 1-day returns multiplied
by sqrt(252) (annualized).

Volatility is a measure of risk. Risk is uncertainty and the larger the range of
possible outcomes, the higher the volatility will be and therefore the greater the
risk. This tends to be borne out when it comes to drawdowns, with higher volatility
securities typically experiencing larger drawdowns than lower volatility ETFs.

ETFreplay has several tools to help investors visualize risk, including:

Monthly Returns View what sort of monthly return is to be expected for the ETFs in your portfolio
and what constitutes an outlier

Down Day Stats See how the ETFs in your portfolio have performed when the broader market has had
a bad day

Return vs. Volatility Evaluate the risk / return performance of the ETFs in your portfolios over any time
period

ETF Volatility Compare the rolling realized volatility of up to five ETFs

All MAs on ETFreplay are calculated using Total Return.
i.e. they include not just closing prices but also account for the receipt and reinvestment
of any dividends and distributions. Accounting for dividends is necessary otherwise
you don't get the full picture and it leads to faulty comparisons between those
ETFs that pay dividends and those that do not.
See Total Returns vs. Price Return

Moving Averages are simple moving averages (SMA), unless otherwise specified. Some
backtests allow the choice of Exponential (EMA) or Simple moving averages.

The length of a simple moving average indicates the number of data points it includes.

i.e.

a 10-month MA contains 10 data points - the month end total return values

a 200-day MA contains 200 data points - the daily total return values

While these two MAs cover a similar period of time, they are not exactly the same.

Can mutual funds be backtested?

A few index mutual funds have been added to fill in spots that don't have 10-15
years of data available, like many ETFs do. The list of index mutual funds can be
found by clicking on the 'Symbols' link when entering tickers on various pages.

Note: These index mutual funds can be used for longer-term backtesting. For shorter-term
backtesting (daily, weekly or semi-monthly rotation / update schedules) users should
really stick to ETFs rather than mutual funds - due to the way that fixed-income
mutual fund accounting does not embed accrued interest in the NAV.

How does adding a quantitative process like ETFreplay.com backtesting help investors?

There is an entire field of study devoted to behavioral finance. All people suffer from various biases
that can range from overconfidence to fear. Adding a quantitative component to your
overall process helps by offering an objective view. This does not mean you MUST
follow your model --- backtesting and models in general should be used as key inputs
to your overall research process. Good judgment will always be a part of balancing
reward and risk.