|
|
General instructions
Run the tester over the specified range of years. Available data
runs from 1986 through 2011 for most screens. You may start at any
month of the year, but other than January can only run through 2010/2011.
Stocks are picked from the screens every holding period (between
1 and 120 months) and held for that period.
Instead of selecting from the available screens, you may choose to
enter codes from any of the backtesters to create hybrid
tests. Simply copy the link listed at the top of a run & paste into the form.
Also see the backtester code guide.
|
|
|
Trading simulation
Start with an initial investment. You may also add more
money periodically. You can also subtract from the portfolio by adding a
negative amount.
Instead of adding or subtracting a fixed amount, you can elect to keep the
portfolio at a fixed allocation. This is useful for testing costs
without considering the effects of compounded growth, to see what your money
might do this year.
Normally, the stocks held in a screen are rebalanced on every trade.
You can also specify to never rebalance, which may allow one stock to dominate
a screen as it grows, or rebalance at a fixed interval (which must be a
multiple of the holding period).
You can pay a fixed commission (which is applied to every buy and
sell) as well as a percentage spread (half applied on buy, half on
sell).
Capital-gains taxes can be applied, both short-term and
long-term. Taxes may be taken out on every trade, quarterly, or
annually.
Screen rankings come out on Friday, and backtests are typically done with
the closing prices on the next Monday, one day after the rankings.
This lag between ranking and trading can be varied from zero days (trade on
Friday) and upward.
|
|
Dozens Backtest instructions
Select to screen to pick from. As with other testers, all stocks are
held for the holding period, but there is also a trading period
shorter than the holding period, which specifies how often new stocks are
selected. The portfolio is split into enough partitions so stocks can be
selected every trade date and held through other trade dates, until the holding
period for those stocks is completed.
There are two ways to select the stocks to hold. First, you can simply
specify the ranks in the screen. Or you can select the
top-ranked stock(s) not already held in a different part of the
portfolio. If all available selections are already held, the highest-ranking
stock will be re-selected.
For a typical "dozens" strategy, the top one stock not held is selected
every month, and held for one year. For a "dual semi" strategy, selected ranks
from the screen are selected every quarter and held for 6 months.
If the trading simulator is being used, the rebalancing of the
Dozens portfolio may be specified. This is independent of any rebalancing done
within the separate start months. Rebalancing between the different parts of
Dozens is more complicated than regular screen rebalancing, as it the entire
portfolio can't be rebalanced at the same time. The tester offers three
different methods of rebalancing. The first and simplest is to never
rebalance.
The second method is to keep a virtual margin account apart from the
portfolio. This is the default if run without the trading simulator. Every
cycle of the tester, money is moved to or from this account to keep the current
partition of the portfolio in proportion with the total (including the virtual
margin account itself). This virtual accounting doesn't change the actual
balance of the screen, but the stocks are all rebalanced to have the proportion
as if the money were taken out.
The third rebalancing method is like the second, except a virtual
cash account is used instead of a margin account. The only difference
is the balance of this account can never go below zero. A variation of the
cash method is to show the cash balance along with the rest of the
screen, and account for it in the CAGR. This will make for slightly reduced,
but more realistic, returns.
|