Limitations and Assumptions
Performance
The simulated performance is quite spectacular.
The actual performance may be different for
the following reasons:
- I do not have data for all delisted stocks (stocks
of companies that were in the index at some time but
that no longer exist).
- Some of the data of companies that merged
may be questionable (e.g. the merging companies
may pick conversion ratios for their stock,
issue new stock for the combined company and continue using one of the two ticker symbols).
- My analysis corrects the downloaded data for obvious errors and omissions (the data may not be up to date, may have missing trading days and may have missing or extreme values). These corrections may not be correct.
- It’s possible that I have made implementation errors and all results are bogus. The probabilty of this is pretty low since I've been following this strategy for quite a while and haven't found any descrepencies.
Number of Trades
A buy is considered a trade and a sell is considered a trade. For example, if you buy and sell 50 stocks in your portfolio and you’re still holding 5 stocks today, you will have made 105 trades (50*2+5).
Annual Return
The annual return in the text summaries is calculated by annualizing the combined return over the entire simulation period.
Profit Rate
The profit rate in the performance charts is calculated by annualizing the profit over the last 90 trading days. You can use the profit rate to determine whether it is below or above the historical average, i.e. the annual return shown in the text summaries.
Buy, Sell and Hold Signals
The charts of individual stocks show "buy", "sell" and "hold" signals but that does not mean that I buy/sell/hold them at that time or that I buy them at all. The analysis will pick only a maximum of 5 stocks using the 3 criteria outlined in the
General Strategy even though many others may have a
"buy" signal at that time as well. Then it holds on to these stocks until a signal
change occurs (i.e. it changes from a "buy" to a
"hold"). If a signal change occurs, then I sell this stock and replace it with another stock that has the highest indicator of all stocks out of the 100 stocks that currently have a
"buy" signal and that is not in the current portfolio.
A result of this approach is that I tend to buy
a stock some time (sometimes a considerable amount of time) after the "buy" signal occurred.
Also, this approach tends to avoid that I buy a stock where the signal flips often.
Buy/Sell Prices
The simulation uses the stock price at the open
(i.e. the price paid for the stock at the first
trade of the day) as the price at which stocks
are bought and sold. For example, if a stock has
a sell signal after the market closed on August
12, 2004 (based on its price at the close on August 12,
2004 as well as its prior closing prices) then the analysis assumes that stock will be sold at the market open on August 13,
2004 whatever that price may be. This is a reasonable assumption because this will allow you to perform an analysis after the market closes for the day, make your investment decision and act on it when the market opens the next day. In practice, it will be impossible to buy or sell stocks at the open price. In some cases you will have to pay more and in other cases less. In the long run this should not have a material effect on the simulation results.
Trading Fees
The trading fees are included in the analysis. I generally use a trading fee of $7 (the fee at Scottrade). You can look up the trading fee used in the simulation among the simulation parameters
near the end of the PDF file.
Margin interest
If you trade on margin then you borrow money from your broker to buy stocks (see
Additional Strategies). In that case you will have to pay interest
on the money borrowed at a certain interest rate. This rate is usually tied to the prime rate (e.g. prime rate plus 1%).
If the analysis uses a target margin ratio then it deducts interest expenses based on the daily margin amount during the preceding month.
The interest rate on each day is determined based on historical data from the Federal Reserve.
Back to Simulated Results
|
