| Chuck LeBeau meint....... |       |
The STC RSI Stock Trading System By Chuck LeBeau
Our Bulletin number 48 about using RSI sparked some very
interesting discussion on our FORUM. One of the messages
suggested that buying stocks when the RSI was below 25 and
then selling them when the RSI went above 45 produced some
excellent historical results, particularly in the Dow
stocks.
I thought I would check it out and ran some historical
tests. Lo and behold it worked remarkably well. I
immediately observed that there was the makings of a viable
system here. All it needed was a little refinement.
After a few days on tinkering I came up with a version of
the system that I think is worth trading. Since our members
on the FORUM originated the basic system, I thought I would
also pass along my work on this system so that all of our
members might benefit.
Here are the stock trading rules I came up with:
ENTRY: When the 14-day RSI is 25 or less, enter an order to
buy tomorrow on a stop at today s close plus some small
amount. Use a couple of cents or .02 units of Average True
Range above the previous close. We need to avoid buying on
weakness because it s a bit like trying to catch a falling
safe. When the RSI is at a low level the stock is likely to
be gapping lower and falling apart to the downside and we
don t want to buy that much weakness. So to avoid buying on
further weakness we require that we must buy only on a price
above today s close. That requirement keeps us out of a lot
of trouble.
PROFIT TAKING: Take profits when the 14-day RSI recovers and
closes at any level above 45. Just sell the next day on the
open. Closing out the trade at such a low level on the RSI
seems counter-intuitive but it works. What tends to happen
is that the RSI can often recover quickly and gap well above
45 so that we are often taking the profits at levels much
greater than 45. I tried a lot of other levels trying to
increase the profits on the winners but selling above 45
makes the most money.
LOSS PROTECTION: After the trade has been open for four days
or more exit on any day that the RSI closes lower than
yesterday. Just sell the next day on the open. This is our
equivalent of loss protection. If we have been in a trade
for four days or more and the RSI has not recovered to 45 we
don t want to hold it while the RSI declines again.
I tested the simple system described above on a portfolio of
28 stocks from 1987 through October 2001. In this more or
less random portfolio there were some Dow stocks plus a few
stocks with more volatility. The results were generally
acceptable but there were some particular results that I
thought were truly amazing.
First the typical performance results from buying $10,000
worth of stock on each signal with no compounding:
Total net profit: $255,557 Number of trades: 1849 Winning
trades: 1124 (60%) Losing trades: 725 (40%) Average winning
trade: $489 Average losing trade: $406 (Ratio avg. win/avg.
loss 1.21) Maximum consecutive winners: 18 Maximum
consecutive losers: 8 Average bars in winners: 3.5 Average
bars in losers: 4.4 Largest intraday drawdown: $28,820 (This
occurred in October 1987) Profit factor (Gross profits
divided by gross losses) 1.87 Sharp ratio: 1.12
Now these results are pretty good overall but what really
got my attention was a couple of additional calculations
that are provided in my BEHOLD software. We like to look at
the number of rolling 2month windows and see how many
12-month periods were possible. (After the first eleven
months each new month provides a new12-month period to be
examined.) There were 161 of those 12-month windows in our
test and every one of them was profitable. This system did
not have a single 12- month period where it lost money.
That is truly remarkable. Although the biggest intraday
drawdown was in October 1987 (no surprise) the system
actually finished that month with a big profit. This is a
system that seems to be able to take advantage of market
weakness.
Another measurement that got my attention was the percentage
of bars in the market and the dollars returned per day in
the market (based on buying $10,000 worth of stock). This
system is not in the market very often; it had trades on
only 7% of the bars. But it returned a profit of $35.64 for
every bar in the market. To put this number in perspective,
any return of $10 per day or more on a $10,000 investment is
considered very good.
Now as always, there are a few warnings. When testing
stocks that have been back adjusted for splits it is nearly
impossible to factor in commissions with any accuracy so no
commissions or slippage are included in these results. But
we are trading stocks now for a penny a share or less and
commissions should not have a big impact because the system
trades very infrequently. (About 4 trades per year per
stock.) If we deduct a few thousand dollars from our
results to cover costs the results are still very
acceptable.
This system has no fixed dollar stop loss so the potential
loss on any trade is unlimited. For this reason you would
want to trade very small positions relative to your total
capital (a good idea in any case). That way if you took a
big hit on one trade it would represent a very small
percentage of your capital. In our historical testing the
largest losing trade was $3,239.
We think this RSI based system deserves further examination. We
like the way it takes advantage of buying during periods of
extreme market weakness. If we combine this system with a
good trend-following system that can make money when the
market is strong we should be able to make money on a very
consistent basis.
Test the system yourself and see what you think. If you
come up with any improvements please let us know.
Good luck and good trading.
|
04.11.2001, 23:10
