StockFetcher Forums · General Discussion · Calculation of RSI(2)<< 1 2 >>Post Follow-up
rharmelink
81 posts
msg #47448
Ignore rharmelink
modified
10/13/2006 4:26:00 PM

gocats,

The problem is in your calculation of the average gain and average loss. What you are computing is the starting point for the average gain and average loss. From that point on, those averages are "smoothed". For example, you used:

AL = (0.03/2) = 0.015 (In the last 2 days there was one red day and fell $0.03)
AG = (0.12/2) = 0.06 (In the last 2 days there was one Green day and rose $0.12)

But, on a smoothed basis, they would be:

AL = 0.0239
AG = 0.1285

I would suggest at least 8 days of smoothing. You should be doing something like this for EZPW (as of 10/02):

AL = (0/2)+(0.03/4)+(0/8)+(0/16)+(0/32)+(0/64)+(1.21/128)+(1.54/256) = 0.02296875
AG = (0.12/2)+(0/4)+(0.02/8)+(0.32/16)+(1.01/32)+(0.76/64)+(0/128)+(0/256) = 0.1259375

Then:

L = 0.02296875 - (0.1259375 / (100 / (100 - 5.1) - 1)) = -2.320456495

...instead of:

L = 0.015 - (0.06/ (100 / (100 - 5.1) - 1)) = -1.1

Much closer to the actual value of -2.38. Of course, adding more terms will increase accuracy, but you can see that each preceeding day's gain or loss becomes less and less important as the denominator doubles.






gocats
22 posts
msg #47451
Ignore gocats
10/13/2006 6:00:09 PM

Thanks rharmelink and Jim.

Now I got a better picture.


rharmelink
81 posts
msg #47608
Ignore rharmelink
10/24/2006 6:27:19 AM

By the way, I ran the calculations on 13 years of SPY data. I determined the buy (or sell) point for the following day using the formulas specified earlier, using RSI targets of 10 and 90. If no position was held, a buy was initiated if the price exceeded the buy point. If a position was held, it was liquidated when the price fell below the sell point.

The return of buy and hold was 15.5% annually over the 13 years.
The return of the RSI(2) was 13.0% annually over the 13 years.

That does assume a 5% return when the money was out of the market. The RSI(2) method was only invested in the market 1012 days out of 3442, so the risk was reduced quite a bit. Overview of major moves:

B&H: $10K (1993) --> $40K (2000) --> $21K (2002/2003) --> $39K now.
RSI: $10K (1993) --> $29K (2000) --> $22K (2002/2003) --> $32K now.

So, as expected, RSI(2) doesn't perform well in a roaring bull market.

Here's a chart of the results:

http://ogres-crypt.com/images/RSI-2-vs-Buy-and-Hold-for-SPY.jpg


contrahawk
68 posts
msg #47623
Ignore contrahawk
10/25/2006 1:10:13 AM

rharmelink says: So, as expected, RSI(2) doesn't perform well in a roaring bull market.

********************************************************************************

Consider this, rharmelink: in 2006, there have been 43 occasions when AAPL has closed the day with daily RSI(2) < 15. Let's call each of those occasions a "signal day" and the next day a "following day". On all 43 of those occasions, the High put in on the following day was above the Close of the signal day.

Do you think I look for days when AAPL closes with RSI(2) < 15? You betcha!

Is it a "lock" that 1) the next time AAPL closes the day with daily RSI(2) < 15 the following day will see a High above the signal day's Close and 2) I will trade AAPL in such a way that I book a profit on the following day? No to both.
However, there is a STRONG likelihood that a close with daily RSI(2) < 15 will be followed by a High > signal day's Close and there IS a profit opportunity there for a trader who works the opportunity wisely/correctly.

This trader would say that RSI(2) "performs well" if performance is defined as pointing out profit opportunities. RSI(2) does not trade for me, however -- I have to do that for myself.

My awareness of RSI(2) came from the postings of TheRumpledOne in this forum though I make use of it in ways that he may not have described.

Jim


rharmelink
81 posts
msg #47625
Ignore rharmelink
10/25/2006 3:33:27 AM

Jim -- I agree with what you're saying. That's why I tried to spell out exactly what I was using as the buy and sell criteria. The generation of a technical indicator is the easy part. There are many ways in which it can be used.

Even though the method I used underperformed buy and hold, I was impressed that it did as well as it did holding the index only a third of the time. Theoretically, the new leveraged (SSO) and inverse (SH or SDS) could be used to enhance those signals. Unfortunately, they don't have enough history to check out on a backtest.

By the way, using the same buy/sell signals on AAPL as I used on SPY gave a -8% compounded return on the stock versus +22% for Buy and Hold...and almost all of the gain for the buy and hold is during 2004 and 2005.

>> Do you think I look for days when AAPL closes with RSI(2) < 15? You betcha! <<

It may be working in 2006, but it didn't in 2005. Here is a table by year showing the success rate of the high exceeding the close of the previous day for that day's RSI(2) value:

Year ..01-15 ..15-50 ..50-85 ..85-99 Total
1993 085.1% 88.6% 94.2% 96.4% 90.7%
1994 095.6% 90.7% 88.4% 94.6% 91.3%
1995 088.9% 87.7% 90.0% 89.3% 88.9%
1996 089.7% 82.6% 87.5% 94.7% 87.4%
1997 088.7% 84.6% 76.5% 90.6% 84.2%
1998 076.7% 85.3% 89.8% 85.7% 86.1%
1999 090.9% 95.8% 83.9% 80.0% 87.3%
2000 091.7% 90.4% 86.3% 90.6% 89.3%
2001 092.7% 85.7% 78.2% 80.0% 83.5%
2002 091.8% 86.3% 81.4% 75.6% 84.5%
2003 097.1% 93.2% 83.1% 87.9% 89.7%
2004 090.6% 84.5% 90.1% 86.8% 87.7%
2005 082.6% 93.2% 89.8% 88.2% 89.7%
2006 100.0% 89.7% 76.1% 86.5% 86.8%
Total 090.5% 88.4% 85.6% 87.1% 87.6%



contrahawk
68 posts
msg #47630
Ignore contrahawk
10/25/2006 8:21:34 AM

rharmelink,

you're a tough one to cross statistics with! Good, speedy analysis! Don't lose sight, though, that even at 75% there are 3 occasions where the High of the "following day" exceeds the Close of the "signal day" for each occasion where it does not.

Happy trading.

Jim


StockFetcher Forums · General Discussion · Calculation of RSI(2)<< 1 2 >>Post Follow-up

*** Disclaimer *** StockFetcher.com does not endorse or suggest any of the securities which are returned in any of the searches or filters. They are provided purely for informational and research purposes. StockFetcher.com does not recommend particular securities. StockFetcher.com, Vestyl Software, L.L.C. and involved content providers shall not be liable for any errors or delays in the content, or for any actions taken based on the content.


Copyright 2016 - Vestyl Software L.L.C.Terms of Service | License | Questions or comments? Contact Us
EOD Data sources: DDFPlus & CSI Data Quotes delayed during active market hours. Delay times are at least 15 mins for NASDAQ, 20 mins for NYSE and Amex. Delayed intraday data provided by DDFPlus