Written by and copyright © 20052017 by Thomas N. Bulkowski. All rights reserved. Disclaimer: You alone are responsible for your investment decisions.
See Privacy/Disclaimer for more information.
This article discusses whether intraday (day trading) pivot points in stocks work.
Pivot Points Summary
Testing using intraday data on the 1minute scale through both bull and bear markets shows that pivots using the standard formulas for pivot points works from 51% to 60% of the time.
Other pivot formulas work better (Modified pivots: up to 61% of the time) and worse (Fibonacci pivots: 28% to 36% of the time).
Pivot Points Background
Pivot points came into being to help floor traders determine support and resistance levels before the trading day began. That allowed them to concentrate on
trading instead of number crunching.
I found three sets of formulas to calculate the pivots, but will only discuss two of them.
Here is the first.
Second resistance, R2 = P + (High  Low)
First resistance, R1 = (2 * P)  Low
Pivot point, P = (High + Low + Close)/3
First support, S1 = (2 * P)  High
Second support, S2 = P  (High  Low)
If the day's high is 12, the close is 11 and the low is 10, R2 = 13, R1 = 12, P = 11, S1 = 10 and S2 = 9. Those become the support and resistance levels for the next day.
The next pivot point formulas are the same, but they look different.
Second resistance, R2 = P + (R1S1)
First resistance: Same as R1 above
Pivot point: Same as P above
First support: Same as S1 above
Second support, S2 = P  (R1  S1)
The formulas for R2 and S2 change, but if you plug in the numbers, you get the same results: R2 = 11 + (12  10) or 13 and S2 = 11  (12  10) or 9. They are equivalent formulas.
I prefer the first set of equations because it is easier to plug in the high and low prices instead of calculating R1 and S1 first.
Pivot Points Example
For example, here is a picture of Alcoa (AA) on the 5minute scale.
The day's high on 2/7/2003 is 19.91, the low is 19.53 and the close is 19.65.
Plugging those numbers into
the pivot formulas, give the pivot lines as shown by the horizontal dashed cyan lines. R2 is not shown because it is above the top of the chart.
The pivots show the predicted support and resistance areas for the next day's trading (2/10/03). Notice that the stock seemed to ignore the pivots by rising above R1 just after the
open, dropping through P, trending up above R1 again, then finding support near P, before closing above R1 again.
The stock traded in a narrow range but with only one valley showing support at P near the close.
Pivot Points Methodology
I used data covering both bull (2/2003 to 12/2005) and bear (2/2007 to 2/2009) markets on 55 stocks on the 1minute scale. I used a 3 cent band above and below (6 cents total)
each pivot point (P, R1, and so on) when trying to determine if price turned near the pivot. If the bands overlapped (including a buffer), I excluded that day's data. In other words,
the stock had to have a tall highlow trading range. The minimum highlow range was approximately 44 cents.
The first test counted how many significant (the highest high or lowest low from 5 bars before to 5 bars after or 11 bars total of a peak or valley).
peaks or valleys appeared within each pivot band. The test showed how often price found support or resistance near a pivot.
The next test counted how often price transited the band, either from above the band to below it, or below it to above it. This test showed how often the pivot failed to show
support or resistance.
Pivot Points Results
I used the pivots as defined by the formulas discussed earlier and found that the pivots worked more often than I expected. Here are the results
How Often Do Pivot Points Work?
 P  R1  S1  R2  S2 
Standard Pivots Below 
Support  55%  53%  58%  55%  54% 
Resistance  56%  60%  51%  56%  51% 
Modified Pivots Below 
Support  57%  55%  59%  46%  48% 
Resistance  58%  61%  53%  51%  43% 
Fibonacci Pivots Below 
Support  36%  35%  33%  29%  32% 
Resistance  36%  34%  35%  34%  28% 
Standard Pivots
The first set of lines show support and resistance at the various pivot points (P, R1 and so on). For example, I found that price formed 80,228 valleys during the
test period and in another 64,844 cases, price sailed right through P without stopping. The ratio, 80,228/(80,228 + 64,844) gives 55% which I show in the cell.
The next line down shows resistance. In a similar manner, I found peaks that showed price turning within the 6 cent bands surrounding P and the other times were
those when price sailed right on through. The ratio, 56%, appears in the table.
Modified Pivots
The two lines below Modified Pivots used the same methodology except that I changed the formulas. Here is what I used for this test.
Second resistance, R2 = High + 2 * (High  Low)
First resistance: R1 = High
Pivot point: P = (High + Low)/2
First support: S1 = Low
Second support, S2 = Low  2 * (High  Low)
I set the day's high to R1, the low to S1, took the average of the day's highlow range for P and doubled that range for R2 and S2. The numbers in the table show
that this worked better in all cases except S2 and R2. I did not choose these formulas for any reason. I just tried them as a type of "control" group to see how they
worked compared to the standard formulas.
Fibonacci Pivots
The last set uses pivots based on a Fibonacci retrace value of 38%. Here are the formulas.
Second resistance, R2 = High
First resistance: R1 = High  38% * (High  Low)
Pivot point: P = (High + Low)/2
First support: S1 = Low + 38% * (High  Low)
Second support, S2 = Low
The day's high becomes R2, the day's low is S2, the pivot remains the same as the prior P formula, and R1 is a 38% retrace of the highlow range down from the high. S1 is the same measure up from
the day's low. These pivots, probably because they were so narrow, flopped. They are the worse performing of the bunch.
 Thomas Bulkowski
Written by and copyright © 20052017 by Thomas N. Bulkowski. All rights reserved. Disclaimer: You alone are responsible for your investment decisions.
See Privacy/Disclaimer for more information.
Never argue with a fool. People may not know the difference.
