Do Pivot Points work?

Testing Pivot Points

Do Pivot/Resistance/Support Points and Levels Work?

This is a very wide ranging question and has many answers ranging from not at all to very well. Trading a strategy that involves using pivots, support, and resistance levels will only be successful if the right money management strategy is used with it. Combining the infinite number of money management strategies into this brief article will make it impossible to reach an end.

This article will therefore, by necessity, limit itself to working out the probability of a reversal happening at a pivot, support or resistance area. This will be compared to the probability of the market reversing anywhere in the range between the lowest support and highest resistance point.

First of all let us simplify the problem to make it possible to gauge the probability of the market reversing close to one of the pivot points. The term pivot points will from now on be used to refer to any of the pivot, support, resistance points.

Formulae used

PP = (HIGH + LOW + CLOSE) / 3 S1 = (2 * PP) - HIGH S2 = PP - RANGE S3 = S2 - RANGE R1 = (2 * PP) - LOW R2 = PP + RANGE R3 = R2 + RANGE

Parameters

  • The high and low of the day were used as the reversal points. These two point are undisputed and not subject to misinterpretation as points of reversal. (The only question I had in using either of these points is when the market closes on one of these points. Because pivots can also be used as targets I consider these valid scenarios in using these points and therefore dismissed this initial concern.)
  • The reversal (high or low) should fall within 0.2% of a pivot in order to qualify. 0.2% is 2 points on the ES when the ES is at 1000. It is reasonable to use percentages instead of absolute points as you would expect larger movements in an index the larger its value is and this will keep the analysis standard across all values and all symbols.
  • The 0.2% figure allows for 0.1% above and 0.1% below the pivot level. In practical terms, this would mean 1 point above and 1 point below the pivot level when the ES is priced at 1,000.

Data

I used data from 10 June 2005 and went back 631 trading days which gave me 630 days of pivot points and relevant data to use. I looked at four commonly traded index futures:

  • E-mini S&P 500: ES (RTH data)
  • Mini sized DJIA: YM
  • E-mini Russell 2000: ER2
  • E-mini Nasdaq 100: NQ (RTH data only)

I had/have a number of analysis tools at my disposal but I chose to use Excel because this could be elegantly and quickly done using Excel and the calculations could be verified on the fly to prevent errors.

I have also selected 2 symbols for which I have used all session data and 2 symbols for which I use RTH (pit traded times) data.

Benchmark

In order to ascertain if the pivot points are working as areas to locate reversal points we need a benchmark against which to measure them.

As a benchmark I've chosen the area bounded by 0.1% above the R3 line and 0.1% below the S3 line. I have then calculated the area that is bound by the bands on either side of each pivot line and added together all of those values. From these "pivot areas" I can calculate the percentage of the area bounded by pivot bands versus the total area bounded by the upper and lower pivots.

I now make the statement that randomly picking 7 areas for reversals in this range should give the same percentage of highs and lows as the area covered by the percentage of those 7 areas versus the total area.

Some of what I said above may be a bit confusing so let's just take a look at a simple example:

If the upper value for the range was 1050 (R3 + 0.1%) and the lower value was 950 (S3 - 0.1%) at the width of each pivot band was 2 points (1 point on either side of each pivot line) and we have 7 pivot lines then our bands take up 14 points in an area of 100 points. If we were to randomly select 7 bands of 2 points each in this area for the high and the low of the day then what I am saying is that there is a 14% chance that the high/low will fall into one of those bands if it falls in our area at all.

These 7 bands are, however, pivot points, so we are assuming that there is a higher probability that reversal points will take place in these bands.

Depending on the width of the bands (they will vary slightly from day to day) and the range between S3 and R3 (this will vary from day to day) the percentage of the area occupied by the pivot bands will change from day to day. I have, therefore averaged this over the period that was tested.

Results

Random High Low
ES 34.2% 34.6% 37.3%
YM 32.4% 29.8% 35.9%
ER2 23.2% 24.9% 24.8%
NQ 21.9% 24.6% 23.7%

Understanding the Results

I would like to be clear about what the results are telling us.

If we take the ES line in the results table it is telling us (from left to right) that on average, 34.2% of the area in the range covered by S3 to R3 was also covered by the pivot bands. That 34.6% of the time the high for the day was in one of the 7 pivot bands and 37.3% of the time the low of the day was in one of the 7 pivot bands.

We can see from the results that in all cases (except for the high for the YM) that the probability of the high or low falling inside one of the pivot bands is higher than would be expected by randomly picking 7 bands for reversals.

We can also see that these pivot bands are only marginally more probable areas for reversal (or at least picking the high and low of the day) than a random system.

Other

This is only one study of the pivot points. There are hundreds of more variables and permutations that you can add to this and other studies. You can make the bands wider and narrower. You can add money management strategies. You can look at other reversals during the day. You can calculate the pivots based on different time frames. You can calculate the highs and lows of the day based on different time frames.