Pivot Stats revealed

[Tubbs]

Right, I've done an analysis on the S&P market from 1982 that's 5954 days.

The test was to use daily data to see what percentage of times the market's high or low bounced off the different often quoted Pivot levels. Put another way, how effective are these levels. Now obviously day traders use these intraday a bit, which would mean the market may temporarily bounce off these levels and may later go through them, meaning my analysis would not show this. I think, though, that the largest moves require that either the high or low of the day hits and bounces off one of these levels.

So I took the usual pivot levels and created a band around then. I used a 10% of the previous range either side of the pivot level. This seemed reasonable.

Then I did a few stats test to see if the level was hit and if so did it act as support or resistance.

The following is what I found. Sorry about the formatting it's driven me crazy. The first number is the occurrance frequency the second is the percentagel.

Pivot Only Stats Number Percent
days total 5954
pivot touch 4551 76.4
Support 764 16.7
Resistance 591 12.9

S1 Stats Number Percent
days total 5954
touched 3008 50.5
Support 809 26.8
Resistance 67 2.2

S2 Stats Number Percent
days total 5954
touched 1186 19.9
Support 591 49.8
Resistance 21 1.7

S2 Stats Number Percent
days total 5954
touched 635 10.6
Support 167 26.2
Resistance 3 0.47

R1 Stats Number Percent
days total 5954
touched 3172 53.2
Support 100 3.1
Resistance 970 30.5

R2 Stats Number Percent
days total 5954
touched 1418 23.81
Support 9 0.63
Resistance 386 27.2

R2 Stats Number Percent
days total 5954
touched 582 9.7
Support 3 0.51
Resistance 166 28.5

So all in all not much to write home about. Would appreciate comments. I've also done a spreadsheet for data mining different zonal open/close and support resistance based on J.T Jackson and or Krausz.

 

[Tubbs]

The best stat is surely that the pivot is touched 76% of the time.

 

[trendie]

Tubbs,

good work! I did something similar a while ago on the Dow and GBPUSD.
( I am assuming that the second S2 is really S3, and the seciond R2 is really R3 )

I put my own analysis aside after getting confused, when looking at it EOD.

areas of ambiguity for me were:
01: there is a difference between Pivot-P1 being "touched" and "run through", so I felt the P1 number was almost meaningless. I see that you also get P1 touched 3 days out of 4.

02: if the previous traded day was tight, then the following days pivots were essentially meaningless, as the pivots S2/S3/R2/R3 are more likely to be hit. the pivots work when days are "average".
similarly, a strong day gave wide pivots for the following day.
If tomorrow turns out to be "average", then the pivots are misleading.

I started going down the road of getting the average daily range of the past 5 days, and getting my pivots from them. ( bear in mind, just because y-day was wide, doesnt mean tomorrow will be. just because y-day was tight, doesnt mean tomorrow will be, so the pivots derived from "out of sample" days could give points that have no real value. )

03: I also found that only with hindsight could I determine whether the pivot got hit and stayed there, or got hit, and rebounded back to P1. ( trend or range day )
I started looking at scenarios where pivot got hit, and then monitored whether it rebounded to previous pivot.
eg; if S2 gets hit, did it carry on, or rebound back to S1. if it hit S1, did it rebound back to P1 etc.

My studies are gathering dust somewhere, awaiting re-analysis.

Strong "out of sample" moves just screws it all up.
All the above was done last year/early this year.

If I ever return to the study, I intend to quantify "out of sample" (NFP etc ), and how to deal with them, since most of the trading/analysis problems occur due to days being abnormally tight or wide.

 

[Tubbs]

Thanks for your reply trendie. You have basically hit the nail on the head with what is wrong with Pivot analysis. Mine and your research really backs this up. I really just wanted to check roughly if there was any validity in the points usually stated. I don't think there is. Looking at the data shows that it is no wonder the pivot is usually hit given its placement in the range. Quite difficult to put into a strategy though given that once it has been hit that is the stat gone, and usually it open right near the pivot. Maybe using globex for a very small gain may be the only way.

Back to the drawing board

 

[egro1egro]

I also think, that pivot points that are not natural (i.e. taken from the price action) but rather calculated with some magical formula, don't make any profit. Who cares about a level based on (H+L+C)/3 plus minus something?
And how to use it even if it has any statistical significance? Buy when it touches, sell when it touches, or goes through, or closes behind? All these variations don't change profitablitiy much.
If a consolidation just happened on one those levels then the price touches the level and goes through and closes above/below 15 times a day without moving in any particular direction. If you play a probability game with this then you are sure to loose even if a really big move comes about eventually. If you limit number of losses by say 3 and then stop, then you are even sure to miss a big move as well. So why bother?
Naturally occuring resistance/support levels are more significant, but the question remains how to trade them? But at least there is a base under them.

 

[Tubbs]

Agreed egro. I think it is strange how these pivot numbers came about. I guess it was someone very clever in the pits back in the thirties that sold a system to pit traders. Probably just stemmed from that. I tried to get the Fisher ACD method to work but that had similar problems. Interesting the Market Profile point of control - similar to a pivot - is equally as hard to trade with. I think I've come to the conclusion that any system with a profit target is doomed. I may be wrong but any good system I have seen has a low win loss ratio but there are occasional very large profitable runs. This is effectively trend following - whatever that means.

 

[egro1egro]

Well, I look at statistically significant patterns in this way.
If in 55% they precede an upmove for at least, lets say, 2 points (or at least 30% of the pattern size, if you don't want fixed numbers) then you also look at retracements before reaching the target (2 points or 30%). If the retracements are mostly 4 points at least (or 60%) and on only a tiny fraction of all cases has small retracements or no retracements then you can successfully trade the opposite side with a predefined target level (which is 3-4 points or 50-60%) and you will win the probability game in this way. If the pattern predicts quite well an upmove in a bigger number of cases with a smaller retracements then you would probably be better off by trading long.
But I have also found that these probabilities don't describe the reality very well. Looking at patterns more attentively, noting many important details about how price action developed into such a pattern can give away much more. And instead of playing very tough probability game you'd better be off by doing some sort of intuitive trading. But it requires a lot of paper trading and real hands-on experience rather then a mechanical system or a set of fixed rules. You can very often notice some common details in all or many failed patterns.
For instance, you are trading a breakout of a resistance level. The price goes down to a support level in a choppy way, then quickly jumps up without small retracements without prior hesitation with nice long same-color candles and takes out the resistance level. I don't expect the move last very long and wouldn't trade such a breakout. It looks more like a false move. Or it was caused by some news while it was at the bottom. In any case, it is not a trend. The trend would have a beginning, some hesitation first or the move has to be steady with some shallow retracements. It should have some slight imperfections. It might have some consolidation before taking out a resistance. If you don't see any ot these why on earth would you jump on it.
It is just an example. The price action might look differently on different markets. Occasionally a new trend might start suddenly on a good news and move in one direction only for some time before retracing a bit. But it might account only for 5% of all breakouts.
Sorry, if it is overloaded.

 

[DeadMoney]

someone just told me they were going to go read a book on flying an airplane, when they were finished I could go with them on their first flight.
I think the stock market works the same way, there are no books or system that will ever make you a profitable trader.
That said, this morning I shorted the emini at 703.50 partly because the pivot point was 704.50
but also because the market had been going up for the first half hour and it was 10 am, which is a time that it is very likely the market will turn.
the market then dropped for just over 2 hours and I got out of my trade at 685.50 because it was a pivot point and the market had been going down for 2 hours and it looked to me like it was going to bounce and go up.
I also stayed in the trade because it broke the 3 lows of the last 3 days, each of which gives the market some more down momentum.

anyway, I made a very nice trade for 18 points today and the pivot points were a big part of the trade. but not that you could ever just buy and sell pivot points and think you were going to make any money, that would never work.
but when they line up with other factors then they can be very powerful trades.
I would say the number one thing I look at when day trading is the time of day, then pivot points, then the high and low of the last 3 or 4 days.
when all of these things line up then it makes some very high probability trades.

the pivot points I use for the emini are here E-mini S&P500 March 2009
and other than that I look at a daily chart and a 5 minute chart.
I also want the dow index, new york index, and spx moving with me.

 

[ektrader]

I could be wrong and if so, may I please stand corrected. I think the pivots are used buy the market makers as goals to push the market too until price action takes over on it's own. That being said it's good to know where they are but you shouldn't trade them blindly. I would pay closer attention to the actual turning points instead. Also.... The market makers and other professional traders "fade" the pivots because they are well aware newbies are using them.

Ektrader

 

[rathcoole_exile]

IThe market makers and other professional traders "fade" the pivots because they are well aware newbies are using them.

so why not go with the "smart" money and fade them yourself ?

 

[ektrader]

so why not go with the "smart" money and fade them yourself ?

You are right. And depending on what I see in the price action, sometimes I do

ektrader