Floor Pivot Probability Trading

[pmsinc 10]

I wanted to look at the historical track record of the Floor Pivot points for the ES Futures. Not having enough daily data I used the SPX index instead. I looked the time period from 01/02/1962 to 03/31/2009 which was available from Yahoo as a free download. Dates prior to 1962 did not have the OHLC data. So I looked at a total of 11,892 days. I read in “Trading in the Zone” Casinos have a 4.5% edge at the blackjack table and make millions due to volume of games played. How much of an edge is needed in ES trading using sound Money Management skills to beat the house?

The Pivot was defined as the previous days (H+L+C)/3.

Standard R1,R2,S1,S2 calculations also

 

The results are as follows:

6,296 days where the open was above the Pivot

5,596 days where the open was below the Pivot

6,288 days closed UP

5,596 days closed DOWN

5,597 days R1 was hit (47%)

1,899 days R2 was hit (16%)

5,170 days S1 was hit (43%)

1,863 days S2 was hit (16%)

924 days R1 and S1 was hit (8%)

2,049 days R1 and S1 NOT hit (Small range day?) (17%)

88 days R2 AND S2 hit (0.7%)

 

2,826 days opened ABOVE pivot and closed DOWN (51%)

2,770 days opened BELOW pivot and closed UP (49%)

3,518 days opened ABOVE pivot and closed UP (56%)

2,778 days opened BELOW pivot and closed DOWN (44%)

 

3,735 days opened ABOVE pivot and R1 hit (59%)

1,407 days opened ABOVE pivot and R2 hit (22%)

3,233 days opened BELOW pivot and S1 hit (58%)

1,216 days opened BELOW pivot and S2 hit (22%)

 

1,947 days opened ABOVE pivot and S1 hit (31%)

1,862 days opened BELOW pivot and R1 hit (33%)

 

So how can we use this information to get a higher probability of trading wins?

As a novice I see:

If we open above the Pivot there is a 59% change we will hit R1 and a 56% chance we will have a close greater than the high. Setting the stop to S2 and we will be stopped out 22% of the time. However a higher Stop would give us a better reward/Risk Ratio and possibly get us out of a down trend sooner.

Setting the stop to S1 and we will have a 31% change of getting stopped out.

 

What other factors can I incorporate to get a higher probablity based on this data?

[Tams 20]

One suggestion:

overlay the Bollinger Bands on the price bars.

observe the pivot behavior around the outerbands (+-2sd).

those are your critical points.

anything in between is just a "hold".

[Trader333 10]

Tams,

 

There are a number of settings that Bollinger Bands could be set to and more specifically how far back to calculate, what would you suggest for this ?

 

 

 

Paul

[BlowFish 10]

You actually don't need a huge edge, however there are a few different variables that are important. One of the most important is risk of ruin, I like this site which talks about that http://www.traderscalm.com/ror0.html

 

Expectancy is another statistic that many people look at. Google will reveal a gazillion pages on that. Sounds like you probably know about that as you where talking about R:R.

 

With an approach based on the stats you will want to look at maximum adverse excursions and minimum adverse excursions. Eliminating outliers that could turn into big losses would be very prudent. Again seems like that is what you where doing in your example looking at S1 or S2 as a stop. There are 'formal' ways to look at these to determine stops that suit you best.

 

As for filters to improve results you could look at daily pivots in respect to weekly and monthly pivots. You could use some sort of 'price action' to trigger a trade, you might have to give up some potential profit to get some confirmation. There's lots you might try.

 

Can I asked do you use Excel to massage your figures or something more sophisticated?

[pmsinc 10]

I used Excel.

I was hoping not to use any lagging indicators. I have considered using a break fo the first 30 or 60 minute range but that data is not in the daily OHLC.

I guess I could also improve the odds by looking at the general trend and only take long positions if the trend is up. But I need to add a filter for that.

I could calculate the R:R as Reward = R1-Open and Risk = Open-S1 if open is above pivot. However this would almost always have a poor ratio.

[Tams 20]

Tams,

There are a number of settings that Bollinger Bands could be set to and more specifically how far back to calculate, what would you suggest for this ?

Paul

 

start with the standard settings:

20MA, +-2SD

[BlowFish 10]

Try messing with weekly and monthly pivots. They 'work' in the same way as ordinary daily ones, just a different sample period no more or less 'laggy'. So normally the PP generated from last week you would project into this week. There is some inherent geometry in pivots S1,R1 etc. Probably why they 'work' just as well on a 7 minute chart as a 3 day chart. If you look hard at the arithmetic you will probably see the geometry they represent.

 

After a wide range day you might want to look at halfway points too. Due to the way they are constructed a wide range up, closing near the high will put R1 off in space. Or maybe look at other filters, stats, or processing based on range, that seems to have a high level of synergy with what you are doing.

 

Do you have access to, and would you consider using volume information? Jerry Perls trading with market statistics threads might be of interest. (In the market profile section) They are a great read whatever your approach, and something quite novel which is a rare thing.