I am going to start posting statistics for ES. I hope you will find them useful. The open is at 4:30pm EST(6pm on sun) and these stats show the probabilities of a target being hit by the CLOSE tomorrow. So if the targets(YH, YO, YL) are taken out in the after hours or PreMarket session those count as well. If they are not taken out, than they are still good for the RTH session. So with that lets go:

From Globex Open @ 1120 on 9-23-10
Distance to Yesterday Low = -2.75 Chance to get down to 1117.25 = 86.23% http://screencast.com/t/NTFhMTM1N

Distance to Yesterday Globex Open = 10 Chance to get up to 1130 = 33.80% http://screencast.com/t/YTEzZGZl
These stats are based on Globex data from Sept 1998, filtered as shown in images.

Laserdan

The following 3 users say Thank You to Laserdan for this post:

Well I dont really USE them per say. But I do think that they can be useful in certain situations. Like today, YL had an 86% chance of being taken out. But it never did...they are just statististics. Highly probable != certain.

Now if i see that pice action is in an up trend AND YH hasnt been taken out AND stats say that YH is a reasonable target with a reasonable chance of success AND its not too far away, than I might, repeat might, put a target up there near YH.

So a lot of things have to line up for me to ue it in my trading. I thought i would share the stats each day. Use them in your trading as you wish,or don't use them at all. just my little contibution to the community.

Laserdan

The following user says Thank You to Laserdan for this post:

I use some similar concepts, but in a different way.

Your statistics are about volatility and conditional probabilities. The volatility is interesting, I use it by plotting yesterday's low and high, pivot levels and pivot ranges. The conditional probability that you indicate is less interesting. When the market opens - or for Fridays after the news release, which is prior to the open - one of several scenarios can be validated.

For example yesterday evening ES tested the Globex low, stopped 1 tick below and immediately snapped back. This is a perfect long setup which was the confirmed by a gap up.

I trade pretty much how laserdan describes it.... I have statistics, and i look for certain conditions that i look for to see if it agrees... If the risk reward is in my favor(usually 1 point risk per 2 point target), then i take it.

1. when you have a 80% chance of hitting a target, 2. a system(not exactly mechanical, could be anything such as volume or price action) that you use which gives you an edge in terms of trading signals, 3. and if( and only if) the risk:reward is in your favor, then you take it.

Im just pointing out how many stars need to be aligned in order for you to take a trade. You honestly only get 2-3 good setups per DAY if you follow these rules... Most people make 20 trades a day. Thats not right( unless your a scalper, and i haven't met any scalpers that are profitable).

I generally think anything less than 1 point is scalping.. more like 1-2 ticks.... but with a day like today when its narrow and slow, 2 points is def not scalping...

Sometimes you have to lower your targets when volatility dies, but when it expands, so does your risk, and so should your targets... so i guess the real definition of scalping is when you constantly go for a few ticks as your target..

Out of all the trading days since sept 97 where
the day before was not an inside day
AND the day before was an Up day
AND it opened extremely high in the day before's range (between 75% and 99%)
AND the distance to the target(YH) was between 2.5 and 5 points

80.98% of those days the target (YH)was filled.

There were 184 days that met all of the criteria above and 149 of them closed the distance to the target (YH).

The most recent ones were:
09/01/2010
08/26/2010
07/27/2010
06/11/2010
06/09/2010

Laserdan

The following user says Thank You to Laserdan for this post:

If you have an up day and overnight price action - Globex influenced by Asian and European Markets - confirms the close/high range of the prior day by producing an open near the highs of the prior day, it is indeed likely that the high will be tested.

The prior direction being up, there are three scenarios for the day that can develop:

(a) Balancing Day: In this case price will either retrace to a 38% or 50% fib level and then test the highs, or do the same thing in an inverted order.

(b) Up Thrust Day: In this case the prior day's high will be exceeded

(c) Down Day: In this case the prior day's high should be at least tested. The test can be a failure, so if the high is nearly reached - just 2 or 3 ticks off - this satisfies the failed test. However the failed test can be produced during the night session, and in this case price never looks back.

yes, I should have said "target filled and/or exceeded" because I measure by what % the pice actually gets to
DATE YH Filled %
09/01/2010 1230%
08/26/2010 260%
07/27/2010 380%
06/11/2010 227%
06/09/2010 441%

Would be interesting to have stats for the daily pivot. Similarly to what you did with the Globex open could we exploit the distance of price at the open at 9:30 EST to the daily pivot. Or if price is above the daily pivot what's the chance to have a green day. Samething when price is below the daily pivot at the open.

Agreed. The problem is I have access to only Globex data. Now if someone could provide me with some RTH ES data, say going back 10 or so years, I could duplicate my database in a flash(and possibly add pivots). hint hint

do you have some stats when price makes a Low during the Globex hours if price test it for the first time but during the RTH what's the probability for this test to fail or for price to bounce off of that Low ? Same thing for the High. The risk/reward of a failed test is very appealing.

I could do this with the current data IF the globex low is not taken out during the premarket and the retest of the Globex Low just so happens to be the Low of the day.

I would just change the Y target filled % to group by a different % ranges. I would think that 98-102% would show these retests depending on the range of the prev day.

Actually today is a perfect example where price went 1 point above YH and retraced
with a range of 18.75 yesterday each point converts to 5.3%
so I grouped Y target filled % by 5%

So the chance of a retracement being the HOD is 3.8% (1.63 + 2.17)
where as the chance of execeeding the HoD by more than 2 points is 77.17%

That does not mean that there wont be a bounce off the HoD...that I can not tell with daily data.

I think what's missing in your stats (which renders them unusefull in their present state) is a second point of reference. Example, the distance to yesterday High = 5.75, what's the probability for price to return to 1146.75 before making an equivalent trip in the opposite direction considering the opening print at 1141. Or if price moves lower after the opening say 1 point lower at 1140 or 2 or whatever value, what's the chance to get up to 1146.75 ? Conditional stats speaks more than generic stats usually. Conditional stats like this provide actionable knowledge that may result in value.

I see your point, and that is the rub with working with OHLC daily values...there is no time component. In your example above the Chance to get up to 1146.75 = 57.02%. No matter how far the price drops, the chance remains the same. Only if we take out the YL target would the % change. Then we could look for the % of both YL and YH targets being hit. See post # 10.

We have only had a couple days (maybe one if I remeber correctly) where the stats have been very useful. Some days they are dead on.

Like I said before...tomorrow is a coin flip as far as direction goes. The predictive value of the stats are limited. If they are useful to you GREAT! If not, thats OK too.